1 A Comparative Analysis of the Socioeconomic Dimensions of LGBTI Exclusion in Serbia July 2019 Table of Contents Executive Summary ................................................................................................................................. 1 Key findings .......................................................................................................................................... 2 Way forward ........................................................................................................................................ 5 1. INTRODUCTION ................................................................................................................................ 8 1.1. Methodology and sample and survey demographics ...........................................................10 1.2. Limitations of the study .........................................................................................................15 2. INCOME AND ECONOMIC WELL-BEING .........................................................................................16 2.1 Personal income ....................................................................................................................19 2.2 Household income .................................................................................................................23 2.3 At-risk-of-poverty rate ...........................................................................................................24 2.4 Financial burden ....................................................................................................................26 2.5 Material deprivation and economic well-being ....................................................................26 3. LGBTI PEOPLE IN THE LABOR MARKET ..........................................................................................30 3.1 Employment status ................................................................................................................31 3.2 Activity, Inactivity, Employment, and Unemployment Rates ................................................34 3.3 Employment sector and types of contracts ...........................................................................35 3.4 Additional jobs and working hours ........................................................................................37 3.5 Changing jobs .........................................................................................................................38 3.6 Long-term unemployment.....................................................................................................39 3.7 Seeking employment and a minimal salary ...........................................................................40 4. LGBTI PEOPLE AND WORKPLACE DISCRIMINATION ......................................................................43 4.1 Being openly LGBTI at work ...................................................................................................44 4.2 Workplace climate and discrimination ..................................................................................48 4.2.1 Consequences of workplace discrimination ..........................................................................51 5. THE QUALITY OF LIFE OF LGBTI PEPOLE ........................................................................................54 5.1 Self-rated health ....................................................................................................................55 5.2 Life satisfaction and emotional health ..................................................................................58 5.3 Support networks ..................................................................................................................63 5.4 Levels of trust and feelings of safety .....................................................................................64 5.5 Perceived impacts of being LGBTI on economic well-being ..................................................66 6. CONCLUSION AND RECOMMENDATIONS .....................................................................................68 6.1. Conclusion ..............................................................................................................................68 6.2. Recommendations and next steps ........................................................................................69 Annex 1. Landing Page of the LGBTI Survey (only text included) ..........................................................73 Annex 2. Detailed Methodology ............................................................................................................74 Annex 3. Detailed Demographic Analysis ..............................................................................................79 Annex 4. Additional Graphs ...................................................................................................................86 Annex 5. Coding Steps for Personal and Household Income ................................................................89 Annex 6. Survey Instrument ..................................................................................................................91 Annex 7. List of Survey Partner Organizations (NGOs) ..........................................................................92 Acknowledgments This report is the product of more than two years of labor on the part of a team of many people and organizations. Most of all, we would like to acknowledge the LGBTI people in Serbia who shared insights into their lives as part of this work. We know this is not always easy and we hope that we have done justice to your experiences. ERA – LGBTI Equal Rights Association for the Western Balkans and Turkey was an essential partner in the realization of this report, in particular, Amarildo Fecanji and Dragana Todorović and their network of partners throughout Serbia, who helped shape the questions and encouraged people to respond. In addition, the following nongovernmental organizations were essential in the implementation of the research: Geten, Egal Serbia, Da Se Zna!, AS Centar, Come Out, Labris, YUCOM, FemSlam, Rainbow Association, Red Line. The survey was implemented, and the results analyzed, thanks to the tireless commitment of a team at IPSOS Strategic Marketing, led by Milena Lazic, Svetlana Logar, Vojislav Mihailovic, and Marko Uljarevic. The guidance provided by the Williams Institute, especially Andrew Park and Andrew Flores, was invaluable in completing the work. At the World Bank, the team was led by Georgia Harley, Runyararo Gladys Senderayi, and Nicholas Menzies (Task Team Leaders), with the unwavering support of Dominik Koehler (Project Coordinator). The team also drew on the expertise of Kristen Himelein Kastelic, Marko Karadzic, and Srdjan Svircev. The report also benefited from Patricia Carley’s excellent editorial insights and incisive comments and suggestions from peer reviewers: Ana Maria Oviedo, Clifton Cortez, and Vuk Raičević, as well as May Olalia. The close collaboration between the World Bank’s Governance and Social Development Global Practices and the Country Management Unit for the Western Balkans was critical to the success of this work. In particular, the team is grateful to Ellen Goldstein, Linda Van Gelder, Adrian Fozzard, Roby Senderowitsch, Tony Verheijen, Stephen Ndegwa, and Goran Tinjic. All errors are the team’s alone, and the views expressed herein do not represent the position of the World Bank or its Executive Directors. Funding for this work was generously provided by the Nordic Trust Fund, a knowledge and learning platform promoting human rights–based approaches to development, the Western Balkans Country Management Unit, and the Social Development Global Practice. The authors can be contacted at dkoehler1@worldbank.org Glossary This glossary seeks to provide a common basis for interpreting the report by describing concepts related to sexual orientation, gender identity, gender expression, and sex characteristics. The glossary contains terms commonly used in the English language at the time of this report. Concepts and terms related to sexual orientation, gender identity, and sex characteristics vary across cultures and languages. Therefore, this list is by no means complete or exhaustive. Further, sex, gender, and sexuality are often understood to fall along a spectrum and be fluid over time. Acronyms SOGI Sexual orientation and gender identity LGBTI Lesbian, gay, bisexual, transgender, and intersex people SGM Sexual and gender minorities General Terms Queer An umbrella term that includes lesbians, gay men, bisexuals, transgender people, intersex people, and others. For decades, “queer” was used solely as a slur for gays and lesbians but was reclaimed by activists as a term of self- identification. Sexual and Persons whose sex, gender, sexual orientation, gender identity, and/or gender Gender Minorities expression differ from those of the majority of the surrounding society. Sex Sex The classification of a person as female, male, or intersex. Infants are usually assigned a sex at birth based on the appearance of their external anatomy. A person’s sex is a combination of bodily characteristics, including their chromosomes (typically XY chromosome=male, XX chromosome=female), their reproductive organs, and their secondary sex characteristics. Sex Assigned at The sex classification of people at birth. This is usually assigned by a medical Birth practitioner after a brief review of a newborn’s genitalia. Sex Each person’s physical features relating to sex, including genitalia and other Characteristics sexual and reproductive anatomy, chromosomes, hormones, and secondary physical features emerging from puberty. Intersex An umbrella term that refers to people who have one or more of a range of variations in physical sex characteristics that fall outside of traditional conceptions of male or female bodies. Some intersex characteristics are identified at birth, while other people may not discover they have intersex traits until puberty or later in life. Note that intersex is not synonymous with transgender. Gender Identity Gender Gender refers to social, behavioral, and cultural attributes, expectations, and norms associated with being male or female. There is increasing consensus that gender goes beyond the binary concept of men and women. Gender Identity Each person’s deeply felt internal and individual experience of gender (e.g., of being a man, a woman, in-between, neither, or something else), which may or may not correspond to the sex they were assigned at birth or the gender attributed to them by society. It includes the personal sense of the body (which may involve, if freely chosen, modification of appearance or function by medical, surgical, or other means) and expressions of gender, including dress, speech, and mannerisms. Note that this sense of self is separate from sex assigned at birth and is not related to sexual orientation. Gender identity is internal and not necessarily visible to others. Gender The way we show our gender to the world around us through things such as Expression clothing, hairstyles, and mannerisms, to name a few. Masculinity/ The socially, historically, and politically constructed qualities associated with Femininity men and women, or maleness and femaleness, in a society at a particular time. The definitions change over time and are different from place to place. Although they seem to be gender-specific, women perform and produce the meaning and practices of the masculine, and men perform and produce those of the feminine. Cisgender Cis or cisgender is used for people whose gender identity is in alignment with the sex assigned to them at birth (“cis” meaning “in alignment with” or “on the same side”). Transgender This refers to a person whose sex assigned at birth does not match their gender identity. The term “trans” is often used as shorthand. Trans Man A person whose sex assigned at birth was female, but who identifies as male. Trans Woman A person whose sex assigned at birth was male, but who identifies as female. Transphobia The irrational fear of those who are gender variant and/or the inability to deal with gender ambiguity. It also describes discriminatory treatment of individuals who do not conform in presentation and/or identity to conventional conceptions of gender and/or those who do not identify with or express their assigned sex. Sexual Orientation Sexual Each person’s enduring capacity for profound romantic, emotional, and/or Orientation physical feelings for, or attraction to, person(s) of a particular sex or gender. It encompasses hetero-, homo-, and bisexuality and a wide range of other expressions of sexual orientation. Lesbian A woman who predominantly has the capacity for romantic, emotional, and/or physical attraction to other women. Gay A man who predominantly has the capacity for romantic, emotional, and/or physical attraction to other men. The term is sometimes used to also describe women who are attracted to other women. Heterosexual People who are attracted to individuals of a different sex and/or gender identity from their own (also referred to as “straight”). Bisexual People who have the capacity for romantic, emotional, and/or physical attraction to person(s) of the same sex or gender as well as to person(s) of a different sex or gender. Homophobia The irrational fear, hatred, or intolerance of homosexuality and/or homosexual people as a social group or as individuals. It also describes discrimination on the basis of sexual orientation. Biphobia The irrational fear, hatred, or intolerance of bisexuality and/or bisexual people as a social group or as individuals. It also describes discrimination on the basis of sexual orientation. Executive Summary 1. The primary objective of this research was to provide Serbian policy makers, civil society, and development partners with evidence of the socioeconomic status of lesbian, gay, bisexual, transgender, and intersex (LGBTI) people in Serbia. The findings of the report, which were derived from a survey of LGBTI people, help to address the significant quantitative data gap on the lives of LGBTI people in Serbia and seek to inform policies and programs to address LGBTI exclusion. The survey was designed to collect and document evidence of the socioeconomic status of LGBTI people in Serbia rather than to explore policy or programming responses. However, the findings revealed that, going forward, certain areas warrant further investigation and greater attention from a policy- making perspective. These include: expanding the LGBTI evidence base; increasing LGBTI diversity in public sector employment; improving trust in government and institutions; and bolstering awareness and capacity to effectively address transgender issues. 2. In addition to the above, the objective of the research was to test a survey methodology that addresses the persistent challenge of generating data that allow robust comparisons between the lives of LGBTI people and the general population. Such comparative data provide more valuable evidence about the unique challenges faced by LGBTI people by framing these challenges within the context of the experiences of the general population. The methodology and statistical analysis used here also offer a technique for other researchers interested in understanding the experience of LGBTI people compared to the population at large. The report presents the results of the survey and, where appropriate, attempts to contextualize the findings. However, where there is insufficient data and information to analyze and provide this contextualization, the report simply presents the results of the survey and does not attempt to explain the differences observed. Going forward, it would be beneficial to collect more data that allow in-depth analyses of the differences observed between LGBTI people and the general population. 3. This research adapted the Survey on Income and Living Conditions (SILC) instrument of the European Union (EU), a standardized household survey used regularly to collect data from the general Serbian population on income, poverty, social exclusion, and living conditions (“general SILC”). The SILC instrument was adapted to include questions on discrimination in the workplace, trust in political and legal institutions, and overall life satisfaction of LGBTI people in Serbia. Between February and April 2017, 998 self-identified LGBTI people answered the adapted “LGBTI SILC” online. 1 Online engagement was chosen to encourage a large number of people to participate privately and confidentially, which is a recognized strategy for reaching LGBTI people who may not otherwise be willing or feel safe in revealing their identity in face-to-face interviews.2 The survey was disseminated using a network of local nongovernmental organizations in Serbia, and respondents were made aware of the survey through social networks, advertisements, and dating apps. The responses were weighted, and a sample matching procedure was conducted using a covariate balancing propensity score to allow a comparison between the two samples: the LGBTI SILC and responses from the Serbian population to the general SILC. The samples were matched on the basis of: age, sex assigned at birth, educational attainment, marital status, region, and area of residence (urban or rural). This generated, for the first time, a data set on the socioeconomic status of LGBTI persons that allows a comparison to Serbia’s general population. 4. This report is part of a broader World Bank research initiative, “Understanding the Socio- Economic Dimensions of LGBTI Exclusion in the Western Balkans.” In addition to this report, the initiative includes two “mystery shopper” experiments on access to primary school education and the 1 The general SILC is a household survey that is carried out in person. 2 See Annex 2 for more details on the methodology, including safety measures. 1 private rental market in Serbia,3 as well as a survey of the experiences of LGBTI people in seven countries across southeastern Europe.4 The multifaceted nature of the initiative has helped to develop a more rounded understanding of the development challenges of LGBTI people as individuals and as participants in the economy and society more broadly. Key findings Income, poverty, and material deprivation 5. The median personal monthly incomes of LGBTI respondents and the general population are similar, but among LGBTI respondents, transgender people earn notably less. The median personal monthly income of LGBTI respondents (€256) was slightly higher than, but statistically indistinguishable from, the comparable general population (€236). Within the LGBTI sample, transgender people earn the least (€203) and less than the comparable general population. Respondents assigned female sex at birth or male sex at birth who are perceived as not gender conforming earn €41 and €81 less, respectively, in their median monthly personal incomes compared to respondents who are perceived as gender conforming. The experience of discrimination clearly has a direct impact on income: LGBTI respondents who had faced discrimination in the workplace earned €24 less than those who had not. Due largely to smaller overall household size, median household income is lower in LGBTI households (€489) for households in the comparable general population (€628). However, even though the median household income for LGBTI respondents is lower than that of the general population, the equivalized disposable household income of both groups is similar, due to different household compositions.5 6. The at-risk-of-poverty rate is similar for LGBTI respondents (17 percent) and the comparable general population (16 percent).6 The at-risk-of-poverty rate for the LGBTI sample does not decrease for older respondents or for those with higher levels of education, as it does in the general population. Transgender respondents have the highest at-risk-of-poverty rate (24 percent) among LGBTI subgroups.7 7. There are no substantial differences between the LGBTI sample (28 percent) and the general population (27 percent) in material deprivation rates. However, variations among LGBTI respondents are notable. Transgender and intersex respondents are at the highest risk of material deprivation (40 percent and 33 percent, respectively). Rates are also higher for older LGBTI respondents (27 percent of those over age 45) and LGBTI respondents with less education (32 percent of those without a college degree). Although there are no striking disparities between the overall LGBTI population and the general population when it comes to material deprivation, there are certain differences in their material possessions and expenses. While a greater percentage of LGBTI respondents state that they cannot afford a car or van, a larger share of the general population claims the same for a week-long vacation. A car might be perceived as more important among the general population, whose households are larger and more often include children. Additionally, a car requires ongoing 3 D. Koehler and others, “Discrimination against Sexual Minorities in Education and Housing: Evidence from Two Field Experiments in Serbia” (Washington, DC: World Bank, 2018), http://documents.worldbank.org/curated/en/509141526660806689/Discrimination - against-sexual-minorities-in-education-and-housing-evidence-from-two-field-experiments-in-Serbia (in Serbian). 4 World Bank, “Life on the Margins: Survey Results of the Experiences of LGBTI People in Southeastern Europe” (Washington, DC: World Bank, 2018), http://documents.worldbank.org/curated/en/123651538514203449/Life-on-the-Margins-Survey-Results-of-the-Experiences- of-LGBTI-People-in-Southeastern-Europe. 5 Equivalized disposable income is an estimate of household income based on the household’s total income divided by the number of members of the household, weighted according to their status and age. 6 The at-risk-of-poverty rate refers to the share of people with an equivalized disposable income below 60 percent of the national median equivalized income, which is calculated at 14,920 Serbian dinars (about €120). 7 This finding is consistent with o ther studies, including evidence from the World Bank report, “Life on the Margins: Survey Results of the Experiences of LGBTI People in Southeastern Europe”. Transgender people are among the most vulnerable groups in the LGBTI community, as they are disproportionately affected by discrimination and suffer greater levels of exclusion. 2 maintenance that might be less economically attractive to or feasible for smaller households with one or two members, which are more common in the LGBTI sample. On the other hand, a one-week vacation each year might be costlier for larger households but more affordable for a greater share of LGBTI respondents. Education and labor market participation 8. LGBTI respondents are more educated and more likely to have attained master and doctoral degrees than the general population. The question emerges whether this can be explained solely by a desire for better employment (which is certainly present) or by a perception that current and future surroundings might be more welcoming—or at least more tolerant—among people with higher levels of education.8 By attaining postgraduate education, LGBTI people could strive to become indispensable in their area of expertise and in so doing, leave less room for discrimination based on their identity. Additionally, unlike heterosexuals, lesbians, gay men, and some bisexuals may have a different pattern of major life events (e.g., not having children) 9 that may allow them to pursue higher education, though studies—and the results here—suggest that there is little economic benefit to sexual minorities for additional education. 10 The fact that LGBTI respondents are more educated than the general population is evidence of the available stock of high-quality human capital from which Serbia could benefit. 9. The unemployment rate for LGBTI respondents is 14 percent, which is less than half of the unemployment rate for the comparable general population (33 percent).11 Also, more LGBTI respondents (30 percent) are students compared to the general population (17 percent). LGBTI respondents are more often in contract work for a finite time period (35 percent) compared to the general population (28 percent), and they are five times as likely to have changed jobs during the past 12 months (26 percent) compared to the general population (5 percent). 10. One specific characteristic of LGBTI respondents in Serbia is that every tenth (10 percent) currently employed person is working in the nongovernmental sector, a percentage that is negligible among the general population. On the other hand, although every third (36 percent) employed person from the general population is working in the public sector, among LGBTI respondents it is one in five (20 percent). LGBTI respondents are far less “out” (or open about their identity) to their colleagues in the public sector and far more likely to be out to their colleagues in the nongovernmental sector. Moreover, LGBTI mistrust toward the state, as well as societal exclusion (as measured through the lack of social connections), has an impact on the way unemployed LGBTI respondents seek new employment. Those who are unemployed in the general population more frequently utilize the National Employment Service or address potential employers directly, and they are also more likely than LGBTI respondents to use informal methods, such as seeking a job through friends, relatives, or trade unions. Discrimination in the labor market 12 11. LGBTI respondents report discrimination and hostile work environments. One out of three (33 percent) LGBTI respondents report a general negative attitude at work toward LGBTI people. This 8 Chung Y. Barry, “Career Decision Making of Lesbian, Gay, and Bisexual Individuals,” Career Development Quarterly 44, no. 2 (1995): 178. 9 Roberto Baiocco, “Sexual Orientation and the Desires and Intentions to Become Parents,” Journal of Family Studies 19, no. 1 (2013): 90– 98. 10 Heather Antecol, Anneke Jong, and Michael Steinberger, “The Sexual Orientation Wage Gap: The Role of Occupational Sorting and Human Capital,” ILR Review 61, no. 4 (2008): 518–43. 11 The official unemployment rate in Serbia at the time of the survey (2017) for people between the ages of 15 and 64 was 14.1 percent (according to 2017 figures from the Statistical Office of the Republic of Serbia). The estimate in the report differs due to the matching process, which adjusts the demographics of the official SILC to match the demographics of the LGBTI sample. 12 These questions are not asked in the general SILC and hence no comparisons are possible between the LGBTI sample and the general population. 3 increases to 62 percent for transgender respondents. This fear of discrimination is evident in the fact that nearly half (47 percent) of L GBTI respondents are completely “closed off” from their work colleagues about being LGBTI. Gay and bisexual men are the most closed off at work. One-fifth (19 percent) of LGBTI respondents report that discriminatory practices are targeted toward the LGBTI population when entering the workplace. Their actual experiences confirm these concerns: 70 percent of transgender respondents, for example, have been asked about their gender identity by their employers or potential employers. Personal experiences of discrimination against LGBTI people once they are at work further demonstrate an unwelcome and stigmatizing environment. One out of five (20 percent) transgender respondents report suffering discrimination based on gender identity. Slightly less than one-fifth (17 percent) of intersex respondents said that they have been discriminated against at work based on their sex characteristics. Approximately one out of seven lesbian (14 percent) and gay (16 percent) respondents note that they have experienced discrimination at work because of their sexual orientation. 12. In 10 percent of cases, workplace discrimination forced LGBTI people to quit their job. The rates for transgender and intersex respondents are even higher (27 and 13 percent, respectively). The consequences of workplace discrimination are pervasive. LGBTI respondents who reported experiencing any kind of workplace discrimination have lower personal and household incomes, have higher rates of material deprivation (52 versus 25 percent), and are more likely to be currently unemployed (24 versus 11 percent) than those who have not experienced workplace discrimination. Given the current legal framework—Article 18 of the Serbian Labor Code prohibits direct and indirect forms of discrimination based on sexual orientation—it is important to further examine whether and how LGBTI people who experience workplace discrimination report such incidents to the authorities. LGBTI respondents’ reportedly low trust in the legal system may lower their likelihood of reporting such incidents. Comparative health status 13. LGBTI respondents (71 percent) less often report their state of health as very good or good compared to the general population (93 percent). Among LGBTI respondents, only 60 percent of bisexual female respondents report good or very good health, and for lesbians and transgender respondents the number is 68 percent. Worryingly, LGBTI respondents were five times more likely than the general population to have neglected to visit a doctor (45 percent compared to 9 percent). Although the LGBTI sample and the general population have different reasons for not visiting a doctor, LGBTI respondents were more likely to report a fear of doctors, hospitals, testing, or treatment. LGBTI trust in institutions, emotional states, and life satisfaction13 14. LGBTI people in Serbia report very low trust in institutions. Overall, 95 percent of LGBTI respondents have low trust in the political system, 93 percent feel the same about the legal system, and 91 percent share this view of the police. LGBTI people also have little social trust, with 69 percent reporting low trust in other people generally. Even though LGBTI respondents do not have trust in state institutions, a large majority feel very or mostly safe in the area where they live (81 percent). 15. Special attention should be paid to the propensity for LGBTI respondents to frequently experience nervousness, depression, and negative emotions. Even though almost half felt calm and collected and one-third felt happy most or all the time in the previous month, a substantial share reported feeling very nervous, depressed, or so despondent that nothing could improve their emotional state. Transgender respondents experience negative feelings most frequently among LGBTI subgroups, followed by bisexual women. One-third (32 percent) of transgender respondents feel very 13 See n. 13. 4 nervous all or most of the time, 42 percent feel depressed all or most of the time, and 31 percent feel so despondent all or most of the time that nothing could improve their emotional state. These negative emotions are more prevalent among those LGBTI respondents who are unemployed or inactive, a group that is even more disadvantaged because they are more likely to indicate that they have no one to ask for help. 16. A summary measure of life satisfaction shows that transgender respondents are the least satisfied among the LGBTI subgroups. Overall, 51 percent of LGBTI respondents report low life satisfaction, while 77 percent of transgender respondents report the same. Unemployed and inactive LGBTI respondents are even more dissatisfied with various aspects of life. Additionally, one-fifth (21 percent) of LGBTI respondents believe that being LGBTI negatively impacts their economic well-being. LGBTI subgroup variation: the situation of transgender people 17. The survey findings make clear the different experiences and outcomes of subgroups within the LGBTI umbrella. What stands out are the more negative results for transgender people compared to LGBTI people as a whole (as well as the general population) on almost all indicators. This survey reveals that transgender people earn less, are at a higher risk of poverty, and experience greater material deprivation. Transgender respondents report more discrimination in the workplace and are almost three times as likely to be forced to leave a job because of discrimination. Finally, transgender people report more frequent negative mental states and lower overall life satisfaction. Although the sample of transgender respondents is small and the results should therefore be taken with care, the statistic supports the findings of other studies on the particular challenges that transgender people face. Way forward 18. The survey results allow policy makers and others to identify the key areas in need of urgent attention to improve the socioeconomic status and inclusion of LGBTI people in Serbia: expanding the LGBTI evidence base; increasing LGBTI diversity in public sector employment; improving trust in government and institutions; and bolstering awareness and capacity to effectively address transgender issues. This is particularly important for Serbia as a candidate for EU membership and in light of the recommendations provided by the European Commission (EC) in its 2019 enlargement report for Serbia, especially with regard to Chapter 23: Judiciary and Fundamental Rights, and Chapter 24: Justice, Freedom, and Security.14 The EC stated that, “fundamental rights are largely enshrined in the legislation in the Western Balkans. However, serious efforts are needed to step up implementation.”15 In its 2019 report on Serbia, the EC specifically highlighted that, “Serbia needs to step up measures to protect the rights of persons facing discrimination, including LGBTI persons, persons with disabilities, persons with HIV/AIDS and other vulnerable individuals.”16 As part of the EU accession process, Serbia has developed action plans to combat discrimination and uphold human rights, including for LGBTI people, as outlined in the Fundamental Rights Charter. This survey provides new data points on implementation gaps and can inform the country’s response to the EC’s recommendations by facilitating dialogue with policy makers and key stakeholders, including LGBTI civil society organizations, to identify specific policy and program interventions and inform future action plans. Furthermore, the data can be used as a baseline to measure the impact of those policies and reforms. 14 European Commission, “Serbia 2019 Report” (Brussels: European Commission, 2019), https://ec.europa.eu/neighbourhood- enlargement/sites/near/files/20190529-serbia-report.pdf. 15 European Commission, “2019 Communication on EU Enlargement Policy” (Brussels: European Commission, 2019) https://ec.europa.eu/neighbourhood-enlargement/sites/near/files/20190529-communication-on-eu-enlargement-policy_en.pdf. 16 European Commission, “Serbia 2019 Report” (Brussels: Eu ropean Commission, 2019). 5 Expanding the LGBTI evidence base 19. This is the first known attempt to collect large-scale survey data comparing the outcomes of LGBTI people with the Serbian population at large. This report presents the key findings, but the data set is rich and could be used for further analyses, including greater detail about different LGBTI subgroups. The full data set is available here17. Researchers, advocates, and policy makers should be encouraged to delve further into the available data to inform their work. 20. This report represents a quantitative baseline on the status of LGBTI people. Over time, Serbia should conduct follow-up surveys to track results on reducing discrimination against, and the exclusion of, LGBTI people and on progress under Chapters 23 and 24. The Statistical Office of the Republic of Serbia should align itself with statistical agencies in other countries and begin collecting LGBTI-disaggregated data that provide regular up-to-date evidence and allow comparisons between the LGBTI population and the general population. This is needed to build more inclusive policies and programs at the country level; it would also allow the measurement of any changes over time to ascertain whether policies and programs are addressing LGBTI exclusion and other concerns. 21. This report revealed the particular challenges faced by transgender people. There are persistent difficulties, also faced when conducting this research, in reaching large numbers of transgender people through surveys. Special data collection efforts should be undertaken to ensure a full response from the transgender community. Furthermore, additional efforts are necessary to ensure that the most marginalized parts of the LGBTI community, for example, LGBTI people in rural areas, LGBTI Roma, and LGBTI persons with disabilities, are included in future data generation initiatives. Increasing LGBTI representation in public sector employment and strengthening workplace protections for LGBTI people 22. This research reveals that few LGBTI people are employed in the public sector. The existing legal protections against nondiscrimination do not appear to be working. State agencies set policies and run programs that serve the whole of society. With fewer LGBTI state employees, it is more likely that these policies and programs will overlook LGBTI Serbians. Furthermore, the public sector represents a source of stable employment. It is thus important that the public sector adopt active programs to increase the number of LGBTI employees through efforts to improve recruitment and workplace culture. To sensitize public servants, regular capacity building and awareness campaigns across all relevant government branches should be conducted in close consultation with Serbian LGBTI organizations. Improving trust in government and institutions 23. LGBTI people’s trust in Serbia’s social, political, and legal institutions is extremely low. The Serbian Government should consider using the survey findings to begin a consolidated outreach program to further understand the most critical factors underlying this lack of trust and take actions to address them. Beyond efforts to understand the lack of trust, the outreach program should be designed to better understand the broader concerns of the LGBTI community with a view to developing an action plan to address those issues also. Specifically, the action plan could include an outreach campaign targeting the general population to facilitate a change in attitudes and perceptions and promote more understanding and acceptance of the LGBTI community. The EU should support these actions through Chapters 23 and 24 of the accession process. Follow-up surveys should be conducted to track progress in improving trust. 17 https://microdata.worldbank.org/index.php/catalog/3486 6 Bolstering awareness and capacity to effectively address transgender exclusion 24. Concerted effort needs to be taken to address the particular disadvantage faced by transgender people. The Government should review the legal code to ensure that gender identity protections are enhanced and are at least on par with those provided for sexual orientation. More importantly, the Government should put in place programs to ensure these protections are actually realized and enforced. Civil society groups that work with transgender populations should be supported to enhance the services they provide. 25. Taking action to promote LGBTI inclusion is the right thing to do and it also makes economic sense. There is increasing evidence that links exclusion with detrimental health, education, and employment outcomes for LGBTI people, aggregating to broader impacts on the overall economy. 18 These effects can be mitigated with increased public acceptance of LGBTI people.19 The data contained in this report provide a sobering view of the challenges experienced by LGBTI people in Serbia. Addressing these challenges will not only ensure that the rights of all Serbia’s citizens are protected, respected, and fulfilled, but will also bring benefits to the country’s society and economy. More open and inclusive societies are more likely to make the most of their entire stock of human capital, more open and inclusive cities are better placed to attract international capital and talent, and more open and inclusive countries make attractive international tourist destinations. The inclusion of LGBTI people is important in and of itself—and it is also the economically smart thing to do. 18 For example, C. Banks, “The Cost of Homophobia: Literature Review of the Economic Impact of Homophobia in Canada,” Community - University Institute for Social Research (Saskatoon: University of Saskatchewan, 2004). See also G. Becker, The Economics of Discrimination, 2nd ed. (Chicago: University of Chicago Press, 1971); and M. V. L. Badgett, “The Economic Cost of Stigma and the Exclusion of LGBT People: A Case Study of India” (Washington, DC: World Bank, 2014), http://documents.worldbank.org/curated/en/527261468035379692/pdf/940400WP0Box380usion0of0LGBT0People.pdf. 19 For example, Banks, “The Cost of Homophobia.” See also Becker, The Economics of Discrimination. 7 1. INTRODUCTION 26. Qualitative studies about the experiences of lesbian, gay, bisexual, transgender, and intersex (LGBTI) people in Serbia have long shown widespread discrimination, exclusion, and violence.20 However, there has been a dearth of quantitative data to better understand the nature, extent, and consequences of these negative developmental experiences. Yet quantitative data can be critical to providing the widespread and more reliable information needed to design national policies and programs to improve the lives of LGBTI people.21 Furthermore, quantitative data can help to assess the impact of these new policies and programs. 27. This survey was conducted to provide policy makers, civil society organizations (CSOs), and development agencies with quantitative data on the socioeconomic outcomes of LGBTI people in Serbia and to compare those outcomes against results from the general population. The data cover several socioeconomic dimensions, including personal and household income, poverty risk and material deprivation, labor market participation and employment discrimination, health status, trust in institutions, and overall life satisfaction. The data were collected through an online survey carried out between February and April 2017. A total of 998 responses were collected. The questionnaire for this survey was developed based on the European Union’s (EU) Survey on Income and Living Conditions (SILC), which is regularly used to collect data on the Serbian general population. 28. The research was conducted as a partnership among the World Bank, the ERA - LGBTI Equal Rights Association for Western Balkans and Turkey (ERA), IPSOS Strategic Marketing, and the Williams Institute. The survey tool was reviewed by community-based LGBTI organizations. The full questionnaire used in the survey is provided in Annex 5. ERA administered the recruitment of participants through 17 local partner NGOs in Serbia.22 Extensive efforts were made to ensure that LGBTI people were aware of the research, motivated to participate in the survey, and encouraged others to take part. The survey was disseminated through social networks (Facebook, Twitter, and national social networking platforms), online banners on major national websites that attract large LGBTI audiences, advertisements placed on gay dating apps such as Grindr and PlanetRomeo, mailing lists, and oral channels. 29. This report provides one of the only comparisons of the socioeconomic status of LGBTI people to the population at large. It also provides the first large-scale quantitative data set on the exclusion of LGBTI people in Serbia. The innovative methodology helps to contextualize the lived reality of LGBTI people in Serbia while contributing to a growing body of evidence on the effects of discrimination based on sexual orientation, gender identity and expression, and intersex status. Perceived hostile workplaces and experiences of workplace discrimination have pervasive effects on employment, income, and life satisfaction. Despite clear antidiscrimination provisions in Serbian law that specifically reference sexual orientation and gender identity, LGBTI respondents still see the 20 For example, UNDP, “Being LGBTI in Eastern Europe: Serbia Country Report. A Participatory Review and Analysis of the Legal, Institutional, Policy and Socio-economic Environment for Lesbian, Gay, Bisexual, Transgender and Intersex People, and Civil Society ” (New York: United Nations Development Programme, 2017), http://www.eurasia.undp.org/content/dam/rbec/docs/undp-rbec-blee- SERBIA%20.pdf; and Cameron Burke, “Examining the Presence and Potential of Restorative Justice in Serbia in Cases of Discrimination,” Local Roots and Global Peace (Yerevan, Armenia: Eurasia International University, 2016): 58 –76. 21 Andrew Park, “An Inclusive Approach to Surveys of Sexual and Gender Minorities: Report of Meeting, Kathmandu, Nepal” (Los Ang eles: The Williams Institute, 2015). 22 See Annex 6 for a list of organizations. 8 workplace as discriminatory against LGBTI people. This report can help inform other research initiatives trying to establish comparable data sets. 30. Access to markets, services, and spaces matters. Social inclusion is at the core of the World Bank’s twin goals of ending extreme poverty and boosting shared prosperity. The 2013 World Bank flagship report, “Inclusion Matters,” provided an analytical framework to better understand the economic effects of exclusion and address the root causes of extreme poverty more effectively.23 Social Inclusion is defined as the ability of people to access markets, services, and spaces. Each of these dimensions provides opportunities for and barriers to inclusion. Individuals and groups can be excluded from these dimensions for a variety of reasons, and exclusion from one area does not necessarily result in exclusion from others. The negative economic effects of social exclusion have been well documented and underline the importance Source: World Bank, Inclusion Matters: The of more inclusive programs and policies. 24 Foundation for Shared Prosperity (Washington, DC: Available data from various countries suggest that World Bank, 2013). sexual and gender minorities are disproportionately overrepresented in the bottom 40 percent of the economy. This research builds on the markets, services, and spaces model 25 established in the “Inclusion Matters” report by collecting data that can help policy makers, development institutions, and civil society groups to better understand the exclusion that LGBTI people face in Serbia. 31. The findings of the survey can be used to improve the situation for LGBTI people in Serbia. The Serbian Government, development partners, and civil society can use the data to advocate for the development of appropriate legal frameworks, policies and programs to ensure that the rights of LGBTI people are adequately respected, protected, and enjoyed. The findings can also contribute to discussions on EU accession and strengthen and facilitate the legislative and policy changes that Serbia needs to carry out as part of that process. This report presents an overview of the findings, comparing the results to the matched sample of the general SILC. The data sets are available online. 32. This report is part of the broader World Bank research initiative, “Understanding the Socio- Economic Dimensions of LGBTI Exclusion in the Western Balkans.” In addition to this survey, there is a large-scale regional survey on the experiences of LGBTI people in southeastern Europe and a report 23 World Bank, Inclusion Matters: The Foundation for Shared Prosperity (Washington, DC: World Bank, 2013). 24 Ibid. 25 M. V. L. Badgett and others, “The Relationship between LGBT inclusion and Economic Developme nt: An Analysis of Emerging Economies” (Washington, DC and Los Angeles: USAID and the Williams Institute, 2014), https://williamsinstitute.law.ucla.edu/w p- content/uploads/lgbt-inclusion-and-development-november-2014.pdf; and M-A. Valfort, “LGBTI in OECD Countries,” OECD Social, Employment and Migration Working Papers (Paris, OECD, 2017), http://www.oecd.org/els/soc/LGBTI-in-OECD-Countries-A-Review- Valfort-2017.pdf. 9 based on two “mystery shopper” experiments that collected data on LGBTI discrimination in the education and housing sectors in Serbia. The multifaceted nature of the research initiative helps to better understand the development outcomes of LGBTI people as individuals, in the economy, and in society. 1. The regional survey adapts the 2012 European Union Fundamental Rights Agency’s Survey of LGBT people in the EU and Croatia and applies that methodology in Albania, Bosnia and Herzegovina, Croatia, Kosovo, North Macedonia, Montenegro, and Slovenia. Because both surveys were fielded at the same time, Serbia was excluded from the regional survey to avoid confusion among respondents and to reduce the risk of low response rates. That report can be accessed here. 2. The two mystery shopper experiments examined the extent of LGBTI exclusion in two key sectors in Serbia: (1) discrimination faced by “feminine boys” when accessing primary education and (2) the experiences of same-sex couples when trying to rent an apartment in the private rental market. That report can be found here. 33. The remainder of Chapter 1 examines the survey sample, summarizes the demographics of the participants, and explains how the sample was compared to the general population. The findings on income and economic well-being are presented in Chapter 2. Chapter 3 explores the situation of LGBTI people in the Serbian labor market. Chapter 4 documents the experiences of LGBTI people with discrimination in the labor market. Chapter 5 presents the results on the quality of life of LGBTI respondents, and Chapter 6 consists of the conclusion with policy recommendations, implications, and next steps for future research. 1.1. Methodology and sample and survey demographics 34. Representative surveys of LGBTI populations are difficult to conduct and online surveys are considered an appropriate method.26 LGBTI people are a hard-to-reach population due to at least two characteristics that make standard random sampling procedures inappropriate: the absence of a sampling frame (i.e., the characteristics of the total population are unknown) and the strong requirement for privacy protection given the stigma and risk of violence LGBTI people often face. This survey was conducted with a self-selected, nonprobability sample.27 As a result, it cannot be said that respondents represent the whole LGBTI population in Serbia. To address this concern, at least in part, the sample was weighted based on a study of the literature.28 Weighting sample characteristics to population targets to correct for over- and/or under-sampled groups can be effective in providing generalizable results, though the process is sensitive to the weighting strategy.29 Due to the lack of administrative data on LGBTI populations, the weighting strategy took into account only sex assigned at birth and sexual orientation. In total, 998 LGBTI respondents took part in the survey. A basic overview is provided in table 1.1.1. When interpreting the results, particular attention should be paid 26 N. S. Koch and J. A. Emery, “The Internet and Opinion Measurement: Surveying Marginalized Populations,” Social Science Quarterly 82, no . 1 (2001): 131–38; J. Rollins and H. N. Hirsch, “Sexual Identities and Political Engagement: A Queer Survey,” Social Politics 10, no. 3 (2003): 290–313; and E. Swank and B. Frahs, “Predicting Electoral Activism among Gays and Lesbians in the United States,” Journal of Applied Social Psychology 43, no. 7 (2013): 1382–93. 27 See Annex 2 for more details on the sample and weighting and Annex 4 for detailed demographics. 28 See Annex 2 for a description of the basis for the weighting. 29 C. Kennedy and others, “Evaluating Online Nonprobability Surveys” (Washington, DC: The Pew Research Center, 2016), https://www.pewresearch.org/methods/2016/05/02/evaluating-online-nonprobability-surveys. 10 to the small number of transgender (57 respondents after weighting, 43 before weighting) and intersex respondents30 (54 after weighting, 53 before). Table 1.1.1. Sample Size by LGBTI Subgroup, Unweighted and Weighted Data Number Percentage Number Percentage LGBTI group (unweighted) (unweighted) (weighted) (weighted) Intersex 53 5 54 5 Transgender 43 4 57 6 Lesbian 164 16 190 19 Gay 512 51 349 34 Bisexual men 111 11 138 14 Bisexual women 115 12 226 22 TOTAL 998 100 1014 100 35. Additional steps were taken to draw more accurate comparisons between the LGBTI sample and the sample of the general population used in the SILC. The SILC in Serbia was weighted to the target demographics in the LGBTI sample. Subsequently, a sample matching procedure was conducted using a covariate balancing propensity score to allow a comparison between the two samples. 31 The matching procedure took the following details into account: age, sex assigned at birth, educational attainment, marital status, region, and area of residence (urban or rural). A weight is estimated from this matching process that can be used for further analysis. The final weight incorporates both LGBTI target demographics and the weights as a result of the matching process. An example is provided in figure 1.1.1, which shows the age distribution of the LGBTI sample and the general population before and after matching and weighting. The LGBTI sample is far younger than the general population prior to matching and weighting, but after the process, there are few differences between the two samples. In addition, figure 1.1.1 shows that the general population sample is adjusted to match that of the LGBTI sample and that the age distribution of the LGBTI sample is unaffected by the matching process. More detailed information on the weighting and matching process is provided in Annex 2. 30 LGBTI organizations felt that the number of respondents who identified themselves as intersex, 53, was higher than what they expected based on their previous involvement with intersex Serbians. The organizations were unsure about the general level of awareness of the terms relating to intersex and sex characteristics. Nevertheless, responses from all respondents who self-identified as intersex were included in the data set used for analysis in this report. 31 Imai Kosuke and Mark Ratkovic, “Covariate Balancing Propensity Score,” Journal of the Royal Statistical Society, Series B 76, Part 1 (2014): 243–63. 11 Figure 1.1.1. Age before and after Matching BEFORE MATCHING LGBTI Sample 55 40 5 General Population 32 37 31 AFTER MATCHING LGBTI Sample 55 40 5 General Population 54 41 5 0% 20% 40% 60% 80% 100% 17-29 30-44 45-67 36. The LGBTI SILC used a set of six self-reporting questions to determine the respondent’s sexual orientation, gender identity, and/or intersex status. The questions were modeled based on international practices32 and covered elements of self-identification, sexual attraction, sexual behavior, and perception. Every person has a sexual orientation (e.g., LGB), a gender identity (e.g., cisgender male, cisgender female, transgender male, transgender female, and gender non-binary), and sex characteristics. However, for the purposes of this report, six mutually exclusive categories of analysis were created: lesbian, gay, bisexual male, bisexual female, transgender, and intersex.33 Each respondent can belong to only one of these categories to ensure results are comparable. Respondents who identified as intersex 34 or transgender35 will only appear as such in the analysis. This analytical categorization should not lead to the conclusion that intersex and transgender people do not have a sexual orientation. For this analysis, the assumption is that the lived experience of a transgender and intersex person is significantly different from that of lesbian, gay, and bisexual people. The full data set is available here36. 37. In summary, the demographics of the respondents are as follows: 37 • Sex: Slightly more than half (53 percent) of LGBTI respondents were assigned male sex at birth and 47 percent were assigned female sex at birth. For the unmatched general population, 51 percent of the respondents are female, and 49 percent are male.38 32 SMART guide for questions on sexual orientation and GenIUSS group recommendations on questions for gender minorities. 33 We also note that there are other (interchangeable and additional) words that people use that are not directly reflected in t he “L,” “G,” “B,” “T,” and “I,” which has led to the more recent adoption of “LGBTI+” (along with many other formulations). We celebrate this diversity and fluidity, but given the empirical focus of this study, we limited ourselves to “LGBTI” as this resonates with other resea rch in the region and creates international comparability. 34 Question A3_B: Some people are assigned male or female at birth but are born with sexual anatomy, reproductive organs, and/or chromosome patterns that do not fit the typical definition of male or female. This physical condition is known as intersex. Are you intersex? 35 Respondents were identified as transgender if they self-identified as such in question A3_A or if their sex assigned at birth was different from the gender identity they currently identify with. 36 https://microdata.worldbank.org/index.php/catalog/3486 37 See Annex 3 for a detailed demographic analysis. 38 Statistical Office of the Republic of Serbia, “2011 Census of the Population, Households and Dwellings in the Republic of Serbia” (Belgrade, Republic of Serbia, 2013) (in Serbian). 12 • Age: LGBTI respondents are significantly younger than the unmatched general population. The average age among LGBTI respondents is 30 (29.6 years), and the unmatched general population is on average 19 years older (48.8 years). o LGBTI respondents whose sex was assigned female at birth are on average much younger than those whose sex was assigned male at birth. Hence, the LGBTI sample contains a higher percentage of young (up to 25 years old) bisexual and lesbian women than bisexual and gay men. Transgender respondents are the youngest group in the sample: 60 percent are below age 26. Intersex respondents, on the other hand, are on average among the oldest. o Difference in age among LGBTI subgroups is one of the possible explanations for some of the differences observed throughout the report. At times, age-adjusted estimates are also presented to account for this imbalance. • Education: LGBTI respondents are not only younger than the unmatched general population but are also more educated. A large percentage of the general population (32 percent) have only primary or incomplete secondary education, yet the percentage is much smaller (3 percent) for LGBTI respondents. In fact, most LGBTI respondents are highly educated (57 percent have graduated from college or have a higher level of education), while only 17 percent of the unmatched general population have a college degree or higher. o Transgender respondents are less likely to have college or higher education degrees (34 percent) compared to other LGBTI subgroups. Otherwise, no differences among LGBTI subgroups are noticeable. • Place of living: Almost all LGBTI respondents live in urban and peri-urban areas, and only 4 percent live in rural areas. Among the unmatched general population, 59 percent live in urban areas while the rest, 41 percent, live in rural areas. o The LGBTI sample is also skewed toward those who live in Belgrade. One out of every two LGBTI respondent (53 percent) lives in the capital, while only 23 percent of the unmatched general population live there. • Household composition: LGBTI households are quantitatively and qualitatively different from households of the general population. Although 25 percent of LGBTI respondents live alone, this is the case for only 2 percent of the unmatched general population. Intersex and gay respondents more frequently live alone compared to other LGBTI groups (34 and 33 percent, respectively), though this is likely related to the age differences among the subgroups. LGBTI subgroup differences are diminished when examining the age-adjusted percent who live alone, though the percentage of intersex and gay respondents living alone remains higher than the average (29 percent for both). o Overall the average size of LGBTI households is 2.8 members, which is significantly smaller than the size of the average unmatched general population household (four members). o LGBTI respondents also live much less frequently with their parents39 compared to the unmatched general population. o Twelve percent of the LGBTI respondents live with their unmarried partner 40 compared to only 7 percent of respondents from the unmatched general population. 39Living with one or both parents does not exclude living also with an unmarried partner. 40Living with an unmarried partner does not exclude living also with parents. Only unmarried partners are taken into considerat ion, given that same-sex partnerships are not recognized in Serbia. 13 Lesbians and bisexual women are almost twice as likely to live with their unmarried partner compared to other LGBTI groups. An experimental study conducted by the World Bank may explain these patterns: in a mystery shopper experiment, same-sex female couples were less likely to be discriminated against during the search for an apartment compared to same-sex male couples.41 Most transgender respondents, likely because of their young age, live with their parents. Age-adjusted percentages, however, still suggest that transgender, intersex, and bisexual male respondents are most likely to be living with their parents. Homeownership status of LGBTI Respondents 38. Potential LGBTI homeowners face unique challenges limiting their ability to purchase homes. The role of the family is important, as homeownership is often passed down from generation to generation.42 Families struggling to accept their LGBTI children might therefore exclude them from this process.43 Same-sex couples certainly face challenges in jointly accumulating assets due to the lack of a formal partnership status in Serbia, which in turn could limit their access to credit or a mortgage and increase the cost of buying a home.44 Homeownership is an important asset, especially during retirement. 39. Significantly fewer LGBTI respondents are homeowners compared to the general population. Among the LGBTI respondents, 66 percent own the dwelling they live in, either as a full- fledged owner or through a mortgage, compared to 86 percent of the general population. LGBTI respondents are more likely to be renting their current home (26 percent) compared to the general population (5 percent) (see figure A.4.1 in Annex 4). 40. LGBTI homeowners and renters are more likely to have had difficulties making their payments in the past 12 months compared to the general population. The percentage of LGBTI respondents who report a difficulty making their mortgage payment (and who are actively repaying their mortgage) is 23 percent compared to 3 percent in the general population. Similarly, when examining LGBTI renters, the number who report difficulties is far higher than in the general population. The percentage of LGBTI renters who report a difficulty making their rent payment is 30 percent compared to 3 percent in the general population. 41. From the above, it is clear that the demographics of the average LGBTI respondent are significantly different from the profile of the average Serbian citizen. After matching both samples based on age, sex assigned at birth, educational attainment, marital status, region, and area of 41 Koehler and others, “Discrimination Against Sexual Minorities in Education and Housing.” 42 For example, Mirjana Bobić, “Applying the Concept of Social Capital in Understanding Marriage and Family in Contemporary Serbia,” in Social and Cultural Capital in Western Balkan Societies, ed. Predrag Cvetičanin and Ana Birešev (Belgrade: Centre for Empirical Cultural Studies of South-East Europe, University of Belgrade, 2012), 41–58. 43 For example, Judit Takács, “Social Exclusion of Young Lesbian, Gay, Bisexual and Transgender (LGBT) People In Europe” (Brusse ls and Amsterdam: the European Region of the International Lesbian and Gay Association and the International Lesbian, Gay, Bisexual, Transgender, Queer Youth and Student Organisation, 2006), https://www.salto-youth.net/downloads/4-17- 948/ReportSocialExclusionIGLYOilga.pdf. 44 For example, in the United States, see M. V. L. Badgett, Money, Myths, and Change: The Economic Lives of Lesbians and Gay Men (Chicago: University of Chicago Press, 2003). 14 residence, the demographic profiles were almost identical, thus allowing for a comparison of economic outcomes and living conditions. 42. Throughout the report, the LGBTI respondents are compared to the matched general population at the national level (unless otherwise stated).45 1.2. Limitations of the study 43. Because there are no data about the LGBTI population and its demographic structure in Serbia, it was not possible to construct a representative sample. Thus, the survey was conducted with a self-selected, nonprobability sample. The survey is therefore not statistically representative of the total LGBTI population; instead it represents the opinions of 998 people who identified themselves as lesbian, gay, bisexual, transgender, or intersex and who were willing to participate. Special attention should be paid to the small number of transgender respondents (57 respondents after weighting, 43 respondents before weighting) and intersex respondents (54 after weighting, 53 before). The sample sizes may be too small to make reliable conclusions on those two groups. 44. An online survey was determined to be the most appropriate method of data collection. It was important to select a data collection method that was likely to yield the greatest number of responses and still protect the identity and privacy of the LGBTI respondents. The survey was thus available only to those who have access to the internet, with the exception of a small number of participants who completed paper versions. As noted, the data indicate that the survey reached respondents who, compared to the general population of Serbia, were younger, had relatively higher levels of education, and tended to live in urban areas, especially Belgrade. The survey likely failed to reach the most marginalized LGBTI people in Serbia, such as those who live in rural areas, are homeless, have disabilities, or are Roma. These patterns may be a consequence of an online survey. Respondents were mainly recruited by nongovernmental organizations (NGOs), leading to a larger number of responses from people who work for NGOs than might otherwise have been expected. Interpreting the Results The results presented in this report represent only those LGBTI people in Serbia who took the online survey. Some of the most vulnerable LGBTI people may not have been reached due to a lack of access to the internet and to the networks used to recruit participants. Because of the uniqueness of the LGBTI sample, which, compared to the general population of Serbia, was younger, had relatively higher levels of education, and was more urban (including especially people from Belgrade), comparisons were drawn to a demographic subgroup of the general population that had similar characteristics to the LGBTI people who responded to the survey. 45. Questions on sexual orientation, gender identity, and/or intersex status are highly sensitive, and disclosing such information can put respondents in danger. Collecting data from LGBTI people 45We recognize that as subset of the respondents to the SILC general survey are LGBTI but as the official SILC does not ask about sexual orientation, we have no way of estimating the percentage of LGBTI respondents in that survey sample. 15 can therefore be very challenging. Standard household surveys are not ideal for collecting information from LGBTI people who often fear discrimination, exclusion, and violence.46 Distinct data collection efforts tailored to reach this population are thus necessary to fill the LGBTI data gap. The method of data collection is substantially different from the original SILC as outlined in the previous section. As a result, the LGBTI SILC reached a distinct subset of the population that differs significantly from the sample of the general population SILC. The following analysis of the survey results and comparisons to the general population should therefore be considered in light of these limitations, even with the subsequent matching strategy. 2. INCOME AND ECONOMIC WELL-BEING 46. Data on the lived reality of LGBTI people are scarce, particularly on their economic situation. Examining the incomes and economic well-being of LGBTI people is important to understand the hardships they face and their general standard of living. The LGBTI SILC examined the personal and household incomes of LGBTI people, standardized measures of household income and material deprivation, and subjective measures of economic well-being. In this way, a fuller assessment beyond income could be performed. The outcomes of LGBTI people are compared to those of the general population, allowing policy makers, CSOs, and development partners in Serbia to better understand how to foster the inclusion of LGBTI people in key markets, services, and spaces. Key terms used in this section • Personal income: The average monthly gross earnings of an individual from all income sources. • Household income: The combined average monthly gross earnings of all individuals residing in the same household. • Equivalized disposable income: An estimate of household income based on the household’s total income divided by the number of members of the household, weighted according to their status and age. • At-risk-of-poverty rate: The share of people with an equivalized disposable income below 60 percent of the national median equivalized income, which is calculated at 14,920 Serbian dinars (about €120). • Financial burden: The inability to pay monthly payments relating to mortgage, rents, or other debts for at least one month in the past 12 months. • Material deprivation: A standard concept developed by Eurostat that relates to a state of economic strain. It is the inability to pay for certain expenses, namely, (1) unexpected expenses; (2) a one-week annual holiday away from home; (3) a meal involving meat, chicken, or fish every second day; (4) the adequate heating of a dwelling; or durable goods like a (5) washing machine, (6) color television, (7) 46Dominik Koehler and Nicholas Menzies, “Surveys, Big Data, and Experiments: How Can We Best Learn about LGBTI Development Outcomes?” (Washington, DC: World Bank, 2017), http://documents.worldbank.org/curated/en/828531501521958802/Surveys-big-data-and-experiments-how-can-we-best-learn-about- LGBTI-development-outcomes. 16 telephone, or (8) car; or being confronted with (9) payment arrears (mortgage or rent, utility bills, hire purchase installments, or other loan payments). • Materially deprived: The percentage of the population that cannot afford at least three of the nine items. • Severely materially deprived: The percentage of the population that cannot afford at least four of the items. • Extremely materially deprived: The percentage of the population that cannot afford at least five of the items. Section highlights • The median personal monthly income of LGBTI respondents (€256) was slightly higher than, but statistically indistinguishable from, the comparable population (€236). • Age, sex assigned at birth, and level of education are the primary factors that influence the monthly income of both LGBTI respondents and the comparable population. • Respondents whose sex assigned at birth was male who self-report that they are perceived by others as feminine, as well as respondents whose sex assigned at birth was female who self-report that they are perceived as masculine, earn lower median monthly incomes (€244 and €203, respectively) than those perceived as gender conforming (€325 for respondents assigned male at birth and €244 for respondents assigned female at birth). • LGBTI respondents who have experienced discrimination earn a lower income (€236) compared to those who have not been discriminated against (€260). • Although personal monthly incomes are slightly higher for the LGBTI sample than for the comparable population, median household income is lower in LGBTI households (€489) than in those of the comparable population (€628). • The at-risk-of-poverty rate among LGBTI respondents is 17 percent, which is similar to the rate for the comparable population (16 percent). The rate for the LGBTI sample does not decrease for those with higher levels of education. The at-risk-of- poverty rate of LGBTI respondents without a college degree is 17 percent (23 percent in the comparable sample) and 18 percent for those with a college degree (12 percent in the comparable sample). The rate of LGBTI respondents over 35 is almost twice (17 percent) that of the comparable population (9 percent). • Rates of material deprivation are higher among transgender (40 percent) and intersex respondents (33 percent), older respondents (27 percent of those over age 45), and respondents with less education (32 percent of those without a college degree). • Rates of material deprivation are higher for LGBTI respondents if they experience workplace discrimination (52 percent) than if they do not (25 percent). 17 47. It should be noted that in addition to the general limitations of the methodology highlighted in section 1.2, there are specific limitations that could affect the results documented in this section also. First, the LGBTI SILC questionnaire is self-reported, while the EU SILC is administrated by interviewers who are trained to collect data on all possible sources of income from the respondents. Second, the method of data collection on the income of LGBTI respondents also differs from the method in the original EU SILC, in which all individuals over the age of 16 in a household can respond to the survey, and personal incomes are combined to make up the household income. In the LGBTI SILC on the other hand, instead of a range of questions on both household and personal income, there was only one question on each. Third, dissemination of the survey relied heavily on NGOs, which likely contributed to the much larger share of respondents who actually work for NGOs (10 percent) than the comparable sample (0 percent). Median incomes are reported,47 which is standard for an analysis of incomes because mean estimates are sensitive to the presence of outliers. Figure 2.1 shows the high prevalence of outliers in both the LGBTI sample and the general population, which suggests that income is highly skewed. Indeed, the skewness 48 of income is 5.2 in the LGBTI sample and 6.5 in the comparable population. Figure 2.1. Box-and-Whisker Plot of Monthly Personal Income (in Serbian dinars) 47 For example, the income gap between the LGBTI sample and the comparable sample is €169 when using the average and €22 when using the median of both samples. 48 Skewness is a statistical metric to determine whether a random variable is distributed evenly about the mean. A normal distribution, for example, has a skewness of zero. 18 2.1 Personal income 48. The median monthly income of LGBTI respondents in Serbia was €25649 compared to €236 for the general population, which does not significantly differ at the 95 percent confidence level.50 Generally, respondents in the LGBTI sample have slightly higher personal incomes than those in the comparable population across most demographics: age, gender, and education. LGBTI respondents who reside in the central region of Sumadija and west Serbia have lower median monthly incomes (€220) than the general population (€226), and LGBTI respondents who reside in south and east Serbia have lower median monthly incomes (€179) than the general population in those areas (€202). LGBTI respondents between the ages of 18 and 25 do not differ in their incomes ( €173) from the general population (€177). The monthly income of LGBTI respondents and the general population is influenced by similar factors such as sex assigned at birth,51 age, and level of education. 49. For both LGBTI respondents and the general population, individuals assigned male sex at birth earn a higher monthly income than individuals assigned female at birth. LGBTI respondents assigned male sex at birth have a significantly higher personal income (€302) than men in the general population (€242). The personal income of LGBTI respondents assigned female sex at birth (€229) is lower than that of both LGBTI respondents assigned male sex at birth and men in the general population but slightly higher than for women in the general population (€202). The gender income gap in the LGBTI sample is 28.9 percent and for the comparable population 20.5 percent. The income gap between men and women could in part also be influenced by age, given that most women in the LGBTI sample are younger than the men,52 and as highlighted above, both younger LGBTI respondents and younger respondents in the general population report lower incomes. That said, the wide gap in income is influenced not only by age but also sex assigned at birth. The income gap, for example, diminishes slightly adjusting for age. The median personal income for men is €317 and for women €268 adjusting for age, which reduces the gap from €82 to €49. 49 Income in the LGBTI SILC includes other earnings besides salary, i.e., income from travel, bonuses, and earnings generated on the basis of copyright or similar contracts; from self-employment, including activities such as giving private lessons; from renting an apartment/house; from profits that have accrued on any grounds; and/or from a pension. 50 Income in the original SILC includes employee cash or near cash income, non-cash employee income, and cash benefits from self- employment. For greater details on coding income, see Annex 4. 51 Sex assigned at birth only reflects differences between people whose sex was assigned female or male; it does not refer to differences based on current gender identity. 52 As already noted in Section 1.1, respondents who were assigned female sex at birth are on average much younger than those who were assigned male sex at birth. 19 Figure 2.1.1. Monthly Personal Income (median) by Education and Residence LGBTI sample General population Median 256 236 Female 229 202 Male 302 242 Aged 18-25 173 177 Aged 26-35 292 202 Aged 36 and higher 407 299 Primary or secondary school education 211 168 College or Higher education 316 266 Big city 285 242 Other 201 194 Belgrade 286 268 Vojvodina 260 193 Sumadija and West Serbia 220 226 South and East Serbia 179 202 0 100 200 300 400 Question: Please state the average total monthly amount of your income in the previous 12 months and the number of months during which you had this income? Base: Those LGBTI respondents and respondents from the general population who reported income/earnings during the past 12 months. 95 percent confidence intervals are represented by the error bars. 50. As expected, age, education levels, and geographic location influence monthly incomes. Those aged 18–25 earn a median salary of €173 for the LGBTI sample and €177 for the comparable population. Respondents over 35 earn the highest monthly income: €407 among LGBTI respondents and €299 for the comparable population. In both samples, those with a college degree or higher education report higher median monthly incomes compared to those with primary or secondary school education. Those who live in urban areas (big cities) earn more than those who live in rural areas. For the LGBTI sample as well as the general population, those who earn the highest median monthly income live in Belgrade: €286 and €268, respectively. 20 51. LGBTI respondents in the private sector earn a lower monthly income (€325) compared to those in the public sector (€390). In the general population, those who work in the public sector also earn more (€325) than those who work in the private sector (€244). Figure 2.1.2. Monthly Personal Income (median) by Workplace LGBTI sample General population 256 Median 236 356 Employee 263 407 Self-Employed 203 325 Private sector 244 390 Public (state-owned) sector 325 244 Nongovernmental sector 358 Contract for indefinite period of time 301 256 Contract for definite period of time 164 0 100 200 300 400 Question: Please state the average total monthly amount of your income in the previous 12 months and the number of months during which you had this income? Base: Those LGBTI respondents and respondents from the general population who reported income/earnings during the past 12 months. 52. Within LGBTI subgroups, there are material differences in personal income, with bisexual and gay men earning the most and transgender people the least. When adjusting for age, differences between lesbians, bisexual women, and bisexual men decrease, though their median personal income remains lower than that of gay men (see figure 2.1.3). Transgender people still earn the lowest median monthly income, but the income gap between gay men and transgender people decreases by about 25 percent after adjusting for age. 53. In addition to the gender gap, the perception of gender has a major impact on income. Those LGBTI respondents who are not perceived by others as being in accordance with their sex assigned at birth report on average much lower median monthly incomes. Even though the gender gap is present here as well (i.e., LGBTI respondents assigned male sex at birth earn more than those assigned female sex at birth), it appears that perceived gender conformity has a significant impact on income. 21 Figure 2.1.3. Personal Monthly Income and Age-Adjusted Personal Monthly Income (median) Personal monthly income Age-adjusted personal monthly income LGBTI median 256 256 Gay 325 322 Bisexual men 285 287 Intersex 260 276 Lesbian 244 264 Bisexual women 220 264 Transgender 203 229 Men perceived as masculine 325 313 Men perceived as feminine 244 273 Women perceived as feminine 244 294 Women perceived as masculine 203 241 Question: Please state the average total monthly amount of your income in the previous 12 months and the number 35 0 105 7 140 0 of months during which you had this income? 245 175280 210 350 315420 455 385490 560 595 525 700 665 630 770 735 Base: Those LGBTI respondents who reported income/earnings during the past 12 months. 54. LGBTI respondents who had experienced workplace discrimination in the past five years report a lower median personal income than those who had not experienced discrimination, though the difference was not statistically significant. The income of LGBTI respondents who had experienced discrimination was €24 lower; they earned a median personal income of €236, while those who had not experienced discrimination earned slightly more at €260. Discrimination in the workplace that results in unequal treatment with respect to employment conditions, such as benefits, pensions, and income, has particularly negative impacts that affect the economic prospects of LGBTI people, which in turn can affect other critical needs such as access to health care and other essential services. Since experience of workplace discrimination is related to a respondent’s age, sex assigned at birth, and LGBTI subgroup membership, taking these factors into account can clarify the magnitude of the effects of discrimination on personal monthly income. If median personal income is adjusted for age, sex assigned at birth, and LGBTI subgroup, the gap between those who experienced workplace discrimination and those who did not widens to €40, which is statistically significant. 22 55. The income distribution of the LGBTI sample is much broader than for the comparable general population. Fewer LGBTI respondents (25 percent) report incomes in the lowest income bracket (up to €160) compared to the general population (36 percent). At the other end of the distribution, a higher number of LGBTI respondents (19 percent) report an income in the highest two income brackets (over €500) compared to the general population (8 percent), as shown in figure 2.1.453 Indeed, 4 percent of the LGBTI sample report incomes higher than €1,250 a month compared to 0.3 percent of the comparable sample. Figure 2.1.4. Personal Monthly Income (percent) LGBTI sample General population 100 80 60 55 57.0 40 36.0 25 20 15 7.0 4 0.3 0 Up to 160 EUR 160 - 500 EUR 500 - 1250 EUR 1250 EUR and more Base: Total LGBTI sample and the general population. 2.2 Household income 56. The median household income for the LGBTI sample ( €489) is less than for the comparable sample of the general population ( €628), and the difference is statistically significant. Although personal income is slightly higher than the comparable sample, household income is not. Unlike the in-person administration of the official SILC, the online format of the LGBTI survey means that LGBTI people from the same household could participate in the survey, which has the potential to double- count household income.54 53 The income scale used in the survey is the same as the scale used in the original SILC: up to 20,000 dinars; 20,001 –25,000; 25,001–35,000; 35,001–45,000; 45,001–60,000; 60,001–80,000; 80,001–100,000; 100,001–150,000; 150,001–200,000; and 200,001 dinars or more. 54 Given that empirically, the LGBTI sample earns a slightly higher median personal income, any double-counting of household incomes would lead to household incomes that are higher than the comparable population. Since this is not the case and since LGBTI people more often live alone, there is less of a concern about the double-counting of some households in the estimate of household incomes. 23 Figure 2.2.1. Monthly Household Income and Equivalized Disposable Household Income (median) LGBTI sample General population 489 Household income 628 Equivalized disposable 275 income of the household 261 0 200 400 600 800 Base: Total target LGBTI sample and the general population. 95 percent confidence intervals are represented by the error bars. 57. The median equivalized disposable monthly household income is €275 in the LGBTI sample compared to €261 in the general population, which is not statistically significant. The equivalized household disposable income is based on the household’s total income divided by the number of members of the household, weighted according to their status and age. 55 Values estimated in such a manner are called equivalized disposable income and are later used for poverty indicators such as the at-risk-of-poverty rate. LGBTI households have a household income of less than €200 at rates similar to the general population (33 and 34 percent, respectively), but 2 percent of LGBTI households have a monthly household income of more than €1,650 compared to 0 households in the general population. 2.3 At-risk-of-poverty rate 58. There is no significant difference in the at-risk-of-poverty rate between LGBTI respondents and the comparable general population . The at-risk-of-poverty rate is the share of people with an equalized income below 60 percent of the national median equivalized income (figure 2.3.1).56 For this report, the at-risk-of-poverty threshold57 is calculated at 14,920 dinars (about €120).58 59. Factors such as age and education that reduce the at-risk-of-poverty rate in the general population do not reduce the rate in the LGBTI sample. In the general population, persons in the youngest age group have the highest rate (21 percent), which declines to 9 percent for individuals over 35 years of age. The LGBTI respondents, however, have similar at-risk-of-poverty rates across age 55 This scale assigns a value of 1 to the household head, 0.5 to each additional adult member (14 years and older), and 0.3 to each child (below 14 years old). Eurostat, “Glossary: Equivalised Disposable Income,” Statistics Explained (Luxembourg: Eurostat, 2018), http://ec.europa.eu/eurostat/statistics-explained/index.php/Glossary:Equivalised_disposable_income. 56 Eurostat, “Glossary: At-Risk-of-Poverty Rate,” Statistics Explained (Luxembourg: Eurostat, 2018)m http://ec.europa.eu/eurostat/statistics-explained/index.php/Glossary:At-risk-of-poverty_rate. 57 The at-risk-of-poverty rate in 2015 at the national level was 25.4 percent. However, as mentioned at the beginning of the paper, characteristics of LGBTI respondents are in line with the more affluent part of the population, and though it is expected that the at-risk-of- poverty rate will be lower than the national rate, for the purposes of comparison we used the matched population. 58 Statistical Office of the Republic of Serbia, “Statistical Release: Poverty and Social Inequality in Republic of Serbia in 2015” (Belgrade: Republic of Serbia, 2016), no. 84: 1–5. 24 groups, and the rate for LGBTI respondents over 35 is almost twice that of the general population (17 percent). The at-risk-of-poverty rate for the LGBTI respondents does not decrease for those with a higher level of education as it does in the general population. The rate for LGBTI respondents without a college degree is 15 percent (23 percent in the general population) and for those with a college degree it is 19 percent (12 percent in the general population). Place of residence is associated with at- risk-of-poverty rates in similar ways for both LGBTI respondents and the general population—living in urban areas reduces the rate in both samples. Additionally, people living in Belgrade are at a lower risk of poverty compared to other regions of the country, especially relative to south and east Serbia, where LGBTI respondents are at greater risk of poverty compared to those from the general population. Still, the at-risk-of-poverty rate for LGBTI respondents in Belgrade (16 percent) is more than twice that of the general population (7 percent). Figure 2.3.1. At-Risk-of-Poverty Rate by Main Demographic Characteristics (percent) LGBTI sample General population Aged 18-25 16 21 Aged 26-35 18 16 Aged 36 and higher 17 9 Primary or secondary school education 15 23 College or higher education 19 12 Living in big city 15 16 Not living in big city 22 18 Belgrade 16 7 Vojvodina 15 19 Sumadija and West Serbia 20 17 South and East Serbia 26 23 0 10 20 Base: Total LGBTI sample and the general population. 60. Self-employed persons in the general population and LGBTI respondents who are independent contractors have higher at-risk-of-poverty rates. The rate is higher for employed LGBTI respondents (14 percent) than for the general population (8 percent) (see figure A.4.2 in Annex 4). LGBTI respondents working in the public sector have a much higher rate (20 percent) compared to public sector employees in the general population (5 percent). No differences can be observed for the private sector. 61. LGBTI respondents who are perceived by others to be gender non-conforming have higher at-risk-of-poverty rates than those perceived to be gender conforming. LGBTI respondents who are not perceived by others to be in accordance with their sex assigned at birth have higher rates (assigned female sex at birth: 22 percent; assigned male sex at birth: 18 percent) compared to those who are perceived by others to be in accordance with their sex assigned at birth (assigned female sex at birth: 17 percent; assigned male sex at birth: 16 percent). Transgender respondents have the highest at-risk- 25 of-poverty rate (24 percent) among the LGBTI subgroups. An age-adjusted rate for the LGBTI subgroups does not change these patterns.59 2.4 Financial burden 62. Surveyed LGBTI households face greater difficulties making household payments on time compared to the general population. LGBTI respondents most often report difficulties making utility payments (40 percent) compared to the general population (31 percent), as well as rent payments (11 percent) and other credits (21 percent), which are 1 and 0 percent, respectively, in the general sample (see figure A.4.3 in Annex 4). A larger share of LGBTI respondents (45 percent) have difficulty making payments in any area relative to the general population (35 percent). LGBTI households more often face difficulties in maintaining their financial commitments given their lower overall household income, despite relatively similar equivalized household income compared to households in the general population and their slightly higher personal incomes. When examining LGBTI respondents who are currently either paying their mortgage and or actively renting, the number who report difficulties is far higher for LGBTI people than the initial response distribution suggests (figure 2.4.2). The percentage of LGBTI respondents who report a difficulty making their mortgage payment and who are actively repaying their mortgage is higher (23 percent) than the initial response set (6 percent), and this percentage is far greater than in the general population (3 percent). Similarly, the percentage of LGBTI respondents who report a difficulty in making their rent payments and who are actively renting is higher (30 percent) than the initial response set (11 percent), and this percentage is also much greater than in the general population (3 percent). Figure 2.4.2. Arrears on Mortgage or Rent Payments among Homeowners Still Paying their Mortage or Renters, respectively (percent) Yes Yes No No Does Not Apply to Me Does Not Apply 23 the whole or paying off LGBTI sample mortgage 67 10 Yes, but the General population 3 17 80 a part of the LGBTI sample 30 Lessee of dwelling 65 5 General population 3 20 77 0% 20% 40% 60% 80% 100% Question: In the previous 12 months, were you in a situation of being late with the payment of costs or the repayment of credits owing to financial difficulties? Base: Total LGBTI sample. 2.5 Material deprivation and economic well-being 59Age-adjusted estimates are the following: transgender, 25 percent; bisexual men, 18 percent; bisexual women, 18 percent; lesbians, 18 percent; gay men, 14 percent; and intersex, 13 percent. 26 64. To assess the economic well-being of households, material deprivation is an important indicator. Material deprivation refers to the affordability of products and not the possession of any particular product, since the possession may be affected by various factors, including but not limited to personal needs and decisions. Material deprivation is a standard concept developed by Eurostat that relates to a state of economic strain;60 it is defined as the enforced inability to possess certain durables (rather than the choice not to do so—see “Key Terms” above).61 65. There are no substantial differences between the LGBTI sample and the general population with regard to their material deprivation across the three categories. A little over one-quarter of LGBTI respondents (28 percent) and the general population (27 percent) are materially deprived, and about 14 percent of LGBTI respondents and 13 percent of the general population are severely materially deprived (figure 2.5.1). A slightly higher percentage of LGBTI respondents (7 percent) are extremely materially deprived compared to the general population (5 percent). Figure 2.5.1: Percentage of Materially Deprived, Severely Materially Deprived, and Extremely Materially Deprived (percent) LGBTI sample General population 28 Materially deprived 27 14 Severely materially deprived 13 7 Extremely materially deprived 5 0 10 20 30 Base: Total LGBTI sample and the general population. 66. Discrimination increases the likelihood of material deprivation among LGBTI respondents. LGBTI respondents who experienced workplace discrimination in the past five years are far more often materially deprived, severely materially deprived, or extremely materially deprived. The rates of material deprivation are two times higher for LGBTI respondents who experienced workplace discrimination in the past five years (52 percent) than for those who did not (25 percent)(figure 2.5.2). The rates of severe material deprivation are also two times higher among LGBTI respondents who experienced workplace discrimination (28 percent) than among those who did not (12 percent). 60 This refers to the ability to afford: unexpected expenses; a one-week annual holiday away from home; a meal involving meat, chicken, or fish every second day; the adequate heating of a dwelling; durable goods like a washing machine, color television, telephone, or car; and payments, such as mortgage or rent, utility bills, hire purchase installments, or other loan payments. 61 Eurostat, “Glossary: Material Deprivation,” Statistics Explained (Luxembourg: Eurostat, 2018), http://ec.europa.eu/eurostat/statistics- explained/index.php/Glossary:Material_deprivation. 27 Extreme material deprivation is almost three times higher among LGBTI respondents who experienced workplace discrimination in the past five years (16 percent) than among the remaining LGBTI respondents (6 percent). Figure 2.5.2. Material Deprivation, Severe Material Deprivation, and Extreme Material Deprivation, by any Workplace Discrimination (percent) No Workplace Discrimination Experienced Workplace Discrimination Materially deprived 25 52 Severely materially deprived 12 28 Extremely materially deprived 6 16 Question: In the past five years, have you personally experienced inequalities at your workplace based on your sexual orientation/gender identity/sex characteristics? Base: Total LGBTI sample. 67. Older LGBTI people and transgender and intersex respondents stand out especially as materially deprived. The deprivation rate decreases with an increase in education and is significantly lower among those living in and around Belgrade and other big cities. Lesbian respondents tend to have the lowest rates of material deprivation, severe material deprivation, and extreme material deprivation in comparison to other LGBTI subgroups. 28 Table 2.5.1. Material Deprivation, Severe Material Deprivation, and Extreme Material Deprivation Rates among LGBTI Respondents, by main background characteristics Severely Extremely Materially Number of materially materially deprived respondents deprived deprived (percent) (unweighted) (percent) (percent) Total 28 14 7 998 Age 18–25 30 15 6 342 26–35 28 15 6 394 36–45 23 12 9 209 46 and higher 27 20 16 53 Highest level of education Secondary school (3 or 4 years) or 32 18 10 411 less College or higher education 24 12 5 579 Refuses to answer (*) (*) (*) 8 Area Big city (150,000 inhabitants or 24 13 5 714 more) Other 39 19 11 284 Region Belgrade 24 13 5 527 Vojvodina 31 14 7 269 Central Serbia 38 17 9 202 LGBTI status Intersex 33 19 11 53 Transgender (40) (18) (10) 43 Lesbian 26 9 3 164 Gay 28 15 7 512 Bisexual women 29 17 8 115 Bisexual men 23 14 8 111 ( ) Figures that are based on 25–49 unweighted cases. *Figures that are based on fewer than 25 unweighted cases. 68. Outside of the material deprivation indicator, more LGBTI respondents cannot afford certain goods or activities compared to the general population. LGBTI respondents, regardless of the item or activity, more often state that those items or activities are not affordable compared to the general population. LGBTI people’s lower reported household income seems to relate to rates of material deprivation (see figure A.4.4 in Annex 4). 29 3. LGBTI PEOPLE IN THE LABOR MARKET 69. Employment is a key factor in the quality of life and economic well-being of individuals. A lack of robust, quantitative data makes it difficult to understand the impacts of labor market discrimination on the socioeconomic outcomes of LGBTI people. For the first time, this survey collected data on LGBTI people in the Serbian labor market that allow for a comparison of labor market outcomes of LGBTI Serbians and their non-LGBTI counterparts. Key terms used in this section • Percent unemployed: The percent of respondents who are unemployed out of all respondents. • Percent employed: The percent of respondents who are employed out of all respondents. • Employed workers: Individuals who are currently working for pay. • Unemployed workers: Individuals who are currently not working but are willing and able to work for pay, presently capable of working, and actively searching for employment. This excludes individuals who are not actively seeking employment, such as students. • Active persons: Individuals who are employed workers or unemployed workers (also referred to as the total labor force). • Inactive persons: Individuals who are currently not working, not willing or able to work for pay, presently incapable of working, and not actively searching for employment. This includes individuals who are not actively seeking employment, such as students. • Activity rate: The percent of active persons divided by the combined sum of active and inactive persons. • Employment rate: The percent of employed workers divided by the combined sum of active and inactive persons. • Unemployment rate: The percent of unemployed workers divided by the number of active persons. • Inactivity rate: The percent of inactive persons divided by the combined sum of active and inactive persons. • Long-term unemployment: When an unemployed worker has been unemployed for longer than a year. 30 Section highlights • LGBTI respondents are less likely to be unemployed (11 percent) and more often still students (30 percent) compared to the general population (30 and 17 percent, respectively). • The employment rate of the LGBTI respondents (56 percent) and the general population (50 percent) is similar. • The unemployment rate for the LGBTI respondents is 14 percent, which is less than half of the unemployment rate for the comparable general population (33 percent). • LGBTI respondents are significantly less often employed in the public sector (20 percent) compared to the general population (36 percent), but they more often work for NGOs (10 percent) than existing data suggest for the general population (1.5 percent). • LGBTI respondents are more often in contract work for a finite time period (35 percent) compared to the general population (28 percent), and they are five times as likely to have changed jobs during the past 12 months (26 percent) compared to the general population (5 percent). • LGBTI respondents less often use the National Employment Service (58 percent) or informal networks (67 percent) to seek employment than the general population (85 and 90 percent, respectively). 3.1 Employment status 70. LGBTI respondents are less likely to be unemployed and more often still studying compared to the general population.62 Almost every third person in the general population is unemployed (30 percent), which is three times more than is the case for LGBTI respondents (11 percent).63 Three out of 10 LGBTI respondents (30 percent) are still students compared to half that number among the general population (17 percent). This could, in part, explain the lower percentage who are unemployed compared to the general population. A similar number among LGBTI respondents and the general population are employed (49 and 46 percent, respectively), and similar percentages of both groups are self-employed. 62As with income, employment percentages and rates are likely to vary by differences in survey mode. For example, the online outreach of the LGBTI SILC may affect the makeup of the survey participants, and this may also correlate to an individual’s employment status. For example, there was a significant portion of LGBTI respondents who are employed by NGOs and are students. 63The official unemployment rate in Serbia at the time of the survey (2017) for people between the ages of 15 and 64 was 14.1 percent (according to 2017 data from the Statistical Office of the Republic of Serbia). The estimate in the report differs due to the matching process, which adjusts the demographics of the official SILC to match the demographics of the LGBTI sample. 31 Figure 3.1.1. Current Main Status (self-declared) (percent) LGBTI sample General population Employee 49 46 Self-employed, including also unpaid family worker 7 5 Unemployed (looking for a job) 11 30 Pupil/student 30 17 Pensioners and other inactive persons 3 2 0 10 20 30 40 50 Question: What is, in your opinion, your current main status? Base: Total LGBTI sample and the general population. 71. Youth unemployment among the general population might be a result of their earlier entry into the labor market. Although two-thirds of young LGBTI respondents (up to 25 years old, 30 percent) are still in education, this is true for only four out of 10 of the general population youth (17 percent). The earlier entry into the labor market of the young general population can explain the large difference in youth unemployment (9 percent for young LGBTI respondents and 28 percent for the younger general population) and likely accounts for a large part of the overall differences registered. The percent of youth employed is the same between LGBTI respondents and the general population. Figure 3.1.2. Current Main Status (self-declared), 18–25 years old (percent) LGBTI sample General population Employee 21 21 3 Self-employed, including also unpaid family worker 3 Unemployed (looking for a job) 9 28 Pupil/student 66 47 Pensioners and other inactive persons 2 2 0 10 20 30 40 50 60 70 Question: What is, in your opinion, your current main status? Base: LGBTI sample and the general population between the ages of 18 and 25. 72. Higher educational attainment increases the likelihood of finding employment. Respondents with college or higher education are far more often employed (LGBTI sample: 83 percent; general population: 58 percent) compared to those with primary or secondary school 32 education (LGBTI: 51 percent; general population: 44 percent). Since the matching of the general population to the LGBTI sample adjusted by levels of education, there are minimal differences in educational attainment and they therefore cannot explain why the percent employed is higher for LGBTI people. 73. Looking at the LGBTI sample, the percent unemployed is about one out of five for LGBTI respondents living in small cities, the outskirts or suburbs, or rural areas (19 percent). In contrast, in big cities (primarily Belgrade), the percent unemployed among the LGBTI sample is about half that (8 percent). This stark difference is not present in the general population, where in both big cities and rural areas, three out of 10 people are unemployed (29 and 30 percent, respectively). 74. Few differences exist between LGBTI subgroups with regard to their employment status, though transgender and intersex respondents are more likely to be independent contractors (20 and 17 percent, respectively) compared to other LGBTI subgroups. Overall, differences among LGBTI subgroups should be interpreted with caution because of the age differences between the groups. After adjusting for age, the percentage of bisexual women who are employed is similar to that of gay and bisexual men (see table 3.1.2), but the percentage of transgender and intersex respondents who are independent contractors is higher than among the other LGBTI subgroups. Table 3.1.1. Current Main Status Transgender Above the average Bisexual Bisexual Intersex average women Lesbian Below the average LGBTI men Gay Employee 40 48 17 48 46 27 39 Self-employed 7 1 4 6 7 7 12 Independent contractor 10 17 20 7 8 9 13 Unemployed (looking for a job) 11 19 16 9 13 7 8 Pupil/student 30 12 37 29 22 49 23 Other – inactive64 3 3 7 2 4 2 5 Question: What is, in your opinion, your current main status? Base: Total LGBTI sample. Table 3.1.2: Age-Adjusted Current Main Status Above the average Bisexual Bisexual Intersex average Transge women Lesbian LGBTI nder men Below the average Gay Employee 40 43 20 49 41 34 39 Self-employed 7 1 5 6 6 9 11 Independent contractor 10 16 23 7 7 11 13 Unemployed (looking for a job) 11 20 16 9 13 7 9 Pupil/student 30 18 27 27 30 36 25 Other – inactive 3 2 9 2 3 3 4 Question: What is, in your opinion, your current main status? Base: Total LGBTI sample. 64 An inactive person means that he/she is not looking for a job. 33 3.2 Activity, Inactivity, Employment, and Unemployment Rates 75. LGBTI respondents have a lower activity rate (66 percent) than the general population (75 percent). This is related to the fact that a larger portion of the LGBTI respondents are currently still in school and thus considered inactive. The employment rate of LGBTI respondents (56 percent) and the general population (50 percent) is similar. 76. The unemployment rate for LGBTI respondents is 14 percent, which is much lower than the unemployment rate in the general population (33 percent). Factors that affected the percent unemployed in Section 3.1 are similar to those that affect the unemployment rate: age, educational attainment, and the region of the country in which people reside. Figure 3.2.1. Activity, Inactivity, Employment, and Unemployment Rates (percent) Chart Title LGBTI sample General population 66 Activity rate 75 34 Inactivity rate 25 56 Employment rate 50 14 Unemployment rate 33 0 20 40 60 80 Base: Total LGBTI sample and the general population. 77. Intersex and gay respondents have higher activity rates than other LGBTI subgroups (83 and 73 percent, respectively), while bisexual women have lower activity rates (49 percent). Additionally, gay men have a higher than average employment rate (61 percent), while bisexual women have a lower than average one (43 percent). Many of these differences may be attributable to age differences among the LGBTI subgroups. Once age is taken into account, LGBTI subgroup differences are overall diminished, though some patterns remain and a new one emerges. Indeed, after adjusting for age, bisexual women still have a lower than average activity rate (59 percent); however, gay men have a higher than average unemployment rate (20 percent). 34 Table 3.2.1. Age-Adjusted Activity, Inactivity, Employment, and Unemployment Rates Transgender Above the average Bisexual Bisexual Intersex average women Lesbian Below the average LGBTI men Gay Activity rate 66 77 64 69 67 59 67 Inactivity rate 34 23 36 31 33 41 33 Employment rate 56 59 50 62 55 52 60 Unemployment rate 14 24 25 11 20 14 11 Base: Total LGBTI sample. 3.3 Employment sector and types of contracts 78. The public sector provides employment to over one-third of the general population (36 percent), while only every fifth LGBTI respondent (20 percent) finds employment in this sector. The private sector provides employment to most respondents (figure 3.3.1). Ninety percent of LGBTI respondents do not have trust in the state,65 which may explain why they are less likely to be employed in the public sector. This argument could be supported by the higher percentage of LGBTI respondents who hide or disguise their sexual orientation, gender identity, and/or intersex status in the public sector.66 Figure 3.3.1. Employment Sector (percent) LGBTI sample General population 58 Private - registered 58 7 Private - unregistered 2 20 Public/state-owned 36 Nongovernmental sector (civil society 10 organizations) Other (e.g., socially owned, cooperative, 6 mixed...) 4 0 20 40 60 Question: What is the form of organization at which you work, i.e., the sector in which you work? Base: Those LGBTI respondents and respondents from the general population who are currently employed. 79. The nongovernmental sector stands out as a key employer of LGBTI people. The available data do not allow for a comparison of the LGBTI sample directly with the general population, as this answer option was unique to the LGBTI SILC.67 These respondents could, however, be compared to 65 See Section 5.4. 66 See Section 4.1, figure 4.1.6. 67 This answer option was added to the LGBTI SILC based on the outcome of consultations with LGBTI organizations in Serbia and their experience in working with LGBTI communities. 35 those from the general population whose economic activities are classified under “activities of membership organizations.”68 Even though this represents a sector much broader 69 than just NGOs, only 1.5 percent of the general population are employed in “activities of membership organizations,” while 10 percent of LGBTI respondents work for an NGO. 80. LGBTI respondents assigned female at birth are employed by NGOs (15 percent) at rates three times higher than LGBTI respondents assigned male at birth (6 percent) (figure 3.2.2). Bisexual women and transgender respondents are most likely to be employed by NGOs (20 and 18 percent, respectively), while gay and bisexual men are least likely (6 and 4 percent, respectively). Figure 3.3.2. Percent of LGBTI Respondents Who Work in the NGO Sector LGBTI average 10 Female 15 Male 6 Bisexual female 20 Transgender 18 Lesbian 11 Intersex 7 Gay 6 Bisexual male 4 Female perceived as masculine 16 Male perceived as feminine 15 0 5 10 15 20 Question: The form of organization at which you work, i.e., the sector in which you work: Base: Those LGBTI respondents who defined their current main status as employed. 81. LGBTI people employed in the NGO sector earn lower incomes and work fewer hours. Those who work in the NGO sector earn the lowest median incomes (€244) in comparison to those in the private (€325) or public sectors (€390). 70 LGBTI people whose primary job is in the NGO sector work on average 29.3 hours per week, which is substantially lower than LGBTI people in general (37 hours per week) or than the general population (43.7 hours). 82. The high number of LGBTI people working in NGOs could have multiple explanations. The survey was disseminated predominantly through local NGOs, which likely led to a disproportionate representation of NGO employees in the sample. Furthermore, LGBTI people might perceive the NGO sector as less discriminatory and thus be more likely seek to work there. This conclusion is supported by the higher number of LGBTI people who feel comfortable being open about their sexual orientation, gender identity, and/or intersex status in the NGO sector compared to the public and private sectors.71 68 Eurostat, “NACE Rev. 2: Statistical Classification of Economic Activities in the European Community,” Methodologies and Worki ng Papers (Luxembourg: Eurostat, 2008), http://ec.europa.eu/eurostat/documents/3859598/5902521/KS-RA-07-015-EN.PDF. 69 It includes the activities of: business and employer membership organizations; professional membership organizations; trade unions; religious organizations; political organizations; and other membership organizations. 70 In the comparable population, there were no respondents who reported that they work for NGOs; however, the official SILC does not document this type of employment and NGOs were added to the LGBTI survey. 71 See Section 4.1. 36 Figure 3.3.3. Openness about Being LGBTI with Current Work Colleagues (percent) None A few Most All Do not apply to me LGBTI average 47 27 9 13 4 Private 47 27 8 13 5 Public/state-owned 56 34 33 4 Non-governmental sector* 9 12 21 58 0 0 20 40 60 80 100 Question: To how many of your current work colleagues are you open/out about your sexual orientation and/or gender identity and/or being an intersex person? Base: Those LGBTI respondents who are currently working. * Sample sizes are small NGO employees (n = 23). 83. Contracts for a defined term are more common among LGBTI respondents. About one out of four (35 percent) LGBTI respondents who are working for an employer have a contract for a definite period compared to 28 percent of the general population (figure 3.2.2). This difference is likely a result of the higher percentage of LGBTI people who work for NGOs, where such contracts are much more common than in other sectors. Figure 3.3.2. Type of Contract, by the Sector in which LGBTI Respondents Work (percent) For indefinite period of time For definite period of time 65 Total (all sectors) 35 27 Nongovernmental sector (civil society organizations) 73 73 Other sectors 27 Question: How do you do your job? 0 20 40 60 80 Base: Those LGBTI respondents who defined their current main status as employed and who have any type of contract (with or without an employment relationship). 3.4 Additional jobs and working hours 84. The general population works an average of 44 hours per week compared to 40.7 hours per week for LGBTI respondents. Private sector employees from the general population work 45.5 hours per week compared to 41.7 hours per week for LGBTI respondents. In contrast, LGBTI respondents employed in the public sector work similar hours per week to the general population (41 and 40.9 37 hours per week, respectively). The NGO sector, which is specific to the LGBTI sample, has the shortest working hours. Every third LGBTI respondent in Serbia has one or more additional jobs (34 percent), while only 2 percent of the general population does. The extra hours spent on these additional jobs is similar in both samples. On average, men work about 2.5–3 hours more per week on their main job than women and about 2.5 hours more if their additional jobs are included. This difference is noticeable in both the LGBTI sample and the general population. Women in the general population work 2.8 hours longer compared to LGBTI respondents whose sex assigned at birth was female. Table 3.4.1. Average Working Hours per Week Above the average LGBTI sample General population Below the average Main job All jobs Main job All jobs Total 37 40.7 43.7 44.0 Female 35.3 39.2 42.3 42.6 Male 38.3 41.8 44.7 45.0 Employee 39 42.4 42.9 43.2 Self-employed, including also unpaid family worker 32.9 37.2 50.7 51.2 Independent contractor 31.4 36.4 N/A N/A Private sector 38.7 41.7 45.1 45.5 Public (state-owned) sector 35.5 41 40.6 40.9 Nongovernmental sector 29.3 33.9 Other (e.g., socially owned, cooperative, mixed...) 36 39.7 49.9 49.9 For indefinite period of time 38.8 42.3 42.9 43.2 For definite period of time 34.5 37.8 42.9 43.2 Question: How many hours during a week do you usually work at your main job? How many hours a week on average do you spend on that additional job/those additional jobs? Base: Those LGBTI respondents and respondents from the general population who defined their current main status as employed. 85. Among employed LGBTI respondents, every fifth (20 percent) person works less than 30 hours per week counting all jobs, which is almost seven times more than among the general population (3 percent). Significantly more LGBTI respondents (22 percent) listed education and training as the reasons for working less than 30 hours,72 while 26 percent of the general population stated that they want to work more hours but could not find a job that offered them (see figure A.4.5 in Annex 4). Although a larger portion of LGBTI respondents work less than 30 hours per week, 17 percent do not want to work longer hours compared to 8 percent of the general population. 3.5 Changing jobs 86. Every fourth employed LGBTI respondent (25 percent) changed jobs in the past 12 months. In contrast, only 5 percent in the general population did (figure 3.5.1). The reasons for changing jobs are similar among both samples. The most common reason was that they found or are presently looking for a better job (LGBTI sample: 46 percent; general population: 55 percent). Employees with contracts for a finite period of time are the most likely to change jobs. For example, 42 percent of LGBTI respondents in a contract with a finite time period changed their job compared to 19 percent with an indefinite contract. The larger share of LGBTI respondents with finite contracts compared to 72 This can be explained by the fact that most LGBTI respondents were students. 38 the general population can therefore, at least in part, provide an explanation for the overall higher percentage of LGBTI respondents who changed jobs compared to the general population. Figure 3.5.1. Changed Jobs during the Past 12 Months (percent) LGBTI sample General population Changed job during the past 12 months 26 5 Found or looking for a better job 46 55 End of temporary contract 13 16 Obliged to stop by employer 16 15 Sale or closure of own/family business 1 1 Moved for family reasons 5 1 Other reasons 20 13 0 20 40 60 Question: Did you change your job during the past 12 months? The main reason why you left your previous job? Base: Those LGBTI respondents and respondents from the general population who defined their current main status as employed; those LGBTI respondents and respondents from the general population who changed their job during the past 12 months. 3.6 Long-term unemployment 87. There is a higher percentage of long-term unemployed73 in the general population than in the LGBTI sample. Of the LGBTI respondents, 32 percent can be considered to be long-term unemployed. In the general population who, as many as 71 percent can be considered to be long-term unemployed. Long-term unemployed LGBTI respondents have gone an average of three years without a job (2.9 years). Figure 3.6.1. Long-Term Unemployment among the Unemployed (percent) LGBTI sample General population 32 Long-term unemployed who at some point worked for at least 6 months 71 0 20 40 60 80 Question: How long ago did you have your last job? Base: Those LGBTI respondents and respondents from the general population who defined their current main status as unemployed, who are actively seeking employment, and who worked at some point at a job for at least 6 months. 73 Defined as unemployed for one year or longer. 39 88. Among LGBTI respondents, a significant gender gap is apparent when looking at the long- term unemployed who at some point worked for at least six months. Four times more LGBTI respondents who were assigned male sex at birth said that they have been unemployed for more than a year (44 percent) than those who were assigned female sex at birth (10 percent). It should be noted that the number of respondents is quite small in this instance and the results are thus to be taken more as a trend than as a reliable indicator. Among the general population, no such gender gap has been registered. Figure 3.6.2. Long-Term Unemployment among the Unemployed (percent) LGBTI sample General population 10 Female 71 Long-term unemployed who at some point worked for at least 6 months 44 Male 71 0 20 40 60 80 100 Question: How long ago did you have your last job? Base: Those LGBTI respondents and respondents from the general population who defined their current main status as unemployed, who are actively seeking employment, and who worked at some point at a job for at least 6 months. 3.7 Seeking employment and a minimal salary 89. Unemployed Serbians, LGBTI or not, are looking for jobs at similar rates. Among unemployed LGBTI people, intersex, transgender, and gay respondents stand out as the three groups most actively seeking employment (67, 51, and 47 percent, respectively). Since a large number of LGBTI respondents report that they are students, the LGBTI average for actively looking for a job significantly differs between those who are currently students and those who are not. Among students, 21 percent are actively seeking a job compared to 75 percent of non-students. 40 Figure 3.7.1. Actively Looking for a Job (percent) LGBTI sample General population Average 36 55 Intersex 67 0 Transgender 51 0 Gay 47 0 Bisexual male 30 0 Lesbian 30 0 Bisexual female 23 0 18-25 25 37 26-35 61 71 36-45 79 76 46 and higher 29 76 0 35 70 105 140 175 210 245 Question: Did you look for a job during the previous four weeks? Base: Respondents who defined their current main status as not employed (unemployed and students). 90. Although the National Employment Service is an important tool for the general population, significantly fewer LGBTI respondents make use of it. LGBTI respondents’ apparent avoidance of the National Employment Service might be related to their low degree of trust in institutions generally (including the political and legal systems and the police).74 For example, 43 percent of LGBTI respondents who have a low trust in the political system use the National Employment Service compared to the overall 58 percent of respondents who use the service. The general population also relies much more on informal networks (90 percent) than the LGBTI sample (67 percent). Overall, the general population uses a greater variety of ways to seek employment compared to LGBTI respondents. 74 See Section 5.4. 41 Figure 3.7.2. Ways of Seeking Employment in the Past Four Weeks (percent) LGBTI sample General population By following ads in newspapers, magazines or on the internet 77 86 Through friends, relatives, trade unions, etc. 67 90 Through ads (by placing an ad or answering the ads in 63 newspapers, magazines or on the internet) 71 Through the National Employment Service 58 85 By addressing employer directly 53 70 By waiting for the results of a job application previously sent 43 32 By waiting for a call from the National Employment Service 34 55 By taking a test, the individual had an interview or was tested 28 24 Through a private employment agency 20 29 By waiting for the results of a selection at a public institution 17 9 By looking for a piece of land, work premises or work 5 equipment 3 By applying for permit, license, financial assets for work 4 2 In some other way 6 24 0 20 40 60 80 100 Question: In which way did you look for a job during the previous four weeks? Base: Those LGBTI respondents and respondents from the general population who defined their current main status as unemployed. 42 4. LGBTI PEOPLE AND WORKPLACE DISCRIMINATION 91. Serbia’s Labor Code prohibits discrimination on a series of grounds, including sex, sexual orientation, marital status, and familial commitments. In addition, the Antidiscrimination Law prohibits discrimination on the basis of sexual orientation, affirms the right to privacy for one’s sexual orientation, and grants the freedom to express one’s sexual orientation ,75 and may be further interpreted to incorporate transgender people. 76 Research in other countries indicates that LGBTI people are often penalized when applying for jobs and discriminated against once in the workplace. Experimental studies in the United States and Greece, for example, have found that openly lesbian and gay applicants had anywhere from a 5 to 40 percent lower callback rate for interviews than heterosexual applicants and were offered lower wages.77 Thus, it is important to examine whether LGBTI respondents are frequently asked about their personal lives and are open about their sexual orientation and/or gender identity in the workplace, whether and to what extent they encounter workplace discrimination, and the effects of workplace discrimination on their lives. 92. The original SILC did not include questions on discrimination, so a comparison to levels of discrimination in the general population is not possible. The following section presents data only from the LGBTI SILC. Section highlights • LGBTI respondents, and especially transgender and intersex respondents, are frequently asked about sexual orientation (30 percent) and gender identity (34 percent, but 60 percent for transgender and 57 percent for intersex respondents) when applying for a job. • Well over half of LGBTI respondents who have been employed in the past five years hide or disguise being LGBTI at work (60 percent). • One out of two (47 percent) currently employed LGBTI respondents are not open to any of their colleagues about their sexual orientation, gender identity, and/or intersex status. • A little more than one-third (36 percent) of LGBTI respondents often or always hear negative comments or see negative conduct directed at a colleague due to the perception that the colleague is LGBTI. • Almost two-thirds (62 percent) of transgender respondents experience negative comments or conduct at work because of being LGBTI. • The workplace climate is more hostile for LGBTI respondents whose sex was assigned male at birth and who are currently perceived by others as feminine: 78 percent witness or see anti-LGBTI conduct directed toward a colleague and 69 percent have experienced a general negative attitude at work toward LGBTI people. 75 ERA, “Country Profile-Serbia” (Belgrade: Equal Rights Association for Western Balkans and Turkey, 2017), https://www.lgbti- era.org/content/serbia. 76 Ibid. 77 András Tilcsik, “Pride and Prejudice: Employment Discrimination against Openly Gay Men in the United States,” American Journal of Sociology 117, no. 2 (2011): 586–626; and Nick Drydakis, “Women’s Sexual Orientation and Labor Market Outcomes in Greece,” Feminist Economics 17, no. 1 (2011): 89–117. 43 • Sexual orientation workplace discrimination has occurred for 16 percent of gay men, 14 percent of lesbians, 10 percent of bisexual men, and 3 percent of bisexual women. • Transgender respondents are the most discriminated against in the workplace: 20 percent have suffered gender identity workplace discrimination. • Almost one in five (17 percent) intersex respondents experience workplace discrimination. • One in 10 LGBTI respondents quit a paid job because of unequal or unfair treatment, and 7 percent took unexpected leave from work for the same reason, with transgender and intersex respondents doing so more frequently (27 and 13 percent, respectively). • The consequences of workplace discrimination may be pervasive. LGBTI respondents who experience any kind of workplace discrimination have lower personal and household incomes, have higher rates of material deprivation (52 versus 25 percent), and are more likely to be currently unemployed (24 versus 11 percent) than those who have not experienced workplace discrimination. 4.1 Being openly LGBTI at work 93. It is common for employers in Serbia to ask LGBTI people about personal information related to their sexual orientation (30 percent) or gender identity (34 percent). Even more frequently, LGBTI respondents are asked questions about their family and marital status that could be used to identify and discriminate against them. In other contexts, however, such questions can be used to discriminate more broadly against other demographic categories. In the United States, for example, the Equal Employment Opportunity Commission has noted that such questions can disparately impact women: “It is clearly discriminatory to ask such questions only of women and not men (or vice-versa). Even if asked of both men and women, such questions may be seen as evidence of intent to discriminate against, for example, women with children.” 78 78U.S. EEOC, “Pre-Employment Inquiries and Marital Status or Number of Children” (Washington, DC: U.S. Equal Employment Opportunity Commission, n.d.), https://www.eeoc.gov/laws/practices/inquiries_marital_status.cfm. 44 Figure 4.1.1: Inquiries into a Person's Personal Information at Work (percent) Yes, more than once Yes, once No Marital status 19 46 36 Family status 20 43 37 Gender identity 13 21 66 Sexual orientation 15 15 70 0 20 40 60 80 100 Question: And has it happened to you personally that, while seeking a job in the past five years, you were requested the following from employers? 94. There are significant differences among LGBTI subgroups in the frequency with which they face questions about their gender identity. Transgender and intersex respondents are more often asked about their gender identity than other LGBTI subgroups (60 and 57 percent, respectively). There are few differences among LGBTI subgroups in the frequency of questions about their marital status, family status, or sexual orientation. 95. Inquiries into the gender identity of the LGBTI respondents are related to their perceived gender conformity (figure 4.1.2). LGBTI respondents who are perceived as different to their sex assigned at birth are much more often asked about their gender identity than others. Almost half (42 percent) of men perceived as feminine have been asked at least once about their gender identity. For women perceived as masculine, 39 percent have been asked. 45 Figure 4.1.2. Inquiries into a Person's Gender Identity at Work, by perceived gender conformity (percent) Yes, more than once Yes, once No Men perceived as masculine 9 18 73 Men perceived as feminine 14 33 53 Gender identity Women perceived as feminine 15 17 68 Women perceived as masculine 11 28 61 0% 20% 40% 60% 80% 100% Question: And has it happened to you personally that, while seeking a job in the past five years, you were requested the following from employers? Base: Those LGBTI respondents who looked for a job in the previous five years. 96. One out of two (47 percent) currently employed LGBTI respondents are not open to any of their colleagues about their sexual orientation, gender identity, and/or intersex status.79 One- quarter (27 percent) of LGBTI respondents who are currently employed are open to a few of their colleagues, and only 13 percent are open to all of their colleagues. Thus, more currently employed LGBTI respondents choose not to disclose their sexual orientation, gender identity, and/or intersex status than to be open to at least a few of their colleagues. 97. Intersex respondents (62 percent), bisexual men (70 percent), and gay men (47 percent) are most likely not to be open to their colleagues compared to other LGBTI subgroups (figure 4.1.3). Transgender respondents, lesbians, and bisexual women, on the other hand, are more open, with 63 percent of trans people, 62 percent of lesbians, and 56 percent of bisexual women open to a few, most, or all of their colleagues.80 79 Only those employed or self-employed were asked questions on workplace discrimination. Employees who are independent contractors as well as inactive respondents did not answer this section. 80 The sample size is small for transgender respondents (n = 10). 46 Figure 4.1.3. Openness about Being LGBTI to Current Work Colleagues (percent) None A few Most All Do not apply to me LGBTI average 47 27 9 13 4 Transgender* 21 29 17 17 17 Lesbian 34 33 10 19 3 Bisexual women 41 33 10 13 3 Gay 47 25 9 15 4 Intersex* 62 19 7 12 Bisexual men 70 16 5 9 Men perceived as masculine 59 20 7 9 5 Women perceived as feminine 35 32 14 17 3 Women perceived as masculine 32 35 11 18 4 Men perceived as feminine 30 40 14 17 0 20 40 60 80 100 Question: To how many of your current work colleagues are you open/out about your sexual orientation and/or gender identity and/or being an intersex person? Base: Those LGBTI respondents who are currently working. * Sample sizes are small for Intersex respondents (n = 28) and transgender respondents (n = 10). 98. In addition to LGBTI subgroups, whether LGBTI people are perceived as masculine or feminine is an important determinant in their degree of openness about their sexual orientation, gender identity, and/or intersex status. LGBTI respondents whose sex assigned at birth is male who believe others perceive them as masculine are far more likely not to be open to their colleagues—six out of 10 men who believe others perceive them as masculine (59 percent) said that they were not open, which differs from LGBTI respondents whose sex assigned at birth is male who believe others perceive them as feminine (30 percent). 99. LGBTI respondents who are living with their (unmarried) partner stand out as more open to all of their work colleagues (27 percent) compared to those residing alone (14 percent) or with their parents (7 percent) (see figure A.4.6 in Annex 4). One-third (31 percent) of LGBTI respondents who live with their partners are not out to any of their colleagues, while about one in two LGBTI respondents who live alone (49 percent) or with their parents (48 percent) are not out to any of their colleagues. 100. LGBTI respondents who are employed in the public sector are less likely to be out than those employed in the private or nongovernmental sector. Over half (56 percent) of LGBTI respondents employed in the public sector are not out to their colleagues compared to slightly under half (47 percent) employed in the private sector. Notably, a majority (58 percent) of LGBTI respondents 47 employed in the nongovernmental sector are out to all of their colleagues, which likely reflects employment in NGOs that focus on LGBTI issues. 4.2 Workplace climate and discrimination 101. One-seventh (15 percent) of the LGBTI respondents have often or always experienced negative comments or conduct at work because of being LGBTI (table 4.2.1). Only 26 percent of transgender respondents have never experienced negative comments or conduct at work, and they are far more likely than other LGBTI subgroups to always or often face these behaviors (33 percent). Bisexual women and intersex respondents are more likely to never experience negative comments or conduct (59 and 52 percent, respectively). About one-third (36 percent) of LGBTI respondents have often or always heard negative comments or seen negative conduct against their colleagues because they are perceived to be LGBTI, while one-quarter (26 percent) have never experienced, seen, or heard such comments or conduct. One-third of respondents (32 percent) reported often or always experiencing a general negative attitude at work toward LGBTI people, which likely explains why most LGBTI respondents in part or fully hide their sexual orientation and gender identity (SOGI) and/or intersex status at work. Table 4.2.1. Discrimination at Work Because of Being LGBTI in the Past 5 Years (% of answers often + always) Transgender Bisexual Bisexual Intersex average women Lesbian LGBTI men Experiencing negative comments or conduct at Gay 15 21 33 17 21 16 31 work because of being LGBTI Hearing or seeing negative comments or conduct against your colleagues because they 36 39 46 39 33 36 36 are perceived to be LGBTI Experiencing a general negative attitude at 32 43 40 33 30 29 32 work against people because they are LGBTI Question: How often during your employment in the past five years have you experienced negative comments or conduct at work because of your sexual orientation and/or gender identity and/or being intersex person? Base: Those LGBTI respondents who were employed at some point in the past five years. 102. Men who believe that others perceive them as feminine stand out with significantly more negative experiences in witnessing anti-LGBTI bias (78 percent) and often or always feeling an overall negative LGBTI environment at work (30 percent). Other differences between LGBTI groups are not statistically significant. 48 Table 4.2.2. Discrimination at Work Because of Being LGBTI in the Past 5 Years (% of answers often + always) Men perceived Men perceived LGBTI average as masculine perceived as perceived as as feminine masculine feminine Women Women Experiencing negative comments or conduct 15 43 29 32 43 at work because of being LGBTI Hearing or seeing negative comments or conduct against your colleagues because they 36 33 28 40 44 are perceived to be LGBTI Experiencing a general negative attitude at 32 15 30 11 21 work against people because they are LGBTI Question: How often during your employment in the past five years have you experienced negative comments or conduct at work because of your sexual orientation and/or gender identity and/or being intersex person? Base: Those LGBTI respondents who were employed at some point in the past five years. 103. The negative workplace climate that respondents perceive translates, although at a much lower rate, into experiences of workplace discrimination: 16 percent of gay and 14 percent of lesbian respondents have personally experienced discrimination in the workplace due to their sexual orientation.81 Bisexual people report much lower rates of workplace discrimination: 10 percent of bisexual men and 3 percent of bisexual women.82 These differences by sexual orientation are unrelated to the age of the respondents, and the patterns remain even after adjusting for age. Men who believe others perceive them as feminine stand out again as particularly vulnerable: one-third experienced discrimination at work because of their sexual orientation (33 percent). 83 104. One-fifth (20 percent) of transgender respondents personally experienced workplace discrimination based on their gender identity and one-fifth (17 percent) of intersex respondents based on their sex characteristics. The difference between the perception of bias in the workplace compared to the personal experience of discrimination is possibly related to a psychological phenomenon commonly known as the “denial of disadvantage.”84 This refers to a situation in which people from marginalized backgrounds perceive their workplaces to be hostile to people with their characteristics, but they also see themselves as capable of navigating their job environments to avoid personally experiencing bias or discrimination. This phenomenon has been extensively documented, for example, in the gender literature on discrimination. 85 Alternatively, workplaces may be viewed as a hostile environment to LGBTI respondents, but existing antidiscrimination laws, when enforced, may reduce actual experiences of workplace discrimination. 81 Base: Respondents who are currently employed or who looked for employment in the past five years. 82 It should be said that this might have something to do with the demographics of bisexual women in the sample —they are predominantly younger, which could ac count for the higher percentage of “don’t know” answers. 83 The small number of transgender and intersex respondents does not allow for a further disaggregation of the data for these populations. 84 Faye Crosby, “The Denial of Personal Discrimination,” American Behavioral Scientist 27, no. 3 (1984): 371–86. 85 Vicki S. Helgeson, Psychology of Gender, 5th ed. (New York: Routledge, 2017). 49 Figure 4.2.1. Personal Experiences of Discrimination at Work Because of Being LGBTI (percent) Gay 16 Lesbian 14 Discrimination based on sexual orientation Bisexual men 10 Bisexual women 3 Discrimination based Transgender 20 on gender identity Discrimination based Intersex 17 on sex characteristics 0 5 10 15 20 Question: In the past five years, have you personally experienced being brought into an unequal situation at your job based on your sexual orientation? In which of the following situations has this happened? In the past five years, have you personally experienced being brought into an unequal position at your job because of your gender identity? In which of the following situations has it happened? In the past five years, have you personally experienced being brought into an unequal position at your job because of your sex characteristics (being an intersex person)? In which of the following situations has it happened? Base: Those LGBTI respondents who were employed or looked for work at some point in the past five years. 105. Among LGBTI respondents who have been discriminated against at work, most have experienced bias in relation to professional advancement or promotions, though specific experiences vary across groups.86 About half of LGBTI respondents (45 percent) report discrimination in professional development or promotions, common specifically because of discrimination based on sexual orientation (49 percent) and on a person’s sex characteristics (51 percent) (figure 4.2.2). Discrimination in hiring is more prevalent based on a person’s sex characteristics (60 percent) or gender identity (39 percent) than sexual orientation (32 percent). In addition, workplace discrimination during the process of being fired is more prevalent based on a person’s sex characteristics (42 percent), and discrimination in other job-related situations is more prevalent based on a person’s gender identity (41 percent). 86The low number of cases prevents us from performing an analysis by LGBTI groups. Therefore, we have only presented an analysis of the level of discrimination based on sexual orientation, gender identity, or sex characteristics. However, it should be kept in m ind that the analysis of discrimination based on gender identity and sex characteristics is also based on a very small number of cases. 50 Figure 4.2.2. Specific Situations in which LGBTI Respondents Experienced Discrimination at Work Because of Being LGBTI (percent) When you were being hired 32 32 39 60 When you were being fired 28 29 27 42 Professional advancement/promotion 45 49 22 51 When the amount of salary was determined 25 25 24 20 Some other job-related situation 27 25 41 12 All LGBTI Sexual Gender Sex persons orientation Indentity characteristics Question: In the past five years, have you personally experienced being brought into an unequal position at your job based on your sexual orientation? In which of the following situations has it happened? In the past five years, have you personally experienced being brought into an unequal position at your job based on your gender identity? In which of the following situations has it happened? In the past five years, have you personally experienced being brought into an unequal position at your job based on your sex characteristics (being an intersex person)? In which of the following situations has it happened? Base: Those LGBTI respondents who named a specific case of discrimination at work because of being LGBTI. 4.2.1 Consequences of workplace discrimination 106. Discrimination has negative effects on victims, including on their health and ability to fulfill their full potential in education or employment. Since measuring these often-long-term negative effects is challenging, respondents were therefore asked about the immediate consequences of the workplace discrimination they experienced. 107. One in 10 LGBTI respondents (10 percent) has quit a paid job, and 7 percent have taken unexpected leave from work due to the discrimination they experienced. There are few differences among lesbians, gay men, and bisexual men. Bisexual women, however, are unique in that only 1 percent reported taking unexpected leave from work due to experiencing unequal or unfair treatment. The data do not allow for any further conclusion that could help better understand these differences. 108. Intersex and transgender respondents far more often quit a paid job due to discrimination. One-fourth of transgender respondents (27 percent) and 14 percent of intersex respondents have quit their jobs due to unequal or unfair treatment. Transgender and intersex respondents also take unplanned leaves of absence from their jobs at slightly higher rates than other LGBTI subgroups (13 and 11 percent, respectively). Due to the small number of respondents, these differences are not statistically significant, yet they should not be disregarded without further inquiry, as other research suggests that transgender people especially often face the most discrimination of all LGBTI subgroups.87 87World Bank, “Economic inclusion of LGBTI Groups in Thailand,” vol. 2 (Washington, DC: World Bank, 2018), http://documents.worldbank.org/curated/en/269041521819512465/pdf/124554-WP-PUBLIC-LGBTI-Report2018-full-report-English- 23March.pdf. 51 Figure 4.2.1.1. Quitting or Taking Unplanned Leave from a Paid Job in the Previous Five Years Due to Experiencing Unequal or Unfair Treatment Because of Being LGBTI (percent) Quit a paid job Taken unplanned leave from a paid job 30 27 25 20 13 14 15 11 11 10 9 9 10 10 7 8 8 6 5 1 0 LGBTI Trans Intersex Bisexual Lesbian Gay Bisexual average women men Question: Have you quit (left) a paid job (of your own accord) in the previous five years due to experiencing unequal treatment (or unfair treatment) because of your sexual orientation and/or gender identity and/or being an intersex person? Have you taken unplanned leave (been absent) from your paid job (in cash, kind, or services) in the previous five years due to experiencing unequal treatment (or unfair treatment) because of your sexual orientation and/or gender identity and/or being an intersex person? 109. The vulnerability noted above of LGBTI respondents who are assigned male at birth who believe others perceive them as feminine is confirmed —they have far more often quit a paid job (14 percent) or taken unplanned leave (19 percent) compared to other LGBTI subgroups. LGBTI respondents whose sex assigned at birth was female who are perceived by others as masculine also quit a job (15 percent) or took unplanned leave (7 percent) at rates greater than LGBTI respondents whose sex assigned at birth was female who are perceived by others as feminine (10 and 2 percent, respectively). There are differences by sex assigned at birth, as LGBTI respondents whose sex assigned at birth was male more frequently take unexpected leave due to discrimination (9 percent) compared to respondents assigned female at birth (4 percent). There are few differences in sex assigned at birth in the rate of quitting a paid job due to unequal or unfair treatment. Figure 4.2.1.2. Quitting or Taking Unplanned Leave from a Paid Job in the Previous Five Years Due to Experiencing Unequal or Unfair Treatment Because of Being LGBTI (percent) Quit a paid job Taken unplanned leave from a paid job Men 9 9 Women 10 4 Men perceived as masculine 8 6 Men perceived as feminine 14 19 Women perceived as feminine 10 2 Women perceived as masculine 15 7 Question: Have you quit (left) a paid job (of your own accord) in the previous five years due to experiencing unequal 0 35 treatment (or unfair treatment) because of your sexual orientation and/or gender identity and/or being intersex person? Have you taken unplanned leave (been absent) from your paid job (in cash, kind, or services) in the previous five years due to experiencing unequal treatment (or unfair treatment) because of your sexual orientation and/or gender identity and/or being an intersex person? 52 110. LGBTI respondents who experience any type of workplace discrimination tend to have lower median personal, household, and equivalized incomes than LGBTI respondents who do not experience workplace discrimination. The median personal income of LGBTI respondents who experienced workplace discrimination is €269, which is less than other LGBTI respondents (€300). The at-risk-of-poverty rate is slightly higher for those who experienced discrimination (20 percent) than those who did not (16 percent) (figure 4.2.1.3). Figure 4.2.1.3. Personal, Houeshold, and Equivalized Household Income, by any Experiences of Workplace Discrimniation (median) Experienced Workplace Discrimination No Workplace Discrimination 269 Personal income 300 407 Household income 493 242 Equivalized household income 285 0 100 200 300 400 500 600 Base: LGBTI respondents who were employed and who looked for employment in the past five years. 111. LGBTI respondents who experience any kind of workplace discrimination are less likely to be currently employed (58 percent) and more likely to be currently unemployed (24 percent) compared to those who did not experience workplace discrimination (65 percent employed and 11 percent unemployed, respectively). Additionally, LGBTI respondents who experienced workplace discrimination are less likely to currently be students (5 percent) than those who did not experience discrimination (14 percent). Figure 4.2.1.4. Current Main Status (Self-Declared) (percent) Experienced workplace discrimination No workplace discrimination Employee 58 65 Self-employed, including also unpaid family 9 worker 9 Unemployed (looking for a job) 24 11 Pupil/student 5 14 Pensioners and other inactive persons 4 2 0 10 20 30 40 50 60 70 Question: What is, in your opinion, your current main status? Base: LGBTI respondents who were employed and who looked for employment in the past five years. 53 5. THE QUALITY OF LIFE OF LGBTI PEPOLE 112. Economic outcomes, such as employment status, income, and material deprivation, are of crucial importance to the well-being of individuals. However, it is also important to consider a person’s quality of life in general. The social and economic exclusion of marginalized populations can manifest in diminished physical and mental health, limited trust in others and institutions, and lower overall life satisfaction. Additionally, how LGBTI people cope in discriminatory environments—such as the existence or lack of support networks—points to patterns of resilience. This section covers a number of key aspects that affect quality of life, including health, support networks, and levels of trust. 113. The original SILC contained questions only on health. The LGBTI SILC, on the other hand, included questions on trust, support networks, and life satisfaction to generate a more complete picture of the lived reality of LGBTI people in Serbia. Therefore, comparisons to the general population can be drawn only for the questions on health (section 5.1). Section highlights • The general population more frequently rate their health as very good compared to LGBTI respondents (55 percent compared to 32 percent), and only 18 percent of transgender respondents do. • The self-assessment of one’s health as bad or very bad progressively increases among LGBTI people who are materially deprived (11 percent), severely materially deprived (14 percent), and extremely materially deprived (20 percent) compared to 5 percent of the overall LGBTI sample. Perceiving one’s health as bad or very bad does not increase in the general population (2 percent for the materially deprived, 3 percent for the severely materially deprived, and 3 percent for the extremely materially deprived). • LGBTI respondents who are at risk of poverty tend to rate their health as good or very good at lower rates than all LGBTI respondents (59 percent compared to 71 percent), while there is no difference among the general population (90 percent for those at risk of poverty compared to 93 percent overall). • Almost half of LGBTI respondents (45 percent) did not visit a doctor when they needed to in the past 12 months compared to only 9 percent in the general population. • Transgender respondents tend to rate their life satisfaction as low at higher rates than overall LGBTI respondents (77 percent compared to 52 percent). • Significant percentages of LGBTI respondents experience negative emotions all or most of the time, as more than one-fourth (28 percent) felt very nervous, 22 percent felt depressed, and 14 percent were so despondent that nothing could put them in a good emotional state. • One-third (32 percent) of transgender respondents feel very nervous all or most of the time, 42 percent feel depressed all or most of the time, and 31 percent feel so 54 despondent all or most of the time that nothing could put them in a good state of mind. • Well over 90 percent of LGBTI respondents do not trust the political system, the legal system, or the police. • One-fifth of respondents believe that being LGBTI negatively impacts their economic well-being (21 percent). 5.1 Self-rated health 114. The general population assess their health more positively than LGBTI respondents. A majority of LGBTI respondents describe their state of health as good or very good (71 percent), but a greater percentage of the general population assess their health in this way (93 percent). At the same time, 5 percent of LGBTI respondents assess their health as bad compared to 1 percent among the general population (figure 5.1.1). Metrics such as self-rated overall health have been validated as indicators of health distress and the presence of disease and can identify groups that are at a heightened risk of mortality. 88 Disparities in health assessments such as the above can have multiple sources, for example, they may be structural and linked to exclusionary policies and/or the lack of societal acceptance of stigmatized populations.89 Figure 5.1.1. Self-Assessment of the State of Health (percent) LGBTI sample General population 32 Very good 55 39 Good 38 24 Solid 6 5 Bad 1 0 Very bad 0 0 10 20 30 40 50 60 Question: What is the state of your health? Base: Total LGBTI sample and the general population. 115. Bisexual female respondents assess their health more negatively than the average for all LGBTI subgroups. Although 71 percent of LGBTI respondents assess their health as very good or good, only 60 percent of bisexual female respondents do (figure 5.1.2). For lesbians and transgender respondents, the number is 68 percent. Bisexual men are most likely to describe their health as good 88 Ellen L. Idler and Yael Benyamini, “Self-Rated Health and Mortality: A Review of Twenty-Seven Community Studies,” Journal of Health and Social Behavior 38, no. 1 (1997): 21–37. 89 For example, Mark L. Hatzenbuehler, Andrew R. Flores, and Gary J. Gates , “Social Attitudes Regarding Same -Sex Marriage and LGBTI Health Disparities: Results from a National Probability Sample,” Journal of Social Issues 73, no. 3 (2017): 508–28. 55 or very good among the LGBTI subgroups (79 percent), followed by gay men (77 percent). A significant percent of bisexual women and intersex respondents self-assess their health as bad or very bad (10 and 7 percent, respectively). Figure 5.1.2. Self-Assessment of the State of Health, by LGBTI Subgroup (percent) Very good/good Solid Very bad/bad LGBTI average 71 24 5 Bisexual men 79 16 5 Gay 77 20 3 Transgender 68 28 4 Lesbian 68 26 6 Intersex 61 30 7 Bisexual women 60 30 10 0 20 40 60 80 100 Question: What is the state of your health? Base: Total LGBTI sample. 116. LGBTI respondents who experienced any type of workplace discrimination are less likely to assess their health as good than LGBTI respondents who did not. Although 71 percent, as noted above, of LGBTI respondents self-assess their health as very good or good, only 65 percent of LGBTI respondents who experienced workplace discrimination do so, and 9 percent of those assess their health as bad or very bad, which is almost double the overall LGBTI sample (5 percent) (figure 5.1.3). LGBTI respondents who did not experience workplace discrimination self-assess their health at rates similar to the overall LGBTI sample. 117. LGBTI people who are materially deprived, severely materially deprived, and extremely materially deprived self-assess their health as progressively poorer. LGBTI respondents who are materially deprived self-assess their health as very good or good 10 percentage points less than the overall LGBTI sample (51–60 percent versus 71 percent overall) (figure 5.1.3). While 5 percent of the overall LGBTI sample assess their health as bad or very bad, 11 percent of materially deprived, 14 percent of severely materially deprived, and 20 percent of extremely materially deprived LGBTI respondents do. In the general population, assessments of one’s health as bad or very bad do not increase among those who are materially deprived (2 percent for the materially deprived, 3 percent for the severely materially deprived, and 3 percent for the extremely materially deprived). Almost half (49 percent) of extremely materially deprived LGBTI respondents do not rate their health as very good or good. 118. LGBTI respondents who are at risk of poverty are less likely to self-assess their health as good than LGBTI people overall. Again, although 71 percent of LGBTI respondents rate their health as very good or good, only 59 percent of LGBTI respondents who are at risk of poverty do so, and 9 percent self-assess their health as bad or very bad (figure 5.1.3). Being at risk of poverty is not a factor among members of the general population, who do not differ in their assessments of whether their health is good or very good (90 percent for those at risk of poverty compared to 93 percent overall). 56 Figure 5.1.3. Self-Assessment of the State of Health (percent) Very good/good Solid Very bad/bad LGBTI average 71 24 5 Did not experience workplace discrimination 69 25 5 Experienced workplace discrimination 65 26 9 Materially deprived 60 29 12 Severely materially deprived 58 28 15 Extremely materially deprived 51 29 21 At-risk-of-poverty 59 32 9 Employed 73 21 5 Self-employed 62 28 9 Independent Contractor 62 31 7 Unemployed 66 30 4 Pupil/Student 74 22 4 Other/Inactive 62 20 18 0 20 40 60 80 100 Question: What is the state of your health? Base: Total LGBTI sample. For workplace discrimination, LGBTI respondents who were employed or looked for employment in the past five years. 119. LGBTI respondents who are unemployed, inactive, self-employed, or independent contractors more often assess their health as poor than do overall LGBTI respondents . LGBTI respondents who are traditionally employed or are currently students assess their health in ways similar to the overall LGBTI sample. Compared to the 71 percent of LGBTI respondents who rate their health as very good or good, 66 percent of unemployed LGBTI respondents and 62 percent of those who are inactive, self-employed, and independent contractors do so (figure 5.1.3). A non-traditional employment setting or being out of the workforce tends to be associated with poorer health among LGBTI people. 120. Almost half of the LGBTI respondents (45 percent) did not visit a doctor when they needed to in the past 12 months compared to only 9 percent in the general population. The main reasons for not visiting a doctor when it was needed are similar for both samples, though there are some important differences. A higher percentage of LGBTI respondents did not visit a doctor because they wanted to wait and see if the situation improved (35 percent versus 21 percent in the general population) (figure 5.1.4). LGBTI respondents were also more likely to indicate that they were afraid of doctors, the hospital, testing, or treatment (12 percent versus 5 percent in the general population). On the other hand, similar percentages of the general population (17 percent) and LGBT respondents 57 (12 percent) did not visit a doctor because they could not afford the medical service. The general population cites other reasons (22 percent) or the existence of waiting lists (14 percent) for not visiting a doctor more frequently than do the LGBTI respondents (11 and 9 percent, respectively). Figure 5.1.4. Main Reasons for not Visiting a Doctor When it was Needed (percent) LGBTI sample General population Wanted to wait and see if the situation was 35 going to get better 21 Could not find the time because of work, care 19 of children or others 20 12 Fear of doctors/hospital/testing/treatment 5 12 Could not afford it (too expensive) 17 9 There is a waiting list 14 3 Did not know of any good doctor or specialist 2 11 For other reasons 22 0 10 20 30 40 Question: What is the main reason you did not visit the doctor? Base: LGBTI respondents and those from among the general population who did not visit a doctor but should have. 5.2 Life satisfaction and emotional health90 121. The LGBTI SILC included questions on life satisfaction, which was measured on an 11-point scale ranging from 0 (“not satisfied at all”) to 10 (“fully satisfied”). Respondents assessed various aspects of life quality, from their current way of life to their financial situation, current job, and relations with other people, among others. For ease of understanding and interpretation and to facilitate analyses that identify drivers of low and high satisfaction, answer categories were grouped into low, medium, and high.91 122. LGBTI respondents are the least satisfied with the ir household’s financial situation. The results on economic well-being previously documented are further reflected in the life satisfaction of the LGBTI respondents: more than a half (54 percent) express low satisfaction with their household financial situation (figure 5.2.1). 90The remaining sections of this chapter provide information only on the LGBTI respondents, as these questions were added to the SILC for the purposes of this study and no comparison can therefore be drawn to the general population. 91The classification follows Eurostat with answers 0-5 as rated as ‘low,’ 6-8 as ‘medium,’ and 9 and 10 as ‘high.’ Eurostat, “Quality of Life: Facts and Views,” Statistics Explained (Luxembourg: Eurostat, 2015). 58 Figure 5.2.1. Level of Satisfaction with Various Aspects of Life (percent) Low Medium High Don't know Financial situation of your household 54 34 12 Your current way of life 44 40 16 Your current job 41 24 17 18 Recreation centers and green areas in the place where you live 50 29 18 3 Level of development of the area in which you live 48 24 26 2 Time you have at your disposal to do the things you like 45 29 25 2 Time spent on the way to work/school and back 37 21 35 7 Your housing 34 37 29 Your relation with relatives, friends, colleagues from work, etc. 34 34 31 1 The things you do in your life are worth the effort 20 31 43 6 0 20 40 60 80 100 Question: How do you assess? Question: All in all, to what degree do you think that the things you do in your life are worth the effort? Base: Total LGBTI sample. 123. A summary measure of life satisfaction92 shows that transgender respondents are the least satisfied among the LGBTI subgroups . Overall, 51 percent of LGBTI respondents report low life satisfaction, while 77 percent of transgender respondents report low life satisfaction (figure 5.2.2). About half of gay men (54 percent), bisexual men and women (50 percent), and intersex respondents (49 percent) have low life satisfaction. The only LGBTI subgroup in which the majority does not report low life satisfaction is lesbians: 52 percent indicate a medium level of life satisfaction. Very few LGBTI respondents have a high rate of life satisfaction. 124. LGBTI respondents whose sex was assigned male at birth and are perceived as feminine are significantly less satisfied with the various aspects of life . Two-thirds (66 percent) of respondents assigned male at birth and perceived as feminine have low life satisfaction, as do 52 percent of respondents assigned female at birth and perceived as masculine. Perceived gender non-conformity can therefore be associated with lower levels of life satisfaction compared to those who are perceived as gender conforming. 92This summary measure was created by: (1) imputing “don’t know” responses via multiple imputation of data sets incorporating household income, demographics, and LGBTI identification; (2) estimating a factor analysis and extracting factor scores; (3) assigning the missingness of the factor score as the average of the 10 data sets; (4) scaling the factor to range from 0 to 10; and (5) making cuts following the 20:60:20 method. 59 Figure 5.2.2. Level of Life Satisfaction, by LGBTI Subgroup (percent) Low Medium High LGBTI average 52 43 5 Transgender 77 18 5 Gay 54 41 5 Bisexual men 50 44 5 Bisexual women 50 46 4 Intersex 49 43 8 Lesbian 43 52 4 Women perceived as feminine 47 48 5 Men perceived as masculine 51 44 5 Women perceived as masculine 52 44 4 Men perceived as feminine 66 28 6 0 20 40 60 80 100 Question: How do you assess? Question: All in all, to what degree do you think that the things you do in your life are worth the effort? Base: Total LGBTI sample. 125. LGBTI respondents who experienced workplace discrimination also tend to be less satisfied with life. About three-quarters (71 percent) of LGBTI respondents who experienced workplace discrimination have a low life satisfaction rate compared to 48 percent of LGBTI respondents who have not experienced it (figure 5.2.3). Unemployed and inactive LGBTI respondents have the lowest levels of life satisfaction—84 percent of those who are unemployed rate their life satisfaction as low (the highest percentage of low satisfaction) as do 61 percent of those who are inactive. 126. Material deprivation and poverty risk are related to how LGBTI respondents rated their satisfaction with life. About three-quarters (79 percent) of materially deprived, 87 percent of severely materially deprived, and 95 percent of extremely materially deprived LGBTI respondents have a low life satisfaction rate (figure 5.2.3). LGBTI respondents who are at risk of poverty also tend to rate their life satisfaction as low (62 percent). 60 Figure 5.2.3. Level of Life Satisfaction (percent) Low Medium High LGBTI average 52 43 5 Did not experience workplace discrimination 48 46 6 Experienced workplace discrimination 71 24 5 Materially deprived 79 21 0 Severely materially deprived 87 Extremely materially deprived 95 50 At-risk-of-poverty 62 36 2 Employed 43 51 6 Self-employed 44 49 6 Independent Contractor 57 38 5 Unemployed 84 14 2 Pupil/Student 50 46 4 Other/Inactive 61 29 11 0 20 40 60 80 100 Question: How do you assess? Question: All in all, to what degree do you think that the things you do in your life are worth the effort? Base: Total LGBTI sample. For workplace discrimination, LGBTI respondents who were employed or looked for employment in the past five years. 127. The emotional health of LGBTI respondents is mixed. Almost half (44 percent) reported that they were calm and collected, while more than one-third (37 percent) felt happy most or all of the time in the previous four weeks (figure 5.2.4). However, 6 percent of respondents did not feel happy at all. On the other hand, more than one-fourth (28 percent) felt very nervous most or all of the time in the previous month, 22 percent felt depressed all or most of the time, and 14 percent were so despondent that nothing could put them in a better emotional state. 61 Figure 5.2.4. Feelings of LGBTI Respondents in the Past Four Weeks, by LGBTI Subgroup (percent) All the time/Most of the time Some time/Little time Did not feel like that Do not know LGBTI average 44 51 4 Calm and collected Intersex 54 40 3 Transgender 36 51 8 Bisexual female 36 58 4 Bisexual male 44 48 3 Gay 48 48 3 Lesbian 43 52 4 LGBTI average 37 55 6 Intersex 39 54 6 Transgender 31 42 22 Happy Bisexual female 36 61 2 Bisexual male 40 48 7 Gay 36 56 6 Lesbian 42 54 4 LGBTI average 28 63 8 Intersex 29 58 13 Very nervous Transgender 35 55 8 Bisexual female 39 55 6 Bisexual male 23 68 9 Gay 22 68 9 Lesbian 26 66 7 LGBTI average 22 52 25 Intersex 19 44 37 In low spirits and Transgender 42 41 17 depressed Bisexual female 32 44 24 Bisexual male 17 57 22 Gay 19 55 25 Lesbian 15 57 27 LGBTI average 14 54 31 a good nothing So despondent that Intersex 12 53 35 put you in a good Transgender 31 46 23 could in much down mood mood you Bisexual female 19 57 23 Sonothing Bisexual male 14 54 30 put Gay 12 52 35 could Lesbian 10 55 35 0% 20% 40% 60% 80% 100% Question: In the past four weeks, how much time were you... Base: Total LGBTI sample. 128. Transgender respondents experience negative feelings most frequently among LGBTI subgroups, followed by bisexual women. One-third (32 percent) of transgender respondents feel very nervous all or most of the time, 42 percent feel depressed all or most of the time, and 31 percent feel 62 so despondent all or most of the time that nothing could improve their emotional state. Similar trends can be observed for unemployed and inactive LGBTI respondents, which likely points to the relationship between economic outcomes and a person’s overall well-being. 5.3 Support networks 129. Support networks are key to the overall well-being of individuals because they offer help and protection from hardship. Due to social exclusion and stigma, LGBTI people can lack certain critical support networks, which may increase the physical and mental health disparities between LGBTI people and the general population.93 The survey shows that the majority of LGBTI respondents (83 percent) have such networks of relatives, friends, or neighbors they can ask for help if needed (figure 5.3.1).94 Figure 5.3.1. Presence of Support Networks, by LGBTI Subgroup (percent) Yes No Do not know LGBTI average 83 10 7 Bisexual men 93 34 Lesbian 90 7 3 Gay 78 13 9 Transgender 76 13 11 Bisexual women 74 14 12 Intersex 71 21 8 Women perceived as feminine 92 6 2 Women perceived as masculine 84 9 7 Men perceived as masculine 78 13 9 Men perceived as feminine 77 15 8 0 20 40 60 80 100 Question: Do you have relatives, friends, or neighbors you can ask for help if needed (moral, finanical, material support or some other help)? Base: Total LGBTI sample. 130. Intersex respondents stand out, as they more often do not have support networks they can rely on when they need help—21 percent are in this situation in comparison to 10 percent of all LGBTI respondents (figure 5.3.1). This is also the case with unemployed and inactive persons. Bisexual women, transgender respondents, and gay men similarly are less likely to have reliable social networks compared to lesbians and bisexual men. There is also a notable difference by sex assigned at birth. LGBTI respondents whose assigned sex at birth was male often have fewer support networks compared to those whose assigned sex at birth was female. 93 For example, Shannon D. Snapp and others, “Social Support Networks for LGBT Young Adults: Low Cost Strategies for Positive Adjustment,” Family Relations 64, no. 3 (2015): 420–30. 94 Examples of cases of needing help include: moral support, financial assistance, or material assistance. 63 131. The majority of LGBTI respondents seek help from their friends and immediate family members. Only a few seek support from others, such as less immediate family members, work colleagues, and neighbors (figure 5.3.2). Figure 5.3.2. People LGBTI Respondents Would Ask for Help if Needed (percent) Friends 79 Parents 63 Siblings 46 Other family members 21 Work colleagues 13 Neighbors 6 Other 6 Wouldn`t ask for help 2 0 10 20 30 40 50 60 70 80 Question: And who would you ask for help if needed (moral, financial, material support or some other help)? Base: LGBTI respondents who say that they have relatives, friends, or neighbors they can ask for help if needed. 5.4 Levels of trust and feelings of safety 132. Trust95 in the police, as well as the political and legal systems, is low among LGBTI respondents, and especially low with regard to the political system (figure 5.4.1). Social trust is tied to civic and political engagement; those who do not trust their social and political institutions may not participate in them or see them as adequate avenues to change their own social situation.96 LGBTI populations may be further impacted by a lack of trust in the police, which might be related to the fear that their SOGI identity will be disclosed to others without their consent, thus creating a strong deterrence to reporting incidents. 97 Overall, 95 percent of LGBTI respondents have low trust in the political system, 93 percent feel the same about the legal system, and 91 percent share this view of the police. LGBTI people also have little social trust, as 69 percent have low trust in other people generally. 95 Trust was measured on an 11-point scale from 0 (“do not trust people at all”) to 10 (“most of the people are trustworthy”) and then grouped according to the same categories of low, medium, and high: low from 0 to 5; medium 6 to 8; and high 9 to 10. 96 See M. Kent Jennings and Laura Stoker, “Social Trust and Civic Engagement across Time and Generations,” Acta Politica 39, no. 4 (2004): 342–79; and Eric M. Uslaner and Mitchell Brown, “Inequality, Trust, and Civic Engagement,” American Politics Research 33, no. 6 (2005): 868–94. 97 Kristen Kuehnle and Anne Sullivan, “Gay and Lesbian Victimization: Reporting Factors in Domestic Violence and Bias Incidents,” Criminal Justice and Behavior 30, no. 1 (2003): 85–96. 64 Figure 5.4.1: Trust in the Political System, Legal System, and Police (percent) Low trust Medium trust High trust Don't know Political system 95 221 Legal system 93 4 21 Police 91 7 21 Trust in people 69 23 7 1 0 20 40 60 80 100 Question: How much trust do you have in: Base: Total LGBTI sample. 133. There are few subgroup differences among LGBTI people in their levels of social and political trust. No subgroup has a medium or high level of trust in any of the measured institutions. Social and political trust also does not vary materially by sex assigned at birth or perceived gender conformity. 134. Even though LGBTI respondents have little trust in state institutions, a majority feel very or mostly safe in the area where they live (81 percent) (figure 5.4.2). There are no statistically significant differences among LGBTI subgroups in their sense of safety, though males who are perceived as feminine more often feel unsafe. Other differences among subgroups include the fact that LGBTI respondents who experienced workplace discrimination or who are materially deprived, unemployed, or self-employed feel less safe compared to other LGBTI respondents. Figure 5.4.2. Feeling Safe in the Area Where LGBTI Respondents Live (percent) Safe Unsafe Don't know LGBTI average 81 18 1 Did not experience workplace discrimination 83 16 1 Experienced workplace discrimination 68 30 2 Materially deprived 73 25 2 Severely materially deprived 64 33 3 Extremely materially deprived 55 39 6 At-risk-of-poverty 77 21 2 Employed 85 14 1 Self-employed 75 25 0 Independent Contractor 80 19 1 Unemployed 76 24 0 Pupil/Student 81 18 1 Other/Inactive 79 15 6 0 20 40 60 80 100 Question: How safe do you feel in the area where you live? Base: Total LGBTI sample. For workplace discrimination, LGBTI respondents who were employed or looked for 65 5.5 Perceived impacts of being LGBTI on economic well-being 135. One-fifth (21 percent) of respondents believe that being LGBTI negatively affects their current economic position (figure 5.5.1). Transgender respondents and respondents whose sex assigned at birth was male and who are perceived as feminine are more likely to report negative implications on their economic well-being (45 and 42 percent, respectively). Bisexual men and women are less likely to see negative consequences in their LGBTI status (10 and 12 percent, respectively). Older LGBTI respondents and those outside of big cities are slightly more likely to feel that being LGBTI negatively affects their economic position than younger respondents and those who reside in big cities. Among those aged between 36 and 45, 28 percent report a negative impact, while 19 percent of those under 25 do. Interestingly, those over the age of 45 are most likely to see their LGBTI status as having a positive impact (15 percent). One in four LGBTI respondents (24 percent) who reside outside of big cities believe that being LGBTI has a negative impact on their economic life compared to one in five (19 percent) of those who reside in urban areas. Figure 5.5.1. LGBTI Identification and Economic Position, by LGBTI Subgroup (percent) It affects positively No, it does not affect It affects negatively Don't know LGBTI average 5 62 21 13 Intersex 12 48 26 14 Transgender 4 42 45 9 Men perceived as feminine 6 43 42 9 Aged 46 and higher 15 51 22 12 Unemployed 5 49 37 9 Other/Inactive 10 42 30 18 0 20 40 60 80 100 Question: All things considered, does the fact that you belong to the LGBTI population affect your current economic position and in which way? Base: Total LGBTI sample. 136. Unemployed and inactive LGBTI respondents more frequently believe that being LGBTI has negatively affected their economic position. About one-third of unemployed (37 percent) and inactive (30 percent) LGBTI respondents have this view compared to one-fifth (20 percent) of those who are employed and 15 percent of those who are currently students. 66 Intersex People Face Particular Challenges Findings from the survey point to the fact that there are marked differences in the lives and experiences of people within the LGBTI population and that each subgroup faces unique challenges and difficulties. Policies and legal frameworks are often not disaggregated and do not take into account the diverse lived realities and varied experiences of each LGBTI subgroup. In Serbia, for example, intersex people are completely excluded from the overall legal and policy framework. Although transgender people generally experience the most difficulties, given the relative invisibility of intersex people and the fact that they are often missed in research, special mention should be made of the experiences and challenges faced by intersex people as determined by the survey. • Intersex respondents, together with transgender respondents, are at the highest risk of material deprivation (33 and 40 percent respectively). • Although workplace discrimination forced one in 10 LGBTI people to quit their jobs, the rate for intersex respondents is higher at 13 percent (and even higher for transgender respondents at 27 percent). The consequences of this workplace discrimination are pervasive and can include lower incomes and higher rates of material deprivation and unemployment. • Intersex and transgender respondents are most frequently asked about sexual orientation and gender identity (57 and 60 percent, respectively) when applying for a job. • When it comes to the existence of support networks, intersex respondents stand out, as they more often do not have networks they can rely on when they need help—21 percent are in this situation in comparison to 10 percent of all LGBTI respondents. 67 6. Conclusion and Recommendations 6.1. Conclusion 137. This report shines a light on the lives of LGBTI people in Serbia and provides empirical evidence on areas where their development outcomes can be improved. The data collected through the survey contribute to the growing worldwide quantitative research base on LGBTI people. The survey also represents an important contribution to the suite of methodologies available to understand the lives of LGBTI people. Globally, there is limited quantitative data on the lived experiences of LGBTI people, even in Organisation for Economic Co-operation and Development (OECD) countries. The methods used to conduct this survey demonstrate that it is possible to make robust comparisons between the lives of LGBTI people and the population at large, even in circumstances where it can be difficult to reach LGBTI people and the characteristics of the LGBTI population at large are unknown. 138. Among other concerns, the survey findings point to serious workplace difficulties for LGBTI people. Because LGBTI people are afraid of a potentially hostile work environment, they rarely disclose their identity to their colleagues and supervisors. Those employed in the public sector are far less likely to be out to their colleagues compared to those in the nongovernmental sector. Specific experiences of discrimination often lead LGBTI people to either quit their current position or take an unanticipated leave of absence. Transgender people in particular face discrimination and harassment when seeking and holding employment in any type of organization. The effects of this workplace discrimination are worrisome, as those who experience it have lower incomes and higher rates of material deprivation and are more likely to be unemployed. 139. Taking action to promote LGBTI inclusion is the right thing to do and it also makes economic sense. There is increasing evidence that links exclusion with detrimental health, education, and employment outcomes for LGBTI people, aggregating to broader impacts on the overall economy. 98 These effects can be mitigated with increased public acceptance of LGBTI people.99 The data contained in this report document the challenges currently experienced by LGBTI people in Serbia. Addressing these challenges will not only ensure that the rights of all Serbia’s citizens , regardless of their SOGI, are protected, respected, and fulfilled, but will also bring benefits to the country’s society and economy and to the wider region. More open and inclusive societies are more likely to make the most of their entire stock of human capital, more open and inclusive cities are better placed to attract international capital and talent, and more open and inclusive countries make attractive international tourist destinations. The inclusion of LGBTI people is important in and of itself —and it is also the economically smart thing to do. 98 For example, C. Banks, “The Cost of Homophobia: Literature Review of the Economic Impact of Homophobia in Canada,” Community - University Institute for Social Research (Saskatoon: University of Saskatchewan, 2004). See also G. Becker, The Economics of Discrimination, 2nd ed. (Chicago: University of Chicago Press, 1971); and M. V. L. Badgett, “The Economic Cost of Stigma and the Exclusion of LGBT People: A Case Study of India” (Washington, DC: World Bank, 2014), http://documents.worldbank.org/curated/en/527261468035379692/pdf/940400WP0Box380usion0of0LGBT0People.pdf. 99 For example, Banks, “The Cost of Homophobia.” See also Becker, The Economics of Discrimination. 68 6.2. Recommendations and next steps 140. The primary objective of the research was to provide Serbian policy makers, civil society, and development partners with evidence on the socioeconomic status of LGBTI people and their development challenges. This evidence should allow policy makers and others to identify areas in need of urgent attention to improve the inclusion of LGBTI people in Serbia. This is particularly important for Serbia as a candidate for EU membership and in light of the recommendations provided by the European Commission (EC) in its 2019 enlargement report for Serbia, especially with regard to Chapters 23: Judiciary and Fundamental Rights, and Chapter 24: Justice, Freedom, and Security.100 The EC stated that, “fundamental rights are largely enshrined in the legislation in the Western Balkans. However, serious efforts are needed to step up implementation.” 101 In its 2019 report on Serbia, the EC specifically highlighted that, “Serbia needs to step up measures to protect the rights of persons facing discrimination, including LGBTI persons, persons with disabilities, persons with HIV/AIDS and other vulnerable individuals.”102 As part of the EU accession process, Serbia has developed action plans to combat discrimination and uphold human rights, including for LGBTI people, as outlined in the Fundamental Rights Charter. This survey provides new data points on implementation gaps and can inform the country’s response to the EC recommendations. The data can also be used as a baseline for future action plans to measure the impact of reforms. Over time, Serbia should conduct follow-up surveys to track results on reducing discrimination against, and the exclusion of, LGBTI people and progress under Chapters 23 and 24. 141. The findings of this study can be used to inform policy dialogue with the Government and LGBTI CSOs in Serbia to identify specific shortcomings in policy and program implementation. Addressing such implementation gaps would help Serbia to respond effectively to the requirements of Chapters 23 and 24 of the EU accession process. Building on this research, the World Bank, EC, and other development partners should work closely together to provide Serbia with the necessary support to ensure the effective implementation of existing laws protecting the rights of LGBTI people. 142. The Yogyakarta Principles are a set of international principles relating to sexual orientation and gender identity. They provide a concise and internationally recognized set of standards that states should comply with to ensure that the human rights of LGBTI people are fully protected. The Yogyakarta Principles complement many of the EC’s recommendations and Serbia is encouraged to implement reforms that are in line with those Principles to address the violence, discrimination, harassment, and stigma that LGBTI people face. Specifically, reforms should focus on promoting and protecting the right to “decent and productive work, to just and favourable conditions of work and to protection against unemployment, without discrimination on the basis of sexual orientation or gender identity.”103 In light of the survey findings, the right to participate in public life, especially “the right to have equal access to all levels of public service and employment in public functions, …. without discrimination on the basis of sexual orientation or gender identity” 104 also requires special attention. 100 European Commission, “Serbia 2019 Report” (Brussels: European Commission, 2019), https://ec.europa.eu/neighbourhood- enlargement/sites/near/files/20190529-serbia-report.pdf . 101 European Commission, “2019 Communication on EU Enlargement Policy” (Brussels: European Commission, 2019) https://ec.europa.eu/neighbourhood-enlargement/sites/near/files/20190529-communication-on-eu-enlargement-policy_en.pdf 102 European Commission, “Serbia 2019 Report” (Brussels: European Commission, 2019) 103 Principle 12. 104 Principle 25. 69 Serbia would also do well to “develop and implement programmes to address discrimination, prejudice and other social factors which undermine the health of persons because of their sexual orientation or gender identity.”105 143. The results of the LGBTI SILC identified a number of areas that warrant special focus: expanding the evidence base; increasing LGBTI representation in public sector employment; improving trust in government and institutions; and bolstering awareness and capacity to effectively address transgender issues. Expanding the LGBTI evidence base 144. Researchers, advocates, and policy makers should delve further into the available data to inform their work. This is the first known attempt to collect large-scale survey data comparing the outcomes of LGBTI people with the Serbian population at large. This report presents the key findings, but the data set is rich and could be used for further analyses, including of the different experiences of LGBTI subgroups. Complex statistical models could, for example, uncover further relationships between income, workplace discrimination, and material deprivation. The full data set is available here106. 145. The LGBTI data gap remains large, and further research and data collection are necessary to better understand the lived experience of LGBTI people and the challenges they face. This report represents a quantitative baseline on the status of LGBTI people in Serbia. The Statistical Office of the Republic of Serbia should align itself with statistical agencies in other countries and begin collecting LGBTI-disaggregated data that provide up-to-date evidence and allow the kinds of comparisons between the LGBTI population and the general population that are needed to build more inclusive policies and programs at the country level. This would allow the measurement of any changes over time to ascertain whether policies and programs are addressing LGBTI exclusion. 146. This report revealed the particular challenges faced by transgender people. There are persistent difficulties, also faced in the conduct of this research, in reaching large numbers of transgender people through surveys. Special data collection efforts should be undertaken to ensure a full response from the transgender community. Furthermore, additional efforts are necessary to ensure that the most marginalized parts of the LGBTI community, for example LGBTI people in rural areas, LGBTI Roma, or LGBTI persons with disabilities, are included in future data generation initiatives. Increasing LGBTI representation in public sector employment and strengthening workplace protections for LGBTI people 147. With fewer LGBTI state employees, it is more likely that policies and programs overlook the needs of LGBTI Serbians. Moreover, the public sector represents a source of stable employment. It is thus important that the public sector adopt active programs to increase the number of LGBTI employees through efforts to improve recruitment and workplace culture. To sensitize public servants, regular capacity building and sensitization campaigns across all relevant government 105 Principle 17. 106 https://microdata.worldbank.org/index.php/catalog/3486 70 branches, including for teachers, social workers, health care providers, and justice sector officials, should be conducted. Such training programs should be designed in close consultation with Serbian LGBTI organizations to ensure sensitivity, relevance, and sustainability. The survey findings clearly suggest that there are implementation gaps in workplace protections for LGBTI people in Serbia. The Government, donors, and NGOs should consider raising awareness of these rights even among LGBTI people themselves. 148. There is a need to combat SOGI discrimination in the labor market. Article 18 of the Serbian Labor Code prohibits direct and indirect forms of discrimination based on sexual orientation. Yet, the survey found that LGBTI people frequently experience discrimination that leads to lower incomes and at times forces them to take unpaid leave or quit their job all together. Although sexual orientation and sex are formally protected categories (and possibly gender and gender identity under “any other personal quality”), there appears to be a disparity between this formal protection and the actual experiences of LGBTI people in the workforce or when seeking employment. It is important to further examine whether and how LGBTI people who experience workplace discrimination report such incidents to the authorities. LGBTI respondents’ low trust in the legal system may lower the likelihood that they will report any violations. Building greater trust in the legal system among LGBTI people, coupled with the successful implementation of Article 18, may allow formal protections to be more greatly enforced. In 2019, the EC confirmed that Serbia had made substantial progress in securing fundamental rights in its laws and institutions but also noted that further efforts are needed to address the disadvantages faced by vulnerable populations. This survey confirms the existence of implementation gaps that could be strengthened for LGBTI people in the workplace. 149. Safe reporting structures are needed to encourage LGBTI people to report workplace discrimination without fear of exposure, retaliation, or further discrimination. Similarly, LGBTI people need to feel assured that their cases will be handled seriously and professionally and that actions will be taken to bring perpetrators to account. The Ministries of Justice and the Interior in Serbia are therefore encouraged to identify ways to improve the treatment of LGBTI people in the justice system. Rights awareness and capacity building are needed for justice personnel, including police, prosecutors, judges, and staff. The EC’s 2018 Annual Enlargement packages for Serbia provide detailed recommendations for governments at all levels on the judiciary and fundamental rights, as well as justice, freedom, and security. Improving LGBTI trust in government and institutions 150. LGBTI people’s trust in Serbia’s social, political, and legal institutions needs to be built so that they can exercise full citizenship and experience all the benefits of development. The Serbian Government should use the survey findings to begin a consolidated outreach program to LGBTI communities to further understand the most critical reasons for this lack of trust and take actions to address them. The EU should support these actions through Chapters 23 and 24 of the accession process. Follow-up surveys should be conducted to track progress in improving trust. Bolstering awareness and capacity to effectively address transgender exclusion 151. Concerted effort needs to be taken to address the particular disadvantages faced by transgender people and other highly vulnerable LGBTI groups. The Government should review the 71 legal code to ensure that gender identity protections are enhanced and are at least on par with those provided for sexual orientation. The health disparities between LGBTI people—especially transgender people—and similarly situated members of the general population indicate that safe spaces should be created where LGBTI people can receive specialized services and support. LGBTI people also report a low life satisfaction, which is particularly low for transgender people. The Government should put in place programs to ensure that legal protections are actually realized and enforced, and general government public health campaigns against (domestic) violence should contain LGBTI components. Civil society groups that work with transgender and other vulnerable LGBTI populations should be supported to enhance their services. Moreover, the Government, development partners, and other donors are encouraged to further build the capacity of existing LGBTI organizations more widely and actively support the creation of services for rural communities in particular. A part of this capacity building should be to engage organizations in the collection of data on LGBTI people, especially those residing in rural and other areas without dependable access to the internet. Governments at all levels should also do more to ensure that LGBTI people, especially those who experience workplace discrimination, receive the official services they need. 72 Annex 1. Landing Page of the LGBTI Survey (only text included) YOUR EXPERIENCE COUNTS! What is this LGBTI survey about? This is the biggest ever survey of lesbian, gay, bisexual, transgender, and intersex people (LGBTI) in this region. The survey is looking for responses from anyone who consider themselves to be LGBT and/or I, who is aged 18 years and above and lives in Albania, Bosnia and Herzegovina, Croatia, Kosovo, North Macedonia, Montenegro, Serbia, or Slovenia. Who is conducting the survey? The survey is being carried out by research agency Ipsos on behalf of the World Bank, in partnership with the Williams Institute at UCLA and ERA - LGBTI Equal Rights Association for Western Balkans and Turkey. Your answers will help us know about the life of LGBTI people in your country. This information will help governments and others know where the biggest problems are and push them to make changes to help LGBTI people. How is your privacy protected? Your answers will be anonymous, and your privacy will be protected. At no point will it be possible to identify any of the respondents personally. The survey is protected by a Linux data server with firewalls installed, which uses: • HTTPS protocol for secure communications, ensuring that the contents of communications between you and the server where the data will be stored cannot be read or forged by any third party. • SSL protocol, creating a secure connection between you and the server, encrypting sensitive information being transmitted through the web page. Your participation in the survey is very important. Please contribute to its effectiveness by responding only once. Your participation in this survey is strictly voluntary. The questions will take up to around 20 minutes to answer. The survey will be open until March 31, 2017. Please note that once you quit the survey, you won’t be able to continue, and it will be necessary to start the survey from the beginning. Throughout the survey, once you click on the NEXT button, it won’t be possible to go back to the previous questions. If you’re filling out the survey on your mobile phone, please do it in landscape mode. If you are experiencing any technical difficulties or have some additional questions, feel free to contact us. 73 Annex 2. Detailed Methodology A systematic analysis of peer-reviewed literature demonstrates a paucity of data on lesbian, gay, bisexual, transgender, and intersex (LGBTI) people outside of the United States and Western Europe in virtually all sectors .107 For example, only 2 percent of the studies identified on suicide risk were in the global south,108 and a review of studies on violence against sexual and gender minorities carried out between 1994 and 2016 reveals that only eight out of 76 occurred in less developed countries.109 No known studies of peer victimization of adolescents due to gender identity and expression, 110 research on disparities in health among lesbian and bisexual women, 111 or reviews of intimate partner violence among lesbians112 come from locations outside of the United States or Western Europe. Another systematic review of intimate partner violence among MSM (men who have sex with men) identified only two such studies.113 Very few data sets and research on employment and identity formation have been produced outside of the United States and Western Europe. Almost all research on discrimination based on sexual orientation in the workplace 114 has been in the United States and the United Kingdom. A global meta-analysis of wage studies indicates that gay men make 11 percent less than their heterosexual counterparts.115 Except for Greece, most data are from advanced economies.116 Furthermore, a review of research on identity formation in sexual minorities noted historically low levels of knowledge about the process of identity formation and how it varies across ethnicity, social class, or region or between Western and non-Western societies.117 Key sections of the Survey on Income and Living Conditions (SILC) were chosen and complemented by a set of additional questions to enable analysis of the responses according to various statistical models. The questions on demographic characteristics were adapted to meet the needs of the survey for LGBTI people: identification of sex assigned at birth, sexual orientation, current gender identity, and intersex status. The questions build on international experience and good practice for surveying LGBTI populations. 107 J. G. L. Lee, T. Ylioja, and M. Lackey, “Identifying Lesbian, Gay, Bisexual, and Transgender Search Terminology: A Systema tic Review of Health Systematic Reviews,” PLoS ONE 11, no. 5 (2016): e0156210. 108 Martin Plöderl and Pierre Tremblay, “Mental Health of Sexual Minorities: A Systematic Review,” International Review of Psychiatry 27, no. 5 (2015): 367–85. 109 Karel Blondeel and others, “Violence Motivated by Perception of Sexual Orientation and Gender Identity: A Systematic Review,” Bulletin of the World Health Organization 96, no. 1 (2018): 29–41L. 110 Kate L. Collier and others, “Sexual Orientation and Gender Identity/Expression Related Peer Victimization in Adolescence: A S ystematic Review of Associated Psychosocial and Health Outcomes,” Journal of Sex Research 50, nos. 3–4 (2013): 299–317. 111 J. M. Simoni and others, “Disparities in Physical Health Conditions Among Lesbian and Bisexual Women: A Systematic Review of Population-Based Studies.” Journal of Homosexuality 64, no. 1 (2017): 32–44. 112 L. Badenes-Ribera and others, “Intimate Partner Violence in Self-Identified Lesbians: A Systematic Review of Its Prevalence and Correlates,” Trauma Violence Abuse 17, no. 3 (2016): 284–97. 113 Ana Maria Buller and others, “Associations between Intimate Partner Violence and Health among Men Who Have Se x with Men: A Systematic Review and Meta-Analysis,” PLoS Medicine 11, no. 3 (2014): e1001609. 114 Emer Ozeren, “Sexual Orientation Discrimination in the Workplace: A Systematic Review of Literature,” Procedia - Social and Behavioral Sciences 109 (2014): 1203–15. 115 Marieka Klawitter, “Meta-Analysis of the Effects of Sexual Orientation on Earnings,” Industrial Relations: A Journal of Economy and Society 54, no. 1 (2015): 4–32. 116 Ibid. 117 Ritch C. Savin-Williams and Kenneth M. Cohen, “Developmental Trajectories and Milestones of Lesbian, Gay, and Bisexual Young People,” International Review of Psychiatry 27, no. 5 (2015): 357– 66. 74 The survey sampling method and recruitment is consistent with previous studies of LGBTI populations.118 Data collection was made possible by programming the questionnaire in the Serbian language using IPSOS’s own data entry program. All the logical checks in the questionnaire were implemented. The data collection program guaranteed full protection of respondents’ privacy and confidentiality. A computer-assisted web interviewing method was used to conduct interviews. The survey was available in all the main web browsers, including Internet Explorer, Mozilla Firefox, Chrome, Safari, and Opera, and was adjusted for use on different types of devices —desktop computers, personal computers/laptops, tablets, and smartphones. Since respondents had to “opt in” to participate in the survey, the collected data is based on self-selected participants and is not a random sample. Sample and weighting The lack of population estimates and basic information about the characteristics of the LGBTI population made it necessary to use a self-selected, nonprobability sample for the survey.119 In total, 998 LGBTI respondents took part in the survey. A basic overview is provided in the tables below — unweighted and weighted data (table A.2.1). Table A.2.1: Sample Realization, Unweighted and Weighted Data LGBTI group Number Percentage Number Percentage (unweighted) (unweighted) (weighted) (weighted) Intersex People 53 5 54 5 Transgender People 43 4 57 6 Lesbians 164 16 190 19 Gay Men 512 51 349 34 Bisexual men 111 11 138 14 Bisexual women 115 12 226 22 TOTAL 998 100 1014 100 The lack of population estimates mentioned above made it impossible to collect a representative sample of the LGBTI population.120 Weighting can be used to adjust sample characteristics to population targets in order to correct for over- or under-sampled groups. The challenge for populations not measured in administrative surveys (e.g., a census) or studies with large sample sizes is that these targets are unknown. To identify appropriate targets, results were collected from about 300 studies of LGBTI populations around the world. These studies were identified by using keyword searches of online databases (Pubmed, JSTOR, Web of Science, Google, and Google Scholar), by examining the references used in studies identified by keyword searches, and by reviewing reports of intergovernmental bodies (the Joint United Nations Programme on HIV/AIDS [UNAIDS], AIDSinfo) and reports of LGBTI and women’s organizations. Key words included: LGBTI (and various combinations of this alphabetism), lesbian, gay, 118 For example, Sandy E. James and others, “The Report of the 2015 U.S. Transgender Survey” (Washington, DC: National Center for Transgender Equality, 2016). 119 Probability sampling, or random sampling, is a sampling technique in which the probability of getting any particular sample may be calculated, which is not the case with non-probability sampling. Namely, in non-probability sampling the samples are gathered in a process that does not give all the individuals in the population an equal chance of being selected. For the LGBTI population, this is the case due to the non-existence of either the population registers or information on characteristics of this population. 120 A representative survey is one in which the sample closely matches the characteristics of the population as a whole. A representative survey of the LGBTI population would mean making a representative sample of LGBTI people that closely matches the characteristics of the LGBTI population. 75 bisexual, transgender, intersex, transgendered, transgenderism, transsexual, transsexualism, trans women, trans men, third gender, non-binary, MSM, WSW (women who have sex with women), same- sex attraction, homosexual, HIV, AIDS, population, prevalence, size, estimation, risk factor, health, MARP (most at risk populations), waria, mak nyah, fakaleiti, hijra, kathoey, and bakla, as well as the translation of English terms such as transgenero, HSH, LSL, МСМ, VIH, SIDA, ゲイ, 同志, 同性戀. About 300 sources of data were identified, including 154 administrative, country-level estimates submitted by national governments to UNAIDS and 150 studies published either as organizational and agency reports or as articles in peer-reviewed journals. Included in all these sources were approximately 520 estimates for various sexual and gender minority groups throughout the world. After excluding studies that were not based on representative samples and studies that were of sexual or gender minority populations outside the scope of the current targets (e.g., MSM), 28 studies about sexual orientation remained in the data set to be used for calculation of the weights. Using data from the 28 remaining studies, a hierarchical, Bayesian meta-analysis was performed on the data from the 28 studies to derive targets, accounting for the diversity of countries and populations included. The sample obtained in Serbia in the current study was reweighted for these targets. The greatest need for, and advantage of, these targets was to ensure that those assigned female at birth were weighted appropriately in the resultant survey data. Some of the outreach methods (such as engaging gay social networks like Grindr and PlanetRomeo) had the potential to recruit more individuals who were assigned male at birth, which could alter the results toward the narratives of that group of individuals. The weights adjust for this, ensuring a narrative common to individuals assigned male or female at birth. A summary of the population targets is provided below (table A.2.2). Table A.2.2. Summary of the Population Targets Assigned male at birth Assigned female at birth Heterosexual or straight 92.8 94.6 Gay or lesbian 5.0 2.8 Bisexual 2.2 2.6 Total 100 100 Additional steps were taken to draw more accurate comparisons between the LGBTI sample and the sample of the general population used in the SILC. We were faced with two options when approaching the weighting adjustments: (1) weight the LGBTI sample to the target demographics in the SILC, a process that has been conducted on other similar samples in other contexts;121 or (2) weight the SILC in Serbia to the target demographics in the LGBTI sample. We chose the latter for the present report because the LGBTI sample is uniquely drawn from urban settings with greater educational attainment than the general population and a greater probability of being single. It was believed, for example, that adjusting the LGBTI sample would up-weight individuals living in rural areas too greatly. A sample matching procedure was performed utilizing a covariate balancing propensity score. 122 We first subset the results of the SILC in Serbia to individuals aged 60 or younger who had a computer in For example, in the United States, see Sandy E. James , “The Report of the 2015 U.S. Transgender Survey.” 121 Kosuke Imai and Mark Ratkovic, “Covariate Balancing Propensity Score,” Journal of the Royal Statistical Society, Series B 76, part 1 (2014): 122 243–63. 76 the household and internet access. We made these data preprocessing decisions in order to more accurately weight to a population that is similar to the LGBTI sample. Since the LGBTI sample was recruited to take an online survey, it made sense to subset the SILC in Serbia to these households. In addition, the vast majority of the LGBTI sample is aged 60 or younger. The matching procedure took the following details into account: age, sex assigned at birth, educational attainment, marital status, region, and area of residence. The model was estimated to the unweighted LGBTI sample. These factors were considered prior to the LGBTI data collection by analyzing the official SILC in Serbia. A machine learning process known as chi-square automatic interaction detection (CHAID) was used on the Serbian general population sample to determine the relevant factors that predict an individual’s employment status. This assisted in identifying key questions that would be included in the LGBTI sample for subsequent weighting. CHAID was selected due to the unordered, categorical nature of the employment variable, which excluded other common machine learning methods such as LASSO. After the model estimated the weights, the SILC in Serbia weights were readjusted for sex assigned at birth to match the weighted distribution of the LGBTI sample. There were very few differences between the two samples after this weighting procedure, offering more accurate comparisons. The following tables summarize demographics in the SILC before and after weighting with the weighted estimates from the LGBTI sample also presented for comparison. As can be seen, unweighted SILC demographics are quite different from the LGBTI sample: the LGBTI sample is younger, more educated, and more often an urban resident (tables A.2.3– A.2.7). The weighted SILC minimizes these differences. Table A.2.3. Age – Structure of the LGBTI Sample, Unweighted and Weighted Official SILC (percent) Age LGBTI Sample Unweighted SILC Survey Weighted SILC Survey 17–29 55.14 32.16 53.78 30–44 39.90 36.52 41.26 45–67 4.95 31.32 4.96 Table A.2.4. Education – Structure of the LGBTI Sample, Unweighted and Weighted Official SILC (percent) Education LGBTI Sample Unweighted SILC Survey Weighted SILC Survey Low 3.06 15.13 2.82 Medium 39.54 62.44 39.37 High 56.65 22.43 57.81 Missing 0.75 0.00 0.00 Table A.2.5. Region – Structure of the LGBTI Sample, Unweighted and Weighted Official SILC (percent) Region LGBTI Sample Unweighted SILC Survey Weighted SILC Survey RS11 52.78 23.48 52.90 RS12 27.79 33.08 27.68 RS21 11.13 22.74 10.89 RS22 8.30 20.70 8.52 77 Table A.2.6. Population Density – Structure of the LGBTI Sample, Unweighted and Weighted Official SILC (percent) Population density LGBTI Sample Unweighted SILC Weighted SILC Survey Survey Densely populated 72.25 36.05 72.16 Medium populated 23.53 27.81 23.63 Sparsely populated 4.22 36.14 4.22 Table A.2.7. Marital/Partnership Status – Structure of the LGBTI Sample, Unweighted and Weighted Official SILC (percent) Marital/partnership status LGBTI Sample Unweighted SILC Survey Weighted SILC Survey Not Married or Partnered 84.67 44.87 86.28 Married or Partnered 15.33 55.13 13.72 78 Annex 3. Detailed Demographic Analysis Sex assigned at birth, gender identity, being intersex, and sexual orientation In the survey, LGBTI respondents were asked to provide information about the sex they were assigned at birth. A slightly higher percentage of respondents were assigned male sex at birth (53 percent) compared to those who were assigned female sex at birth (47 percent). In line with the higher percentage of those assigned male sex at birth, a higher percentage of respondents describe their gender as male (51 percent), while 39 percent describe their gender as female. Of the total number of respondents, 4 percent identify as transgender and 6 percent as gender fluid. Among respondents who were assigned female sex at birth, four out of five describe their gender as female (80 percent), while 5 percent describe themselves as male, 6 percent as transgender, and 9 percent as gender fluid. On the other hand, among respondents who were assigned male sex at birth, almost all describe their gender as male (93 percent), while 2 percent describe themselves as female, 2 percent as transgender, and “only” 3 percent as gender fluid (three times less than those who were assigned female sex at birth). Figure A.3.1. Gender Identity of LGBTI Respondents (percent) Female gender Male gender Transgender Gender fluid LGBTI average 39 51 4 6 Female at birth 80 5 6 9 Male at birth 2 93 23 0 20 40 60 80 100 Question: How would you describe your gender identity? Base: All LGBTI respondents. Out of all LGBTI respondents, 5 percent described themselves as intersex. There are no differences between those who were assigned female or male sex at birth when it comes to the percentage of intersex respondents among them (5 percent of those who were assigned female sex at birth and 6 percent of those who were assigned male sex at birth). Focusing on sexual orientation, the highest percentage among respondents identify as bisexual (42 percent) and a slightly lower percentage as gay (37 percent). Additionally, every fifth LGBTI respondent identifies as lesbian (19 percent). It should be noted that 2 percent of the respondents described themselves as heterosexual or straight but were included in the survey because they identified as either transgender or intersex. Certain differences concerning the sexual orientation of LGBTI respondents are noticeable between those who were assigned female sex and those who were assigned male sex at birth. Although those who were assigned male sex at birth predominately 79 describe themselves as gay (69 percent), and in a much lower percentage as bisexual (30 percent), those who were assigned female sex at birth describe themselves more often as bisexual women (55 percent) than as lesbian (40 percent). In addition, 3 percent of LGBTI respondents who were assigned female sex at birth identify as gay, while 0.4 percent of those who were assigned male sex at birth identify as lesbian. No differences related to the percentage of those who describe themselves as straight were registered. Figure A.3.2. Sexual Orientation of LGBTI Respondents (percent) Heterosexual/Straight Lesbian Gay Bisexual LGBTI average 2 19 37 42 Female at birth 3 40 3 55 Male at birth 1 69 30 0 20 40 60 80 100 Question: Do you consider yourself to be…? Base: Total LGBTI sample. Sex assigned at birth, age, and education LGBTI respondents are mainly young. The average age is 30 (29.6), which is far younger than among the general population (48.8123).124 Three out of four respondents are younger than 35 years (39 percent aged 18–25 and 38 percent aged 26–35), while one in five is aged 36–45 years (19 percent). Only 5 percent of LGBTI respondents are older than 45. Respondents who were assigned female sex at birth are on average much younger than those who were assigned male sex at birth. Although half of those who were assigned female sex at birth are 25 years old or younger (50 percent), this is true of less than a third of those who were assigned male sex at birth (29 percent). In line with this, among those who were assigned male sex at birth, there is a higher percentage of older LGBTI respondents. Every fourth LGBTI person who was assigned male sex at birth is between 36 and 45 years old (24 percent), while only 14 percent of respondents who were assigned female sex at birth are in that age group. The same trend is visible in the oldest age group (over 45 years old)—7 percent of respondents who were assigned male sex at birth are over 45 years old compared to just 2 percent among those who were assigned female sex at birth. Statistical Office of the Republic of Serbia, “2011 Census of Population” (see n. 37). 123 The average age refers to the LGBTI respondents and those within the general population who are 18 or more years old, given that only 124 people over 18 participated in the LGBTI survey. 80 Figure A.3.3. Age of LGBTI Respondents (percent) 18-25 26-35 36-45 46 and higher LGBTI average 39 38 19 5 Female at birth 50 35 14 2 Male at birth 29 40 24 7 0 20 40 60 80 100 Question: Year of birth? Base: Total LGBTI sample. Among LGBTI respondents, a higher percentage of the young (up to 25 years old) are bisexual and lesbian women (54 and 45 percent, respectively) compared to bisexual and gay men (37 and 26 percent, respectively). Transgender respondents are also young, with 60 percent under 26 years old. By contrast, intersex respondents are on average among the oldest in the survey (only 25 percent of intersex respondents are 25 years old or younger). This age distribution is one of the possible explanations for some of the differences in income and employment, among other factors. It is hard to say whether those differences have more to do with LGBTI subgroups themselves or with age difference by sex assigned at birth, but nevertheless, age differences should be kept in mind when drawing conclusions about differences among LGBTI groups. At times, age-adjusted estimates for LGBTI subgroups account for the distributional differences in age. Figure A.3.4. Age of Different Groups of LGBTI Respondents (percent) 18-25 26-35 36-45 46 and higher LGBTI average 39 38 19 5 Transgender 63 21 14 2 Bisexual women 53 37 10 Lesbian 43 35 20 2 Bisexual men 36 32 23 9 Gay 26 44 23 7 Intersex 25 36 34 4 0 20 40 60 80 100 Question: Year of birth? Base: Total LGBTI sample. 81 LGBTI respondents are not only younger than the general population, they are also more highly educated. Only 3 percent of respondents have only primary or incomplete secondary school education, while in the general population this percentage is 10 times higher (32 percent125). Four out of 10 respondents have secondary school education (40 percent), which is similar to the general population (51 percent), but the highest percentage of LGBTI respondents have college or higher education (57 percent). In contrast, among the general population, only 17 percent pursued college or higher education. No differences among LGBTI groups concerning education are registered, with the exception of transgender respondents who more often than others do not have college or higher education. Figure A.3.5. Education of LGBTI Respondents (percent) Primary school 1 Incomplete secondary school 2 Secondary school (3 or 4 years) 40 College or higher education 42 Master 13 Doctorate 2 Refuse to answer 1 0 5 10 15 20 25 30 35 40 45 Question: What is your highest acquired education level? Base: Total LGBTI sample. Place of living Almost all LGBTI respondents live in urban areas, and only 4 percent in rural areas. This is quite different from the general population, 41 percent of whom live in rural areas. 125 Statistical Office of the Republic of Serbia, “2011 Census of Population.” 82 Figure A.3.6. Place of Living of LGBTI Respondents (percent) Big city (150,000 inhabitants or 72 more) Outskirts or suburb of a big city 4 Small city (fewer than 150,000 19 inhabitants) Rural area 4 0 20 40 60 80 Question: Where do you currently live? Base: Total LGBTI sample. In addition to being more urban, the LGBTI sample is also skewed toward those living in the Belgrade region, with every second LGBTI respondent living in the capital (53 percent). Little more than a quarter of respondents live in Vojvodina (28 percent) and the rest live in Sumadija and western Serbia (11 percent) or south and east Serbia (8 percent). This again greatly differs from the unweighted general population, one-fourth of whom live in each region (23 percent in Belgrade, 27 percent in Vojvodina, 28 percent in Sumadija and west Serbia, and 22 percent in south and east Serbia). This dominant position of the capital city (Belgrade) is in line with the previously noted skewness toward urban areas, since two-thirds of those who reported living in a big city are actually living in Belgrade (68 percent). Demographic profile of the average LGBTI respondent The average LGBTI respondent is about 30 years old, has a higher education, and lives in a big city (predominantly Belgrade). This profile differs from the profile of the average citizen of Serbia (the general population), who is about 50 years old, has a secondary education, and has an equal chance of living in an urban or rural area or any region in Serbia. Characteristics of the average LGBTI respondent correspond to the characteristics of the (economically) better-off part of the general population. In light of this fact, all analysis is focused on comparisons between the position of the LGBTI respondents and the matched general population at the national level (unless otherwise stated). Since the official SILC does not document sexual orientation, matching to the general population makes for more accurate comparisons.126 Additional details of working experience and the average number of paid jobs of LGBTI respondents On average, LGBTI respondents in Serbia (excluding those who have never worked) have about eight years of work experience (7.9 years). Certain gender differences can be noticed (8.8 years for men and 6.8 years for women), as well as some differences among LGBTI groups (from intersex 126Since the official SILC does not ask about sexual orientation, we have no way of estimating the percentage of LGBTI responden ts in the official SILC sample. 83 respondents with 9.2 years of experience to bisexual women with 6.1 years). It should be kept in mind that these observed differences are most likely caused by age differences between LGBTI respondents of different gender and different LGBTI subgroups. Younger generations (aged 18–25) have on average two years of experience, while those aged over 45 have 10 times as much work experience (22.4 years). Figure A.3.7. Total Number of Years at All Paid Jobs LGBTI average 8 Intersex 9 Bisexual men 9 Gay 9 Transgender 7 Lesbian 7 Bisexual women 6 Female 7 Male 9 18-25 2 26-35 6 36-45 14 46 and higher 22 0 5 10 15 20 25 Question: How many years in total have you worked at all of your paid jobs? Base: Those LGBTI respondents who worked at some point at a job for at least six months. LGBTI respondents in Serbia had on average four different jobs (3.8). As expected, those with more work experience had a greater variety of jobs during their career. Some gender differences are noticeable (men had on average 4.1 jobs and women 3.6 jobs), as well as some differences among LGBTI groups (from bisexual men with 4.2 jobs to lesbian women with 3.5 jobs on average), but those are caused by age differences between LGBTI respondents of different gender and LGBTI subgroups. Although younger generations (aged 18–25) on average had already changed jobs twice (2.3), those aged 26–35 had changed jobs twice as many times (4.2) and those over 35 had changed jobs on average as many as six different times in their working careers (5.8). Those who are self-employed on average have changed more frequently (six times) than those who have piecework contracts or contracts for temporary and occasional jobs (5.4). Also, those working in the private sector on average had the same number of jobs in their working careers (4.8) as those working in the nongovernmental sector (4.7). 84 Figure A.3.8. Average Number Paid of Jobs LGBTI average 3.8 Bisexual male 4.2 Gay 4.1 Transgender 4.1 Intersex 3.7 Bisexual female 3.6 Lesbian 3.5 Female 3.6 Male 4.1 18-25 2.3 26-35 4.2 36-45 5.8 46 and higher 5.7 Employee 4.4 Self-employed 6.0 With contract, but without emp. relationship 5.4 Public (state-owned) sector 4.0 Private sector 4.8 Non-governmental sector 4.7 0 1 2 3 4 5 6 Question: How many years in total have you worked at all of your paid jobs? Base: Those LGBTI respondents who worked at some point at a job for at least six months. 85 Annex 4. Additional Graphs Figure A.4.1. Ownership of the Dwelling Respondents Live In (percent) LGBTI sample General population 60 Yes, full-fledged owner 85 6 Yes, but paying off the mortgage 1 26 Lessee of the whole or a part of the dwelling 5 7 The housing is free 10 0 20 40 60 80 100 Question: Are you or some of the members of your household the owner of the dwelling you live in or do you lease the dwelling? Base: Total LGBTI sample and the general population. Figure A.4.2. At-Risk-of-Poverty Rate by Main Demographic Characteristics (percent) LGBTI sample General population Average 17 16 Employee 14 8 Self-employed 17 21 With contract, but without emp. relationship 20 Unemployed and inactive 27 25 Pupil/student 17 22 Private sector 13 11 Public (state-owned) sector 20 5 Nongovernmental sector 18 Contract for indefinite period of time 12 9 Contract for definite period of time 22 14 0 10 20 30 Base: Total LGBTI sample and the general population. 86 Figure A.4.3. Arrears in Paying Mortgage or Rent Payments, Utility Bills, Hire Purchase Installments, or Other Loan Payments (percent) Yes No Does Not Apply Arrears in LGBTI sample 45 51 4 category any General population 35 65 LGBTI sample 40 55 5 services Utility General population 31 68 1 of other Rent for a LGBTI sample 11 57 33 dwelling General population 3 20 77 mortgage Repayment LGBTI sample 21 54 26 credits General population 9 31 60 Repayment LGBTI sample 6 51 43 dwelling for the of the General population 3 17 80 0% 20% 40% 60% 80% 100% Question: In the previous 12 months, were you in a situation of being late with the payment of costs or the repayment of credits owing to financial difficulties? Base: Total LGBTI sample and the general population. Figure A.4.4. Percentage of Respondents Who Cannot Afford Specific Items (Not Included in the Material Deprivation Indicators) (percent) LGBTI sample General population 24 Regular activities during free time 12 20 Replacing worn-out clothes 9 Spending a smaller quantity of money without obligation 20 to consult with anyone 14 Two pairs of footwear (including one pair which is for all 13 weather conditions) 9 13 Meeting with friends/family for a drink/lunch/dinner 7 0 5 10 15 20 25 Question: Can you afford? Base: Total LGBTI sample and the general population. 87 Figure A.4.5. Reasons for Working Less than 30 Hours (percent) LGBTI sample General population That is full time 36 34 22 Education and training 16 17 Do not want to work longer 8 Cannot find a job with longer working hours 16 26 Family or personal reasons 6 3 Illness or inability 3 2 1 Other reasons 12 0 4 8 12 16 20 24 28 32 36 40 Question: Main reason why you worked for less than 30 hours? Base: Those LGBTI respondents and respondents from the general population who defined their current main status as employed and who reported working less than 30 hours per week. Figure A.4.6. Openness about Being LGBTI with Current Work Colleagues (percent) None A few Most All Do not apply to me LGBTI average 47 27 9 13 4 Live alone 49 25 7 14 5 Live with parents 48 28 12 7 5 Live with unmarried partner 31 25 14 27 3 0 20 40 60 80 100 Question: To how many of your current work colleagues are you open/out about your sexual orientation and/or gender identity and/or being intersex person? Base: Those LGBTI respondents who are currently working. * Sample sizes are small for Intersex respondents (n = 28) and Transgender respondents (n = 10). 88 Annex 5. Coding Steps for Personal and Household Income Personal income The LGBTI respondents were asked to report their average monthly income. If they did not know or refused to answer the question, they were asked in a follow-up question to identify their average monthly income within the following ranges: up to 20,000 Serbian dinars, 20,001–25,000, 25,001– 35,000, 35,001–45,000, 45,001–60,000, 60,001–80,000, 80,001–100,000, 100,001–150,000, 150,001– 200,000, and 200,001 or more. Income was imputed as a continuous variable for respondents who refused to answer the initial question and answered this follow-up question. Respondents who selected an income range were randomly assigned an income to be within the income bracket relying on a random uniform distribution (i.e., there were equal probabilities of having an income within the interval). The top-code for the highest bracket was placed at 300,000 Serbian dinars. For LGBTI respondents who refused to answer both income questions (n = 84, 11 percent of the sample), a hot-deck imputation process was used to impute these values. 127 Since income is not normally distributed, a median regression was fitted taking into account age, LGBTI identification, sex assigned at birth, educational attainment, and location of residence (e.g., big city versus a rural area).128 Once the median regression was fitted, predicted values from the regression were used as estimates of income for the 84 LGBTI respondents who did not answer this question. The median income of the sample without this imputation was €252, while the median was €256 with the imputation. Household income The LGBTI respondents were asked to report their household’s average monthly income , excluding their own income. If LGBTI respondents did not know or refused to answer the question, they were asked in a follow-up question to identify their average monthly income within the following ranges: up to 20,000 Serbian dinars, 20,001–25,000, 25,001–35,000, 35,001–45,000, 45,001–60,000, 60,001– 80,000, 80,001–100,000, 100,001–150,000, and 200,001 or more. This variable was initially cleaned because it was apparent that some respondents incorporated their personal income into their household income. Household income was then imputed as a continuous variable for respondents who refused to answer the initial question and answered this follow-up question. Respondents who selected an income range were randomly assigned an income to be within the income bracket relying on a random uniform distribution (i.e., there were equal probabilities of having an income within the interval). The top-code for the highest bracket was placed at 300,000 Serbian dinars. To arrive at an estimate of the total household income, these estimates were combined with the personal income estimates. For LGBTI respondents who refused to answer both household income questions (n = 118, 12 percent of the sample), a hot-deck imputation process was used to impute these values. Since income is not normally distributed, a median regression was fit taking into account age, LGBTI identification, sex assigned at birth, educational attainment, location of residence (e.g., big city versus 127 Teresa A. Myers, “Goodbye, Listwise Deletion: Presenting Hot Deck Imputation as an Easy and Effective Tool for Handling Missi ng Data,” Communication Methods and Measures 5, no. 4 (2011): 297–310. 128 A Vuong closeness test comparing the log-likelihoods of a traditional ordinary least squares regression to a median regression suggested that the median regression is closer to the true data generating process (t = -14.7, df = 711). Quang H. Vuong, “Likelihood Ratio Tests for Model Selection and Non-Nested Hypotheses,” Econometrica 57, no. 2 (1989): 307–33. 89 a rural area), and personal income. Once the median regression was fitted, predicted values from the regression were used as estimates of income for LGBTI respondents who refused to answer the income questions. The median monthly household income of the sample without this imputation was €504, while the median was €489 with the imputation. Due to item response missingness on the demographic variables used in the hot-deck imputation, three respondents were dropped from both the personal income and household income estimates. The hot-deck imputation strategy deviates from the practices initially recommended by the Statistical Office of the Republic of Serbia. Its recommendation was to impute respondents who refused to report their income from a random uniform distribution with a minimum of 20,001 dinars and maximum of 80,000 dinars. This strategy, however, systematically increased the median personal income of the LGBTI respondents to €282. Rather than relying on a random number generator, hot- deck imputation imputes values for income based on the data collected and respondent demographics. This should more accurately reflect the distribution of income for the LGBTI respondents, as the recommendations from the Statistical Office may be a practice more reflective of non-LGBTI populations. Since the official SILC collects data from in-person surveys, instances of response missingness on numerous variables are fewer. Thus, a different strategy was selected. 90 Annex 6. Survey Instrument https://microdata.worldbank.org/index.php/catalog/3486 91 Annex 7. List of Survey Partner Organizations (NGOs) • Geten • Egal Serbia • Da Se Zna! • AS Centar • Come Out • Labris • YUCOM • FemSlam • Rainbow Association • Red Line 92 94