Assessing the effects of the fiscal system on gender disparities in Armenia Pilot-study for the Engendered CEQ methodology Alan Fuchs Tarlovsky & Maria Fernanda Gonzalez Icaza POVERTY, EQUITY AND GENDER GLOBAL PRACTICE (EECPV) FEBRUARY 2023 Abstract This report applies the Engendered Commitment to Equity (E-CEQ) methodology to understand the effects of the fiscal system on gender disparities in Armenia. We leverage data from the Living Conditions Survey (ILCS), the Labor Force Survey (LFS), and administrative sources, and we follow recommendations from the recently produced Methodological Paper on Engendered CEQ, as well as specific recommendations from the Technical Advisory Panel, to empirically test the E-CEQ method in Armenia. Since budget and fiscal data disaggregated at the intra-household level is unavailable in Armenia, as in many other countries, we develop a typology of households, aiming to identify characteristics associated with potential gender disparities for each household surveyed in the ILCS. In addition to calculating the standard CEQ estimators for Armenia, we disaggregate the effects of the fiscal system by household type. We find that the fiscal system in Armenia—particularly, direct taxes—reduces gender disparities by affecting the incomes of households that rely economically on women earners less than proportionally to those that rely on men’s earnings. The results, however, do not necessarily imply that Armenia’s fiscal system reduces overall gender disparities. The lower tax incidence on women may reflect their exclusion from formal labor markets. The results also suggest that childcare and elderly care constrain mostly women’s economic opportunities. This report constitutes a pilot for testing the first generation of E-CEQ studies. The results and scope of this assessment should be expanded in subsequent research in Armenia and other low and middle-income countries. Acknowledgments The authors are grateful to the members of the Technical Advisory Panel, Carolina Sanchez-Paramo, Benu Bidani, Caren Grown, Sarah Iqbal, Jon Jellema, Matthew Wai-Poi, Sailesh Tiwari, and Mariano Sosa for their inputs and comments. The team is also grateful for the financial support generously provided by the Hewlett Foundation. Contents I. Introduction .......................................................................................................................................... 2 II. Poverty and inequality .......................................................................................................................... 4 III. Gender disparities ................................................................................................................................. 5 IV. Overview of the fiscal system .............................................................................................................. 16 V. Standard CEQ results........................................................................................................................... 19 VI. The Engendered (E-CEQ) methodology............................................................................................... 20 VII. Data sources........................................................................................................................................ 23 VIII. Engendered household typologies ...................................................................................................... 23 IX. Engendered CEQ Results ..................................................................................................................... 29 X. Discussion............................................................................................................................................ 35 References................................................................................................................................................... 36 Appendix ..................................................................................................................................................... 38 Page 1 Figures Figure 1. National poverty, 2008-2020.......................................................................................................... 5 Figure 2. Gini coefficient, 2019-2020 ............................................................................................................ 5 Figure 3. Labor force participation rates, 2012-2019 .................................................................................... 7 Figure 4. Gender ratios in tertiary education and economic participation ................................................... 7 Figure 5. Net wage earnings by gender, cohort, and educational attainment .............................................. 8 Figure 6. Population out-of-the-labor force, by category ............................................................................ 10 Figure 7. Population out-of-the-labor force, by reasons for not seeking employment ............................... 11 Figure 8. Youth not in Education and not in Employment (NEET) by age groups, 2019 .............................. 11 Figure 9. Changes in national poverty ......................................................................................................... 20 Figure 10. Changes in inequality.................................................................................................................. 20 Figure 11. Interaction of the fiscal system with gender and intrahousehold dynamics .............................. 21 Figure 12. Share of the population .............................................................................................................. 23 Figure 13. Share of the poor........................................................................................................................ 23 Figure 14. Pre-fiscal poverty, by household typologies ............................................................................... 27 Figure 15. Income sources by household type ............................................................................................ 28 Figure 16. Fiscal incidence on gender gaps, by income level ...................................................................... 31 Figure 17. Gender gaps across the fiscal system ......................................................................................... 34 Figure 18. Marginal income effects, by fiscal interventions and household type ....................................... 34 Tables Table 1. Employment of men and women in 2019 and 2020, by activity ...................................................... 8 Table 2. Gender disparities in time allocation, 2022 ................................................................................... 14 Table 3. Classification of households into categories based on gender relations ....................................... 22 Table 4. Household types based on the ILCS 2020 ..................................................................................... 24 Table 5. Socioeconomic distribution of household types ............................................................................ 26 Table 6. Consumption shares across household types ................................................................................ 28 Table 7. Income estimates by household type and direct fiscal interventions ............................................ 29 Table 8. Income estimates by household type and indirect fiscal interventions ......................................... 30 Table 9. Gender gaps across the fiscal system ............................................................................................ 33 Boxes Box 1. Survey on Gender Disparities in Time Allocation and Household Responsibilities, 2022 ................. 13 Page 2 I. Introduction This report analyses the welfare and distributional incidence of the fiscal system of Armenia from the perspective of gender disparities. The analysis builds on the methodological approach developed by the Commitment to Equity Institute at Tulane University and the World Bank, under the guidance and advice of the Technical Advisory Panel (TAP). Along with companion pilot analyses in Jordan (Wai-Poi and Woodham, 2022) and Uruguay (Llovet Montanes, Tuzman & Rodriguez-Chamussy, 2022), this constitutes a pilot study of the Engendered Commitment to Equity methodology (E-CEQ).1 This pilot aims to apply the recommendations and suggestions of the E-CEQ methodology and leverage available data, to assess the overall impacts of the fiscal system on gender equity in Armenia. Fiscal systems impact women and men differently, as they experience fiscal policies—both taxes and benefits—via different channels, behaviors, and interactions.2 Gender gaps can arise in the outcomes and opportunities enjoyed by men and women across several dimensions, including education, earnings, access to productive assets, political representation, and bargaining power within the household (WDR 2012). Gender gaps affect individual well-being at the micro-level, as well as economic institutions, inclusive growth, and poverty reduction, at the aggregate level. Fiscal systems directly affect individuals’ and households’ income sources, consumer prices, tax liabilities, and social benefits. Hence, fiscal interventions change incentives for people and their households to allocate assets —including time use—and make decisions regarding social reproduction,3 paid and unpaid work, education, savings, investments, and entrepreneurship, among others. The fiscal system, by providing public transfers and public goods, also affects economic opportunities, with potentially more significant impacts—both positive and negative— for vulnerable groups. In turn, fiscal systems may also be affected by gender relations and existing structural gender inequalities in the economy and society. Traditional fiscal incidence analyses (FIAs), nonetheless, are limited in their ability to capture the effects of taxation and transfers on gender equity. The CEQ methodology4 approaches the effects of the fiscal system from the perspective of changes in household income, at different stages of their interaction with taxes and transfer. Welfare and income measures—including household income, headcount poverty, and the Gini Index of inequality—are traditionally aggregated and measured at the household level. In contrast, using these aggregations, the individual welfare of family members within households is difficult to assess in traditional FIAs. Moreover, gender disparities are often hidden and contingent, limiting the scope for conceptual and empirical assessments. While most countries—including Armenia—have implemented 1 This report will reference the “Methodological Paper on Engendered CEQ” as E-CEQ 2022. 2 Gender is “a normative social construct defining and differentiating the roles, rights, entitlements, responsibilities, and s ocial obligations of women and men; it also provides positive and negative incentives for broad-based compliance with those roles, responsibilities, and obligations” (E-CEQ, 2022). 3 Social reproduction consists of activities ranging from fuel and water collection, to cooking and cleaning to caring for children, the elderly and other dependents (E-CEQ Concept Note). 4 The CEQ methodology aims to address four key questions: How much income redistribution and poverty reduction is being accomplished through fiscal policy? How equalizing and pro-poor are specific taxes and government spending? How effective are taxes and government spending in reducing inequality and poverty? What is the impact of fiscal reforms that change the size and/or progressivity of a particular tax or benefit? The methodology has been extended to and adapted for over 85 low- and middle- income countries over the past decade. For methodological details, see Lustig (2018). Page 3 important efforts to eliminate gender biases in law and regulation (explicit biases), other forms of differential treatment for men and women persist implicitly across public and fiscal policy (Stotsky 1996). This paper provides evidence of the effects of the fiscal system on gender gaps in Armenia. The country has made considerable progress over the last decade in reducing gender inequalities. However, certain gender gaps persist, mostly disfavoring women.5 Meanwhile, the fiscal system has undergone relevant changes over the past years, including the restructuring of the personal income tax (PIT) implemented in 2020. Following guidance from the E-CEQ (2022), this report aims to explore the effects of Armenia’s fiscal system on (1) Promoting gender equality, while (2) Reducing poverty, vulnerability, and exposure to livelihood risks. Second, as a preliminary but necessary step to fully understand and contextualize the E- CEQ in Armenia, the paper reviews existing evidence and analyzes new survey data on the individual and household decisions, perceptions and preferences that affect gender disparities in the fiscal system and beyond. The evidence includes selected results from the household survey for “Assessing Gender Disparities in Time Allocation and Household Responsibilities in Armenia” (2022).6 The subsequent sections are structured as follows. Section II provides a brief context of the trends in poverty, inequality, and gender disparities in Armenia, followed by an overview of the country’s fiscal system. Section III introduces the theoretical framework behind the E-CEQ. Section IV briefly describes the sources of data collected for this empirical exercise in Armenia. Next, a typology of households is constructed for the context of Armenia, and differences in poverty incidence are identified as an indication of potential gender disparities. After presenting the national aggregated effects of the fiscal system of Armenia, the fiscal effects on different household types are presented in Section VIII. The analysis of gender gaps at each stage of the fiscal continuum (from pre-fiscal to consumable incomes) aims to identify sources of gender disparities within the fiscal system. The last section discusses the main takeaways from this application of the E-CEQ in Armenia and proposes a future research agenda. II. Poverty and inequality Armenia experienced rapid poverty reductions driven by economic growth over the past decades. Between 2004 and 2018, the national poverty rate decreased by 30 percentage points—from 53.5% to 23.5%.7 Living conditions among the poor have improved, as the depth and severity of poverty have declined (World Bank 2020). Between 2004 to 2008, a third of the population escaped poverty, however, poverty reduction slowed down after the Global Financial Crisis (GFC). The national poverty rate was 26.4% and 27.0% in 2019 and 2020, respectively (Figure 1).8 5 The analysis focuses on gender disparities hindering the wellbeing and economic welfare of Armenian women and girls. In many instances, men—rather than women—may suffer because of gender disparities. The effects of the fiscal system on Armenian men are, nonetheless, outside the primary focus of the analysis. 6 The survey was designed and contracted by the World Bank Poverty and Equity team, with partial financial support from the Hewlett Foundation in May-June 2022. The questionnaire design and survey methodology were tailored to fill in long-standing knowledge gaps about time use and gender disparities in Armenia, as well as to contextualize and match the specific research questions and results from this E-CEQ paper. 7 The lower-middle-income poverty threshold of US$3.2/day decreased from 55.6% in 2001 to 13.0% in 2018 (Povcalnet, 2022). 8 In 2019, Armenia updated the national poverty lines and methodology, and adopted the average poverty line, equivalent to AMD 44,482 per adult equivalent per month in 2020. Poverty and inequality indicators before and after 2019 and not strictly comparable. Page 4 Incomes from labor markets and social protection were fundamental in reducing poverty. Wages, self- employed earnings, and agricultural incomes have been the three main drivers of income growth among Armenian households. Wage income accounted for more than half of the total increase in income, though the benefits were concentrated among the top-60 of the population.9 Pension receipts also had a meaningful contribution to total income growth, reflecting Armenia’s relatively generous pension payments.10 Figure 1. National poverty, 2008-2020 Figure 2. Gini coefficient, 2019-2020 Source: SCRA. Notes: Poverty rates for 2019-2020 use the Source: SCRA. Social Snapshot of Armenia 2021. Based on updated poverty lines and poverty measurement methodology data from the ILCS 2019 and 2020. (ILCS 2019 baseline). Estimates before and after 2019 are not directly comparable. Despite significant progress in poverty reduction, large sectors of the population remain vulnerable to impoverishment in Armenia. After the GFC’s severe and long-lasting impact on the bottom 40% of the population (World Bank 2017), shared prosperity in Armenia remained negative from 2009 to 2018.11 About a third of the population still lives below the poverty line, and a large share of households is at risk of falling into poverty in the face of aggregate or idiosyncratic shocks, such as unemployment and severe illness. Between 2010 and 2016, one in three people who escaped poverty eventually fell back into poverty (Fuchs et al., 2019). III. Gender disparities Armenia has also made considerable progress in reducing gender inequalities over the last decade. In 2016- 2018, the sex ratio at birth—a common indicator of gender disparities and social preference for sons over girls—was 1.11 boys per girl. By comparison, the international average is lower at 1.06. The sex ratio at birth in Armenia, however, has improved significantly from 1.17 in 2001 (Demographic Handbook of 9 Income from self-employment was most relevant for households in Yerevan and less so in other locations (World Bank 2020). 10 Datt and Ravallion (1992) decompositions show that the relative contribution of growth was most significant, while redistribution had a small—but positive—effect on poverty reductions (World Bank staff calculations, based on microdata from ILCS 2015-2018). 11 The shared prosperity premium is calculated as the difference in the income growth of the bottom 40% of the population and the overall income growth of the whole population. Page 5 Armenia 2019).12 Gender equality remains one of the priorities of the Government of Armenia, anchored in the national gender strategy for 2019-2023. Despite recent progress, certain gender gaps persist, disfavoring mostly Armenian women. Gender disparities slow down the country’s inclusive growth and achievement of its full economic potential. Solely the gender gap in economic participation is estimated to reduce Armenia’s GDP by 14% (Cuberes & Tiegnier, 2016). Other sources of gender disparities are often hidden and difficult to quantify. The following subsections present a brief overview of gender disparities across dimensions of economic development and wellbeing. 3.1. Poverty incidence At first sight, gender disparities are not apparent within the poor population. In Armenia 2020, 20% of men and 19% of women lived in poor households.13 The gender gap in multidimensional poverty was 3 percentage points in 2018, with one quarter of women and girls living in multi-dimensionally poor households (SCRA, 2019). Nonetheless, uneven distribution of resources—as well as production decisions—within households can potentially disadvantage women and girls, limiting their current wellbeing and future opportunities. Hence, assessing and improving the impacts of fiscal interventions on gender—and ultimately on poverty and inequality of any vulnerable group—requires opening the household “black-box” and understanding intra-household dynamics and policy impacts. An in-depth analysis of poverty trends by household typologies below seeks to unveil the impact and interaction of poverty with household composition and gender. 3.2. Human Capital Women in Armenia achieve high levels of educational achievement and human capital, compared to men. Access to basic education (grades 1-9) is close to universal in Armenia. Boys and poorer students tend to drop out of upper secondary, more often than girls. Tertiary net enrollment rates are high and favor women (61% vs. 42% among men) (WDI 2020). Beyond school coverage, learning and overall human capital accumulation are higher among women than men. Girls perform better than boys across several indicators of the Human Capital Index (HCI), including years of schooling, Harmonized Test Scores, adult survival rates, and stunting. Overall, an Armenian girl achieves 60% of her potential human capital accumulation by age 18 (HCI=0.60). A boy would only achieve 56% of his potential (HCI=0.56) (HCP, 2018). Although the observed “reversed” gender gaps affecting boys and men lie outside the scope of this report, gender disparities in human capital—regardless of their direction—are a major concern for long-term growth and equity in Armenia. Demographic challenges in Armenia contribute to a gendered population structure and high dependency ratios. Armenia has faced demographic challenges due to low fertility rates and high outmigration (World Bank 2017). The population shrank from the 1990s until 2011. Since 2012, the population growth rate has been positive but close to zero (0.19% in 2020) (WDI, 2022). Women and girls account for 52% of the 12 Concerningly, the ratio (and son preference) increases from 1.04 for the first-born child to 1.40 for the third child (SCRA, Demographic Handbook of Armenia 2019). Interpreting the skewed sex ratio at birth as “missing girls” highlights the loss of girl births due to a “preference for sons” and thus recourse to sex-selective abortion (World Bank 2016). 13 Based on the national average poverty line, equivalent to AMD 44,482 per adult equivalent monthly in 2020. The differences in monetary poverty between women and men under this indicator are not statistically significant at standard confidence levels. Page 6 national population, however, women represent 61% of the elderly, measured by both thresholds of 65 or 80 years and older (ILCS 2020). Moreover, premature mortality is high among young men. In 2019, the average life expectancy at birth was 73.1 and 79.5 years for men and women, respectively (Demographic Yearbook of Armenia 2021). These demographic patterns lead to a high economic dependency ratio on the working-age population, with high dependence of elderly women on future fiscal sustainability. 3.3. Labor markets Despite their high educational attainment, Armenian women have low labor force participation rates. Less than 50% of working-age women participate in the labor force, compared to 71.5% of men (WDI, 2019).14 Moreover, the gender gap in labor force participation has remained stagnant over the past two decades, at over 20 percentage points (Figure 3). While this gender contrast is not uncommon in the ECA region, Armenia performs worse than several peers in integrating an educated female population into labor markets (Figure 4).15 Working women face high unemployment, under-employment, and barriers to accessing high-quality jobs. Unemployment is high among young women aged 15 to 24, at 45%, compared to 33.3% of men in the same age group (ADB, 2020). Women are also more likely (34%) to work part-time positions than men (18%). Analyses of equality of opportunity based on the Human Opportunity Index (HOI) highlight the incidence of gender as an “unfair” determinant of access to high-quality jobs (SCRA and World Bank, 2021). Figure 3. Labor force participation rates, 2012- Figure 4. Gender ratios in tertiary education and 2019 economic participation Source: SCRA. Notes: From 2018 onwards, the methodology of the Source: WDI. Notes: Latest available year for each country; Armenia Labor Force Survey was adapted following the “Resolution concerning (2017). The ratio of female to male labor force participation rate “is statistics of work, employment and labour underutilization” adopted calculated by dividing female labor force participation rate by male by the 19th International Conference of Labour Statisticians. labor force participation rate and multiplying by 100” (WDI 2020). Occupational segregation affects women’s access to high-paying jobs. Women are under-represented in science, technology, engineering, and math (STEM) fields, accounting for only 38% of graduates in these fields (World Bank Gender Portal, 2018). In 2019, one-quarter of women workers were employed in agriculture, while women are virtually excluded from industrial activities, such as construction and mining 14National estimates for 2019. Working-age population defined as people aged 15 and above. 15Armenia has a higher ratio of female to male enrollees in tertiary education (1.3) than Georgia (1.1), Russian Federation (1.2) and Ukraine (1.2). However, Armenian women are less likely to participate in labor markets than women in those countries. Page 7 (Table 1). On the other hand, almost one-fifth of female workers are employed in education (in contrast to less than 4% of male workers) and almost one-tenth in the health sector (vs. only 2% of men). The share of women is also higher than the share of men employed in wholesale and retail. Table 1. Employment of men and women in 2019 and 2020, by activity (% employed population) Men Women Total 2019 2020 2019 2020 2019 2020 Agriculture, forestry and fishing 20% 23% 24% 20% 22% 22% Mining and quarrying 1% 2% 0% 0% 1% 1% Manufacturing 11% 10% 9% 10% 11% 10% Electricity, gas, steam, and air conditioning supply 3% 4% 1% 1% 2% 2% Water supply; sewage, wage management and remediation 0% 0% 0% 0% 0% 0% Construction 16% 13% 0% 0% 9% 7% Wholesale and retail trade; repair of motor vehicles 12% 11% 13% 13% 12% 12% Transportation and storage 8% 7% 2% 1% 5% 5% Accommodation and food services activities 2% 2% 4% 3% 3% 3% Information and communication 3% 3% 3% 2% 3% 3% Financial and insurance activities 1% 2% 2% 3% 1% 2% Real estate activities 0% 0% 0% 0% 0% 0% Professional, scientific, and technical activities 2% 1% 2% 3% 2% 2% Administrative and support service activities 1% 1% 1% 1% 1% 1% Public administration and defense; compulsory social security 10% 10% 6% 7% 8% 9% Education 3% 4% 18% 19% 10% 11% Human health and social work activities 2% 2% 8% 10% 5% 6% Arts, entertainment, and recreation 2% 2% 2% 2% 2% 2% Other service activities 1% 1% 3% 3% 2% 2% Activities of households as employers 0% 0% 1% 1% 0% 0% Activities of extraterritorial organizations 0% 0% 0% 0% 0% 0% All activities 100% 100% 100% 100% 100% 100% Source: SCRA. Notes: Red highlight refers to higher shares, green highlight to lower shares. Figure 5. Net wage earnings by gender, cohort, and educational attainment (Mean monthly net wages, AMD 1,000s) Source: Based on SCRA, Labor Market in Armenia, 2021. Note: pay gap on monthly salaries, unadjusted for hours worked. The unexplained gender wage gap is estimated at 20% in Armenia. Female employees earn only 74% of male net wages, at an average of AMD 103 thousand vs. AMD 139 thousand (Labor Force Survey, 2019). On average, men consistently earn higher wages than women, regardless of age and educational Page 8 attainment (Figure 5). Even tertiary-educated women employees earn only 72% of the average wage earnings of tertiary-educated men. Women’s under-representation in occupations associated with higher wages can partially explain their lower earnings. Rodriguez-Chamussy, et al. (2018) estimated that women face an average unadjusted earnings gap of 33%. Econometric adjustments for differences in education, age, experience, and other observable characteristics reduce the wage gap. Nonetheless, the adjusted wage gap is still high at 20% (Ibid). This gap points to unexplainable factors that can reflect gender discrimination in labor markets in Armenia. The estimated gender gap varies significantly across sectors and locations. Women observe smaller wage disparities in agriculture—although the gender gap is still 15.5% of men’s earnings—and in education, health, and social work (17.8% gender gap). In addition to the traditional exclusion from industrial activities, women also face a gender gap of almost 30% in this sector. This evidence suggests that women are segregated both across and within sectors of employment, facing exclusion from higher-paying sectors and higher-paying positions. Geographically, the wage gap in hourly pay is greatest for women in secondary cities (26%), and lower in Yerevan (14%) and rural areas (17%). Employment in the public sector is associated with a lower gender gap, compared to non-public sector jobs (Ibid). 3.3. Legal protections and other incentives across the lifecycle There are weaknesses in the legal and policy frameworks to protect gender equity in Armenia. According to the World Bank’s Women, Business, and the Law 2021 report,16 Armenia scores poorly in laws protecting women’s decisions to work. women's pay, women's work after having children, and entrepreneurship (WBL 2021). Armenia scores particularly low in workplace indicators and laws affecting women’s work. For example, Armenia lacks a legal mandate for equal remuneration for work of equal value.17 On the other hand, the lack of legal prohibition of gender-based discrimination in access to credit affects Armenia’s performance in Entrepreneurship. Armenia ranks 98 out of 153 countries on the World Economic Forum’s 2020 Global Gender Gap Index, holding the lowest ranking among countries in ECA (24th of 26).18 The country’s standing is explained primarily by low rankings in “health and survival” (148th of 153, driven by the skewed sex ratio at birth), and “political endowments” (114th of 153, affected by women’s low representation in governance and leadership positions). There are de jure provisions to assist women and families during motherhood and re-insertion into labor markets, nonetheless, women's economic opportunities across the lifecycle may be de facto constrained by explicit and implicit incentives. Maternity leave in Armenia guarantees 140 days of paid leave to women, including 70 days during pregnancy and 70 days after childbirth.19 As a positive development, Armenia approved paid paternity leave in 2021. Nonetheless, paid paternity leave is limited to only 5 days for working fathers. This pattern of shorter paid leaves for fathers than mothers is common across countries 16 WBL analyses eight indicators: Mobility, Workplace, Pay, Marriage, Parenthood, Entrepreneurship, Assets, and Pension. 17 To improve on the Workplace Indicator, Armenia may wish to consider enacting legislation protecting women from sexual harassment in employment, Legislation and criminal penalties 18 World Economic Forum, Global Gender Gap Report 2020. http://www3.weforum.org/docs/WEF_GGGR_2020.pdf 19 The maternity leave can increase up to 55 days in case of complicated childbirth. Page 9 (Rubiano-Matulevich, 2020),20 and it can disproportionately allocate childcare responsibilities to women from childbirth. Armenia has government-provided childcare services, paid leave to care for sick relatives, and free and compulsory basic education (USAID, 2019). There are also financial incentives for families with young children. Parents receive non-tax benefits for children under 6 years old (child allowances). Similarly, the legal framework in Armenia guarantees that mothers can return to an equal position after maternity leave. Nonetheless, de facto and non-legal barriers discussed below may limit women’s ability to re-enter labor markets and high-quality jobs. 3.4. Housework responsibilities and time use Household responsibilities—including caring for young children and elders—impose important barriers to women’s economic opportunities. Almost 40% of women out-of-the-labor force (OLF) are classified as housekeepers in the LFS, while less than 2% of men OLF are housekeepers (Figure 6). Armenian women are much more likely than men to miss economic opportunities due to childcare and family responsibilities. Over one-third of OLF women declared that they stayed out of labor markets because of “family circumstances, childcare responsibilities, or care of elderly family members who are sick or suffer disabilities” (Figure 7). In contrast, only 2% of men declared any of those reasons. Although the details behind “family circumstances” may vary, the evidence supports the view that Armenian women take up the largest share of household responsibilities. Figure 6. Population out-of-the-labor force, by category Source: Based on data from SCRA, Labour Market in Armenia, 2021. 20 105 economies (out of 189 with data) guarantee at least one day of paid paternity leave for the birth of a child or reserve a portion of paid parental leave specifically for fathers through fathers' quotas. Armenia lies at the median duration of that leave, at five days. In contrast, 184 countries guarantee at least one day of maternity leave and the median length is 98 days. Page 10 Figure 7. Population out-of-the-labor force, by reasons for not seeking employment Source: Based on data from SCRA, Labour Market in Armenia, 2021. Childbirth and motherhood coincide with the time for the first entry to labor markets or first years of employment for a large share of young Armenian women. Armenian women between the ages of 20 to 24 years have the highest fertility rate (1.20 in 2020, increasing from 1.10 in 2000 and 2010).21 The fertility rate decreases to 1.06 among women ages 25 to 29 years old (in 2020) and drops significantly for older women (Demographic yearbook of Armenia 2021). Regardless of the exact cohort, young women are always more likely than young men to be out of education, employment, and training (NEET). The SCRA estimates that 43% of women and 29% of men between 20 and 24 years old were NEETs (Figure 8). The sharp increase in the share of women NEET for ages 25 to 29 is likely related to motherhood (men in the same age group observe a decrease in the share of NEETs). Figure 8. Youth not in Education and not in Employment (NEET) by age groups, 2019 Source: Based on data from SCRA, Labour Market in Armenia, 2021. 21The SSCRA calculates the age-specific fertility rates “as a ratio of annual number of births to woman at the given age group to the midyear number of women at the same age”. Demographic Yearbook of Armenia 2021. Page 11 Women experience “time poverty” linked to the responsibilities for unpaid domestic tasks and narrow use of childcare services. Women’s time poverty is accentuated when they do enter the labor force, but their domestic workloads do not diminish. A contributing factor to women’s burden is the low enrolment of children in preschool institutions, at only 30% in Armenia and as low as 17% in rural areas (Ibid). Data collected from an independent mixed methods study suggests that most Armenian parents (57% of respondents) use a combination of formal and informal childcare for children 7 years old and younger (World Bank, 2017).22 On the other hand, 41% of respondents use informal childcare arrangements only, and a minority of respondents (3%) rely exclusively on formal care.23 Although there are government- provided childcare services, childcare payments are not tax-deductible and private childcare centers do not receive non-tax benefits (USAID, 2019). Most households rely on women as primary care providers in children’s rearing and development, while fathers contribute to a lesser extent. According to a study by World Vision (2017), close to 80% of households rely on women to provide daily care for children. In only 16% of interviewed households, men and women provide daily childcare equally (Appendix 1). Similarly, at least 60% of households rely on women to stay home to care for sick children. And women are also responsible for helping children do their homework in 50% to 70% of Armenian households. Fathers are more involved in picking up children from school, playing, and taking children to leisure activities, though women are still the primary contributor to these tasks (World Vision Armenia, 2017). Men tend to overestimate their contribution to household responsibilities (relative to women’s perceptions). Special survey on time allocation and gender disparities in Armenia New evidence collected on time allocation and gender disparities in Armenia confirms that women are overburdened by household and family responsibilities, including care for children and elderly. The household phone survey was designed and contracted by the World Bank to assess “Gender Disparities in Time Allocation and Household Responsibilities in Armenia” in 2022. The survey aimed to complement the results from this analysis and address current knowledge gaps on the decisions and perceptions of adults in Armenia, related to: (i) time use, (ii) gender roles and responsibilities within the households, (iii) childcare access, (iv) access to elderly care, and (v) opinions and preferences on gender disparities. Box 1. Introduces the structure of the survey and selected results are summarized below. 22 On the other hand, the share of households relying exclusively on informal care is lower in Armenia, as compared to Serbia (46%), Bosnia and Herzegovina (60%), Macedonia FYI (63%) and Kyrgyz Republic (71%). The study was based on a mixed-methods dataset exploring childcare and eldercare in selected countries of Europe and Central Asia (ECA). 23 An upcoming survey on time use and gender disparities collected in Armenia by the World Bank will attempt to fill in knowledge gaps on costs, quality, social norms, and accessibility issues surrounding child and elderly care in Armenia. Page 12 Box 1. Survey on Gender Disparities in Time Allocation and Household Responsibilities, 2022 The survey for “Assessing Gender Disparities in Time Allocation and Household Responsibilities in Armenia” was conducted in 2022, by the World Bank, in partnership with CRRC Armenia. The aim of this panel study was to assess the impacts of household responsibilities and care needs and access on women’s economic opportunities, time-poverty, and general well-being. Two rounds or contacts were conducted, aiming to maintain a panel structure and optimize the time and quality of each round.1 Questionnaires were designed and tailored based on a review of international best practices and previous work in Armenia. 1 The sample collection was based on Computer-assisted telephone-interview (CATI) technique for data collection. The interviews lasted one month, from 22nd of May to 25th of June, before the end of the school year in Armenia. The survey is representative at the national, regional, urban-rural levels, as well as female-headed households, and households with children. The survey sampling was based on Random Digit Dialing for sampling. The first round achieved a sample size of 3,059 adult individuals with a cellphone in Armenia. All individuals with completed first-round interviews were re-contacted within a period of 7 to 14 days.1 Due to attrition, the second round collected a sample of 1,603 individuals. Two sets of Sample weights were calculated for households and for the adult population in Armenia. The identification of gender of the household head was self-reported by the interviewee. The fieldwork was conducted during the school year, aiming to capture individual and household decisions under “regular” circumstances and tasks related to children. 1 Results from the survey confirm that women in Armenia tend to allocate a much larger share of their time to childcare, elderly care, and domestic tasks. The time allocation to childcare and elderly care (as share of total hours of activity in a regular weekday) is three times as large for women than men (Table 2). The pattern is independent of geographic location, though individuals with high (tertiary) educational achievement spend, on average, more time working and less time in care and household tasks. The gender gap is much more significant in households with children, and women in households with young children allocate the largest share of time to household and care tasks. Men consistently spend more time on leisure activities than women, regardless of the disaggregation. The higher leisure time for men is also reported during the weekends (Appendix 2). Women with higher educational attainment are more likely to spend time at work, nonetheless, they face a double burden of employment and domestic responsibilities. Women with tertiary education spend much longer hours at work than women with no tertiary education (16.5 vs 11.6 percent), while men tend to spend close to 27% of their time in employment, regardless of their schooling. Nonetheless, women with tertiary education (vs. no tertiary education) observe only a small reduction in the share of their time allocated to childcare, elderly care, and household chores. Elderly family members assist with household and childcare responsibilities, but only marginally. The survey did not find strong evidence that the elderly fills in gaps for childcare within the household (or family). Interviewed women and men ages 65 and above allocate less than 5% of their time to childcare (women only allocate 4%, while men allocate 3.5% of their time). Women leaving in households with young children reduce their time allocated to childcare slightly in the presence of elderly members (from 21% to 18%). Page 13 Table 2. Gender disparities in time allocation, 2022 Average share of time allocation, regular weekday Work and Domestic Sleeping Elderly care Childcare commuting Studying Leisure Selfcare Transport Others tasks to work Location Female respondent Armenia 31.3 1.9 9.8 14.4 4.8 17.7 12.6 4.9 2.1 0.4 Male respondent Armenia 32.7 1.3 3.5 26.8 3.1 19.8 3.2 4.7 4.4 0.4 Female respondent Yerevan 29.7 1.4 7.9 16.4 6.6 18.3 11.1 5.3 3.1 0.2 Male respondent Yerevan 31.9 0.7 2.9 27.6 4.8 19.7 3.1 4.6 4.4 0.2 Female respondent Other Urban 31.4 1.8 10.6 13.3 4.5 18.6 13.1 4.8 1.6 0.4 Male respondent Other Urban 32.6 1.5 3.4 26.3 3.1 20.3 3.0 5.0 4.5 0.3 Female respondent Rural 32.5 2.4 10.8 13.7 3.7 16.6 13.4 4.8 1.7 0.5 Male respondent Rural 33.4 1.4 4.1 26.7 2.1 19.6 3.3 4.5 4.3 0.6 Individual characteristics Female respondent No Tertiary Education 32.9 2.1 10.9 11.6 3.7 17.7 14.1 4.7 1.9 0.5 Male respondent No Tertiary Education 33.3 1.4 3.7 26.8 2.2 19.8 3.3 4.8 4.3 0.3 Female respondent Tertiary Education 30.1 1.9 9.0 16.5 5.8 17.7 11.4 5.1 2.3 0.3 Male respondent Tertiary Education 32.1 1.1 3.4 26.8 4.1 19.9 3.0 4.5 4.6 0.5 Female respondent 18-24 28.7 0.9 8.6 9.8 14.2 17.6 9.2 6.9 4.0 0.0 Male respondent 18-24 32.3 0.6 1.8 24.2 7.9 19.2 2.9 6.6 4.4 0.2 Female respondent 25-34 27.9 1.9 17.8 13.0 4.2 14.9 13.2 5.0 2.0 0.2 Male respondent 25-34 29.8 1.3 3.6 31.1 3.3 17.1 3.1 4.9 5.7 0.2 Female respondent 35-44 29.2 2.1 13.7 18.1 3.2 12.9 13.5 4.6 2.3 0.4 Male respondent 35-44 31.6 2.1 4.6 29.2 2.2 17.3 2.9 4.9 4.8 0.4 Female respondent 45-64 32.1 2.3 5.5 18.4 3.5 18.2 12.9 4.6 1.8 0.6 Male respondent 45-64 34.4 1.3 3.4 26.2 2.1 21.0 3.2 3.9 3.9 0.6 Female respondent 65+ 39.0 1.8 4.2 5.7 4.0 26.7 12.6 4.5 1.1 0.4 Male respondent 65+ 37.5 0.3 3.5 17.5 2.5 28.1 4.0 3.7 2.2 0.7 Household composition Female respondent Household no children 34.5 2.0 0.7 15.9 5.5 21.7 11.8 5.2 2.3 0.5 Male respondent Household no children 33.6 1.3 0.7 24.9 3.9 22.1 4.1 4.8 4.2 0.4 Female respondent Household with children 29.1 1.9 16.1 13.3 4.4 14.9 13.2 4.8 2.0 0.3 Male respondent Household with children 32.0 1.2 5.7 28.3 2.6 18.1 2.5 4.6 4.6 0.4 Female respondent Household no elderly (65+) 29.9 1.5 11.2 15.5 5.0 16.2 12.9 5.0 2.4 0.4 Male respondent Household no elderly (65+) 31.9 0.9 3.7 28.1 3.4 18.8 3.1 4.9 4.8 0.4 Female respondent Household with elderly (65+) 33.8 2.7 7.5 12.3 4.5 20.3 12.2 4.7 1.6 0.4 Male respondent Household with elderly (65+) 34.5 1.9 3.3 24.2 2.5 22.0 3.3 4.2 3.6 0.5 Female respondent Household no young children (0-6) 32.9 2.1 4.9 16.0 5.5 18.9 11.9 5.1 2.3 0.4 Male respondent Household no young children (0-6) 33.5 1.2 2.2 25.8 3.5 20.8 3.5 4.8 4.2 0.5 Female respondent Household with young children (0-6) 28.0 1.6 20.7 10.8 3.3 14.9 14.2 4.5 1.6 0.2 Male respondent Household with young children (0-6) 31.1 1.3 6.4 29.1 2.3 17.9 2.5 4.4 4.8 0.3 Female respondent No young children & no elders 31.1 1.7 5.3 18.3 5.8 17.2 11.9 5.3 2.8 0.5 Male respondent No young children & no elders 32.7 0.9 2.2 27.0 3.9 19.6 3.4 5.2 4.7 0.5 Female respondent No young children, elders 35.5 2.7 4.3 12.5 5.0 21.5 11.9 4.8 1.6 0.3 Male respondent No young children, elders 35.1 1.9 2.0 23.4 2.9 23.0 3.7 4.2 3.2 0.5 Female respondent Young children, no elders 27.9 1.2 21.6 10.6 3.5 14.3 14.6 4.6 1.6 0.1 Male respondent Young children, no elders 30.5 1.0 6.4 30.2 2.6 17.3 2.5 4.3 4.9 0.2 Female respondent Young children & elders 28.2 2.7 18.3 11.5 2.9 16.5 13.2 4.3 1.7 0.6 Male respondent Young children & elders 32.7 2.1 6.5 26.1 1.6 19.4 2.4 4.4 4.6 0.4 Female respondent No children & no elders 32.4 1.7 0.8 19.7 6.1 19.0 11.6 5.4 2.8 0.5 Male respondent No children & no elders 32.7 1.0 0.6 26.7 4.5 20.6 3.8 5.1 4.7 0.4 Female respondent No children, elders 37.4 2.5 0.5 11.0 4.7 25.2 12.1 4.8 1.5 0.4 Male respondent No children, elders 35.5 2.1 0.9 21.5 2.8 24.9 4.6 4.2 3.2 0.3 Female respondent Children, no elders 28.5 1.5 17.1 13.2 4.4 14.5 13.6 4.8 2.1 0.2 Male respondent Children, no elders 31.4 0.9 5.8 29.1 2.7 17.5 2.6 4.8 4.9 0.3 Female respondent Children & elders 30.4 2.9 14.0 13.5 4.3 15.8 12.3 4.6 1.8 0.4 Male respondent Children & elders 33.6 1.8 5.3 26.5 2.3 19.4 2.2 4.3 3.9 0.6 Proxy welfare status of the household Female respondent Household no extreme poverty 30.8 1.9 10.2 14.7 5.2 17.2 12.4 5.0 2.3 0.3 Male respondent Household no extreme poverty 32.6 1.1 3.4 27.6 3.2 19.4 3.1 4.6 4.5 0.4 Female respondent Household in extreme poverty 34.0 2.1 8.1 12.5 3.1 20.0 13.9 4.6 1.3 0.5 Male respondent Household in extreme poverty 33.5 2.3 4.7 21.3 2.4 22.5 3.9 5.0 3.6 0.8 Source: Based on data from the survey for “Assessing Gender Disparities in Time Allocation and Household Responsibilities in Armenia ”, 2022. Notes: Time allocation shares are calculated as percentage of the total time (in hours) reported by each respondent in the survey. This adjustment corrects for respondents reporting less or more than 24 hours of activity in one day, a common issue faced by time-use surveys. Extreme poverty status is proxied by the respondent’s self-assessment that her/his household cannot afford sufficient food. Individual weights applied to represent the national adult population of Armenia. Page 15 3.5. Social norms and other vulnerabilities Social norms may also play a role in defining women’s roles in economic participation and household responsibilities. Around 60% of Armenians still consider that men should be the “breadwinner” of the family (Caucasus Barometer, 2019). Despite improvements in the sex ratio at birth, some evidence points to a persistent preference for sons. Having a boy is still preferred by one-third of adult respondents, over having a girl or no gender preference (Ibid). The son preference is highest in rural areas and among male respondents. Some evidence also highlights the low agency of Armenian women in taking decisions over economic opportunities. For example, 52% of interviewed men and only 38% of interviewed women believed that the decision for women to work was shared with their partners (World Vision Armenia, 2017). Finally, the Covid-19 pandemic may have widened gender disparities in endowments, economic opportunities, and women’s agency, with long- to medium-term impacts. In addition to the short-term shocks to incomes and labor conditions, the pandemic and lockdown measures are expected to exacerbate women’s vulnerabilities, including, female segregation in informal employment and unpaid family work; risks of gender-based violence (GBV), and increased responsibilities to care for children staying at home or sick family members (World Bank, 2020). IV. Overview of the fiscal system24 Armenia’s tax system can be categorized as mixed, as it combines individual-based taxation with elements of a family-based system. Personal income tax is collected at the individual level.25 The tax code establishes that all resident individuals receiving income—from sources other than Armenian legal entities recognized as tax agents—must pay PIT through self-declaration and submit an Annual Income Tax Declaration by 20th April of the following calendar year. There are no reporting obligations for income received from tax agents.26 Hence, the individual tax obligations in Armenia avoid disincentives of household-based taxation discouraging the labor supply of lower earners, who tend to be women. Similarly, the fiscal system does not impose explicit gender differences in taxes regimes, there are no specific tax deductions or tax credits that apply exclusively to men or women in Armenia (USAID, 2019). On the other hand, in practice, several social benefits are allocated at the family level, introducing elements of a mixed system. Child allowances, and the Family Benefits Program are, in practice, household-level social assistance. The main elements of the fiscal system considered in the analysis are outlined below. 4.1. Revenues Armenia observed significant changes in its tax structure over the past years. Income and trade taxes have increased their contribution to total taxes since 2010. Reliance on indirect taxes, in contrast, has been less significant than in peer countries of Europe and Central Asia (ECA). Personal income taxes (PIT) and 24 For more details, please see World Bank. 2021. Armenia Fiscal Note. 25 Under individual taxation, each person’s tax is determined based on their own income, regardless of marital status. Household- level taxation allows to aggregate household incomes under a combined scheme. These systems tend to create incentives to file jointly and raise the marginal tax rate for secondary earners. Hence, household-based tax systems implicitly bias household decisions against women’s labor supply, who tend to be the lower earner within the household. Delgado Coelho et al. (2022). 26 https://home.kpmg/xx/en/home/insights/2021/07/armenia-thinking-beyond-borders.html corporate income taxes (CIT) account for approximately 40% of total tax revenues. Other direct taxes include social payments, property, and land taxes (World Bank, 2021). A major reform to the PIT introduced a flat rate of 23% in 2020. The income tax rate was formerly structured into three progressive brackets—with 24.4%, 26%, and 36% rates, respectively—and became a uniform tax rate of 23% beginning in January 2020. The rate is to be reduced by 1 percentage point annually until reaching 20% in 2023. According to ex-ante simulations, PIT reform and the gradual decrease in the flat rate from 23% to 20% will bring small income windfalls for households by 2023; albeit these would be higher for better-off households and negligible for the bottom-10 (Carrasco et al. 2021). The 2014 national pension reform, another major fiscal policy change, introduced mandatory social security contributions (SSCs) called the “social targeted payment”. The reform aims to transition towards a fully funded pension system. SSCs are paid on top of the unified income tax (PIT) by all formal employees born after January 1974. The social targeted payment by employees will increase from 2.5% in 2020 to 5% of their incomes by 2023 (World Bank, 2021). The contribution is matched (one-to-one up to a defined ceiling) by budget sources, resulting in a total of 10% of income channeled to individual pension accounts.27 Indirect taxes include Value-added-taxes (VAT), excises, and trade taxes. VAT has been the largest tax instrument in Armenia, accounting for 8% of GDP (2014-2018 average). Nonetheless, the decline in VAT revenues in recent years has been partly explained by its falling tax efficiency (World Bank, 2021). The standard VAT rate on domestic sales of goods and services and imported goods is 20%, while exported goods and related services benefit from a zero rate. Armenia holds a generous list of VAT exemptions, which includes most financial, education, and health services. A recent assessment of the fiscal system found that VAT exemptions and high thresholds cause economic distortions and incentivize tax evasion (World Bank, 2021). It is estimated that VAT exemptions resulted in foregone revenue of 5.7% GDP (or 80% of tax expenditures) in 2021 (Ibid). Revenue from other indirect taxes has increased over time, aided by an increase in statutory excise tax rates and trade volumes. Excises are applied to alcoholic beverages, tobacco products, and petroleum products. Their share in total tax collection increased from 5% (in the early 2010s) to 8.4% (2018-2019). A new excise fixed rate—which also eliminated the ad valorem component—was implemented in 2020. Alcohol and tobacco excise rates will increase annually by 30% and 15%, respectively, until 2023. Finally, trade taxes (import and export duties) are the third largest component of indirect taxation, averaging 1.2% of GDP (2014-2018). Their increase since 2015 has been linked to increased imports after Armenia acceded to the Eurasian Economic Union (EEU). Armenia’s revenue performance has progressed significantly, though important improvements are pending (World Bank, 2021). Armenia has maintained a trend of increasing tax revenues, despite declines during periods of crisis. In 2019, the tax-to-GDP ratio stood at 22.5% (WDI 2022). By 2021, the main areas of improvement in revenue performance included (1) Enhancing the productivity of PIT, CIT, and especially VAT, by closing policy gaps—e.g. high thresholds and generous exemptions—and administrative gaps; (2) 27The 2014-reform imposed employee contributions of 5%. Nonetheless, strong resistance delayed the reform implementation until July 2018. Moreover, the new government—after the 2018 Velvet Revolution—reduced employees’ contribution to 2.5% and increased the government’s matching to 7.5% for 2020 (World Bank, 2021). Page 17 Improving fairness by raising the tax burden on high net-worth individuals; and (3) Expanding the use of excise taxes, such as those on tobacco, sugar-sweetened beverages, and green taxes (Ibid). 4.2. Public expenditures Before the Covid-19 outbreak, Armenia’s State Budget expenditures accounted for 27.2% of GDP per annum from 2015-to 2020. Approximately one-third of the national budget expenditures (7% of GDP, in 2018) are allocated to social protection programs. Government spending on education and health has remained significantly low by international standards, at 8.7% of government expenditures (or 2.7% of GDP, in 2020) and 5.3% of government spending (or 1.2% of GDP, in 2018), respectively (data from WDI, 2022). In response to the Covid-19 pandemic, government expenditure played a critical role to support households, businesses, and financial institutions. The policy response to Covid-19 included 25 targeted packages, equivalent to 1.6% of GDP in direct spending, 1.1% of GDP in tax deferrals, and 2.8% of GDP in support through the financial sector. Additionally, the Government of Armenia (GOA) also increased health spending by AMD 36.4 bn to mitigate the public health emergency. Nonetheless, the value of fiscal support measures in response to Covid-19 was modest compared to peer countries. Increased emergency spending coupled with declining revenues, pushed public debt to 67% of GDP in 2020 (MPO, October 2021), leading to plans for rapid fiscal consolidation in the coming years. 4.3. Social protection and assistance Armenia has a complex system of over a hundred contributory and non-contributory social protection programs (MOL, World Bank, and Unicef, 2020). Pensions take up two-thirds of total social spending, while one-third is allocated to different social allowances. Around 3% of the state budget was allocated to cover the state’s obligatory contribution of 75% to SSCs in 2019. Reforms to pensions and social assistance since 2000 have shifted benefits towards families with many children (Attah and Sammon, 2020). Family and child benefits represented about 50% of total social allowances in 2011, although their share has decreased to about 30% in 2019, mostly due to the introduction of new benefits. There are de facto coverage gaps and exclusion errors in social protection, linked to (a) eligibility dependent on formal employment, (b) lacking coverage to address lifecycle events, and (c) financial constraints. The social protection legal framework recognizes gender equality, in practice, nonetheless, women face exclusions due to employment status and life cycle events. The principles of “gender equality, non - discrimination, and special needs are properly addressed in SP legislation and, in general, are maintained during the practical implementation” (MOL, World Bank & Unicef, 2020). Nonetheless, the system offers “no guaranteed social protection for all throughout the life cycle, including children, persons with disabilities or informal workers”. Pensions—including old-age, disability, and survivor pensions—and maternity benefits are all linked to formal employment (Attah and Sammon, 2020). Hence, the extreme poor and informal working poor tend to be excluded from social protection (MOL, World Bank, and UNICEF, 2020). Finally, available financial resources are insufficient to implement “adequate” social protection policies to address the needs of the population (MoL, World Bank & Unicef, 2020). 28 The size of minimum benefits--meant to cover the minimum food basket—remains very low and lacks an indexation rule (Ibid). 28 Overall, social protection expenditures in Armenia remain low, at half of the OECD average, as a share of GDP. Page 18 The overall impact of social protection programs on the well-being and economic incentives of women and other vulnerable populations requires further analysis. The assessment by the Ministry of Labor, World Bank & Unicef (2020) determined that a separate analysis is needed to fully understand “the extent to which the SP system is disability and gender-sensitive, responsive, and transformative” (MOL, World Bank & Unicef, 2020). It is also argued that the current social protection system creates incentives that may expand gender disparities. For example, the design of social assistance pays little attention to behavioral incentives for workers to accept formal employment and for employers to register workers (Ibid). Low salaries and fear of losing eligibility for social assistance benefits may disincentivize beneficiaries of social assistance benefits from accessing formal employment (Morgandi, Posadas, and Damerau 2014). 4.4. Impacts on poverty (CEQ 2017) Fiscal interventions have traditionally played a crucial role in reducing poverty and inequality in Armenia. An assessment based on the CEQ Methodology by Younger and Khachatryan (2017) found that Armenia’s fiscal system improves equity and reduces poverty.29 Fiscal interventions led to a decrease in the Gini coefficient by 0.11 points–from 0.469 to 0.357. Measured at the international threshold of $2.5/day (2005 PPP), the poverty headcount declined from 39.3% to 30.9% when moving from market income to consumable income. However, if measured at the threshold of $4.0/day (2005 PPP), the poverty headcount slightly increased from 58.3% to 60.2%. The bottom 60% have benefited from fiscal interventions more than the top 40%. The first quintile of the population captures significant income gains from the fiscal system, their average final income (including in-kind benefits) is 247% higher than their market income. However, as income increases, wealthier Armenian households become net payers. Overall, the bottom 60% record an increase in income, although it is already low (only 2%) for the third quintile. On the other hand, the final income after fiscal interventions of the top two quintiles is lower by 15% and 23%, respectively, relative to their market incomes. In-kind transfers are also well-targeted to the poor. Most in-kind education benefits, especially primary and middle-school education, and vocational training are concentrated among poorer households. On the other hand, pre-school, secondary schooling, and the value of in-kind health benefits are spread evenly across the income distribution. V. Standard CEQ results Fiscal policy and pensions contributed to reducing poverty at the national level in Armenia. Poverty estimations from the standardized CEQ methodology are presented in Figure 9.30 Like previous applications of the CEQ in Armenia, most of the poverty reduction effect in 2020 took place because of contributory old-age pensions (Carrasco Nunez et al. 2020).31 Armenia’s fiscal system reduced national poverty by 9.4 29 The authors assumed that pensions in Armenia are public transfers, as opposed to differed private incomes. Their analysis used data from 2013, before the introduction of the latest pension reform. 30 The poverty headcount at disposable income (25.1%) is close but slightly above the official poverty headcount in 2020 (27.0%). 31 Compared to World Bank staff estimates for 2017, the poverty-reducing effect in 2020 seems much stronger (for example, in 2017, the effect was calculated at 4.3 percentage points. Nonetheless, year-to-year comparisons are not possible due to changes in the national poverty lines, national consumption aggregate, ILCS collection methodology, and the incidence of Covid-19 and associated fiscal responses by the Government of Armenia. Page 19 percentage points, from 37.4% to 28% in 2020, when going from market income (pre-fiscal income, excluding contributory pensions) to consumable income (post-fiscal income, after all direct and indirect taxes and transfers, but excluding in-kind transfers) (Figure 9). On the other hand, if contributory pensions were considered entirely as an individual’s differed income, the estimated taxes and transfers of the fiscal system would lead to a higher headcount poverty rate, from 22.8% (at market income plus pensions) to 28.0% (at consumable income). The fiscal system also had a significant impact on reducing inequality in 2020. The Gini index decreases by 13.2 Gini points (from 35.8 to 22.6) when going from gross market (pre-fiscal income) to final income (post- fiscal income, accounting for in-kind transfers in education and health) (Figure 10). Accounting for contributory pensions led to the largest reduction in inequality (by 7.8 Gini points). Direct taxes and direct transfers were progressive as of 2020. The most important indirect tax—the VAT—was mildly progressive according to the Kakwani Index. Excise taxes on tobacco and alcohol and petroleum were highly progressive. In-kind education benefits were progressive. Nonetheless, in-kind health expenses were highly regressive. Figure 9. Changes in national poverty Figure 10. Changes in inequality (Headcount poverty rate for average poverty line) (Gini index) Source: Authors’ estimation based on the ILCS 2020. Note: Source: Authors’ estimation based on the ILCS 2020. Note: Based on the national average poverty line. Based household income aggregates per adult equivalent. VI. The Engendered (E-CEQ) methodology 5.1 Theoretical framework Standard applications of fiscal incidence analysis (FIA)–including the CEQ methodology (Lusting, 2018)— assume a uniform distribution of burdens and benefits from the fiscal system across all members of the household. This practical assumption—partially driven by the lack of widespread data at the individual level in national household surveys—nonetheless, neglects “different economic, financial, and social needs of male and female subpopulations and households of different demographic compositions” (E-CEQ 2022). Fiscal policies can have heterogeneous effects on individuals’ welfare, decisions, and interactions within households. Fiscal interventions affect gender relations and disparities, for example, by modifying incentives for labor force participation and other household production functions; influencing optimal time Page 20 allocations for work, childcare, and elderly care; changing relative prices of consumer goods and services; providing household businesses with access to finance and other assets; impacting the costs and returns to investing in asset accumulation (including in human capital); affecting the bargaining power and relationships within the household, and influencing reproductive decisions. In turn, many fiscal policy outcomes (for example, effective revenue collection) are affected through a feedback loop, by gender disparities and relations within households and societies. The E-CEQ adapts the traditional CEQ approach to assess the interaction of the fiscal system with gender disparities and intra-household dynamics. Figure 11 showcases how the different income stages analyzed as part of the CEQ methodologies can affect—and be simultaneously affected—by gender disparities and intra-household dynamics. Hence, the Engendered CEQ aims to explore these different “access points” to identify and quantify the fiscal benefits and burdens experienced by women and men, along their life cycle and contingent on their individual and collective circumstances and decisions. Figure 11. Interaction of the fiscal system with gender and intrahousehold dynamics Source: Authors, adapting from Engendered CEQ Methodological Note (2021) and Lustig (2018). Note: Rent—generally included in market income—is not imputed in Armenia’s consumption aggregate. Page 21 5.2. Empirical strategy to identify gender disparities Identifying the different experiences and interactions of men and women with the fiscal system is, nonetheless, often constrained by data limitations. Household members tend to pool incomes, expenditures, and assets together, whether those are contributed by one individual (e.g., one member’s monthly salary or her PIT payments) or several household members (e.g., agricultural sales produced collectively, or utility bills). Moreover, most microdata on incomes and expenditures from Household Budget Surveys (HBS)—including the Integrated Living Conditions Survey (ILCS) in Armenia—is collected at the household level. Consequently, for both conceptual and data limitations, the first problem in integrating a gender angle to FIA is the identification of women’s and men’s interactions within the household and with the fiscal system. Following the E-CEQ identification strategy, the application in Armenia will adapt “the CEQ Assessment FIA toolkit to estimate fiscal policy impacts on households according to their sex-disaggregated composition or sex-disaggregated- age- and employment-based characteristics” (E-CEQ, 2022). The E-CEQ, hence, proposes to analyze several “access points” that define the interactions of households and the fiscal system and can provide evidence of the fiscal incidence on gender equity. Households interact with the fiscal system when they consume goods and services; make reproductive decisions; participate in labor markets; engage in production for market sales or own consumption; allocate time between labor, study, and other activities (including household tasks and care); investment in human capital and other assets, among others. All those decisions are affected by taxes and transfers of the fiscal system, as well as implicit and explicit biases and incentives related to gender. Hence, the E-CEQ recommends classifying households into typologies to highlight gender dimensions and potential gender disparities. A basic framework to construct typologies under the E-CEQ is summarized in Table 3. Table 3. Classification of households into categories based on gender relations Demographic Headship Income contributions Mixed typology composition Description Most common Exploits variation in Exploits variation in intra- Combining more than one and classification. consumption patterns household bargaining typology to further identification Generally, self- and gender-specific power. Exploits variation in disaggregate household reported or needs. gender-specific market characteristics. enumerator opportunities. assigned in the survey. Pros and Pros: Available in Pros: Easily available from Pros: May disclose power Pros: A more detailed cons most surveys the household roster. dynamics within the analysis accounting for Cons: Definitional Accounts for eligibility to household that affect interactions in gender and issues. Does not certain fiscal policies. decisions. household vulnerabilities. account for intra- Cons: It may still be Cons: The definition of Cons: Cells may be empty household insufficient to capture income thresholds may not or the sample size dynamics and intra-household dynamics be straightforward. insufficient for statistical allocations. and allocations. inference. Basic gender • Female head • Female majority • Earners are a female • Earners are female typologies • Male head • Male majority majority majority + Dependent • Earners are a male majority Page 22 Expansions • Pay attention to • Consider only adults • Consider labor earnings vs. children + Dependent & present vs. • Separate dependent other incomes elderly refinements migrant household elderly (non-WAP) • Adjust by share of total • Earners are male (Armenia) head • Separate young (below household incomes majority + Dependent the age of 6) vs. other children + Dependent children elderly Source: Adapted and expanded from Grown and Valodia (2010). VII. Data sources This Engendered CEQ pilot in Armenia leverages a variety of macroeconomic, microeconomic data, and administrative sources. The main E-CEQ model is applied to household and individual-level microdata from the ILCS 2020. The ILCS is the national household budget survey, collected since 2001 by the Statistical Committee of the Republic of Armenia (SCRA), with national, regional, and urban-rural representativeness. Administrative data were mostly retrieved from budgetary reports by the Ministry of Finance. The analysis also borrowed extensively from analytical inputs and research performed in previous CEQ applications for Armenia by Younger and Khachatryan (2017) and Carrasco Nunez et al. (2021). VIII. Engendered household typologies 7.1. Proposed typologies A preliminary demographic analysis of the ILCS highlights that the largest share of Armenian households (28%) is composed of a couple with children, with other adults also living in the household (Figure 12).32 This household type is also overrepresented among the poor population, accounting for almost 40% of poverty in the country (Figure 13). Most single-headed households—where the self-identified household head is not living with a spouse—are sustained by females. These single female-headed households account for almost one-fifth of all households (9% of households have a single female without minors, but other adults; another 9% are households with a female head, children, and other adults). Figure 12. Share of the population Figure 13. Share of the poor Source: Authors’ calculation, based on ILCS 2020. Source: Authors’ calculation, based on ILCS 2020. Note: Based on the national average poverty line. 32This preliminary typology while helpful, resulted in 24 potential categories that could not be individually analyzed under the E- CEQ framework, due to sample size issues. Page 23 The preliminary analysis also shows that households with children are overrepresented among the poor. A sensitivity analysis was conducted to understand the correlates of poverty and the presence of children and elderly, of different age groups, in the household. Graphs illustrating the analysis are presented in the appendix. The presence of elders in the household is not correlated with a significant increase in the likelihood of poverty either. On the other hand, 58% of Armenian households have children (below 18 years old). However, these households with children are overrepresented among the poor, accounting for 76% of poor households. The presence of younger children (younger than 12 years old) is not correlated with higher poverty incidence than the presence of any child (younger than 18). Four main household typologies are proposed and tested for the E-CEQ analysis.33 Based on the CEQ-2020 methodology, typologies recommended by Grown and Valodia (2010), and data availability for Armenia (ILCS 2020), the four household typologies are identified (Table 4). It is worth noting that these definitions aim to identify heterogeneous gender characteristics within households while maintaining a sufficient sample size for valid statistical inference. Unfortunately, other combinations and typologies that are interesting did not yield enough observations per cell or category. 34 Each typology—from T1 through T4—integrates a higher degree of granularity to identify potential gender disparities in households’ interactions with the fiscal system. T1 refers to demographic composition only and categorizes households according to the “majority” gender of adults. T2 restricts the gender majority to adults with labor earnings, thereby introducing the dimension of women’s economic opportunities and intra-household roles. T3 goes a step forward and assumes that not all household earners have equal contribution or bargaining power within households. Female-sustained households are defined as those with female earnings accounting for at least 60% of household labor incomes. Male-sustained households are those where male adults account for at least 60% of labor incomes. Additionally, T4 implicitly identifies the presence of children and the elderly to capture the household’s care needs and domestic responsibilities, and their interaction with incomes and labor decisions. T4 is used as the preferred typology in the remaining of this report, though T3 will be alternatively used in some analyses due to sampling size limitations. Table 4. Household types based on the ILCS 2020 Typology Name Household types Definition Traditional definition (not recommended) (T0) Gender of A. Female head • Identification of household head as (self- household head B. Male head reported) in household interview roster C. No household head is present Demographics only (as a proxy for needs and opportunities) (T1) Gender A. Majority of female adults • Calculated from the gender of household majority B. Majority of male adults members (18+ years) C. No majority of one gender Demographics + Income (proxy for intra-household roles & bargaining power) 33 The rest of the report refers to “typology” as the criteria (or set of rules) to identify households based on variables related to female or male characteristics. Typologies are referred to as T1, T2, T3, and T4. In turn, a household “type” is a specific case within the typology. For example, households with “Majority of female adults” is a type. 34 For example, a typology disaggregating the gender of breadwinners, as well as three age groups of children and two groups of elderly members was tested. Nonetheless, the combinations yielded 112 theoretical types, with many empty data cells in the ILCS. Page 24 A. (Labor) earners are a female majority • Calculated from the gender of household (T2) Gender of B. (Labor) Earners are a male majority members who report positive labor earnings. household C. Equal number of male and female • Labor earnings include salaries and self- earners in labor labor earners employed earnings; exclude incomes from markets D. No (labor) earners in the household agricultural sales and own consumption. A. Larger share of female earnings. • Calculated over total labor earnings reported (T3) Gender (>60% of household labor earnings) by gender of household members. sustaining B. Larger share of male earnings. (>60% • Labor earnings include salaries and self- household of household earnings) employed earnings; exclude incomes from earnings C. No earners in the household agricultural sales and own consumption. Demographics + Income (proxy for intra-household roles & bargaining power) + Presence of children & elderly (proxy for care needs and domestic responsibilities) • Typology (3) of gender sustaining • Calculated from gender and age of all (T4) Gender household labor earnings household members (self-reported) in sustaining • Combined with the presence of retirees household interview roster; labor earnings household who may contribute to domestic tasks reported by gender of household members. earnings + and care presence of • Combined with the presence of children & dependent household members: elderly o Children <6 years old o Elderly 80+ years old Source: Adapted and expanded from Grown and Valodia (2010). Notes: Income sources available in the ILCS that can be attributed at the individual level include salaries, self-employment earnings, remittances, pension incomes, and incomes from social assistance programs (including the Family Benefit program). Other income sources are only reported at the household level, and hence, they are excluded from the gender typologies above. These include incomes from agricultural sales and incomes from self- consumption. Household total earnings are calculated as the sum of all individual-level incomes. 7.2. Basic poverty and sociodemographic statistics by typology Table 5 summarizes the distribution of household typologies across income groups and urban-rural locations. The purely demographic types under T1 of the majority of adults are evenly distributed across income groups and locations. 16% of Armenians live in households with only female earners, while one- third in households with male-only earners (Panel A). In total, one-fifth of households rely on female workers (contributing >60% to labor incomes). However, households sustained by females and also including elderly members and dependents (under T4) are the smallest household type, accounting for 1.2% of the population. Male-sustained households are not more likely than female-sustained households to belong to higher-income groups in the population. 61% (65%) of male-sustained (female-sustained) households belong to the top-60 national incomes. However, due to their larger number, households sustained by male earners represent 60% of the national distribution of top-60 incomes (Panel B). Page 25 Table 5. Socioeconomic distribution of household types (Cells represent % of the total population) Panel A. Share of household type in income groups and location Household typology All Q1 Q2 Q3 Q4 Q5 Bottom-40 Top-60 Urban Rural Gender of household adult members (T1) Members are female majority 100% 20.7% 18.8% 20.5% 18.9% 21.1% 39% 61% 63% 37% Members are male majority 100% 17.3% 21.3% 18.7% 21.3% 21.4% 39% 61% 51% 49% Equal number of females and males 100% 20.8% 20.2% 20.2% 20.1% 18.7% 41% 59% 60% 40% Gender of household earners (T2) Earners are female only 100% 20.9% 14.9% 23.1% 18.2% 23.0% 36% 64% 66% 34% Earners are male only 100% 19.3% 22.5% 20.9% 18.8% 18.6% 42% 58% 54% 46% Earners are both female and male 100% 16.1% 19.3% 20.2% 23.6% 20.7% 35% 65% 65% 35% No labor earnings 100% 26.7% 21.6% 16.2% 17.2% 18.3% 48% 52% 50% 50% Gender sustaining household earnings (T3) Female (>60% earnings) 100% 20.0% 15.3% 23.1% 18.9% 22.6% 35% 65% 67% 33% Male (>60% earnings) 100% 17.6% 21.0% 20.3% 21.4% 19.7% 39% 61% 59% 41% No labor earnings 100% 26.7% 21.6% 16.2% 17.2% 18.3% 48% 52% 50% 50% Gender sustaining household earnings + presence of children & elderly (T4) Female (>60% earnings) 100% 22.3% 14.8% 20.6% 26.2% 16.1% 37% 63% 54% 46% Dependents + Male (>60% earnings) 100% 27.3% 24.2% 15.5% 16.3% 16.8% 51% 49% 56% 44% Elderly No labor earnings 100% 39.7% 21.3% 14.6% 12.9% 11.5% 61% 39% 46% 54% Female (>60% earnings) 100% 24.7% 21.2% 17.3% 19.8% 16.9% 46% 54% 57% 43% Dependents, no Male (>60% earnings) 100% 20.3% 24.8% 20.6% 18.9% 15.4% 45% 55% 54% 46% Elderly No labor earnings 100% 35.1% 23.5% 15.6% 12.6% 13.2% 59% 41% 43% 57% Female (>60% earnings) 100% 22.4% 12.2% 30.7% 20.0% 14.6% 35% 65% 71% 29% No Dependents, Male (>60% earnings) 100% 20.2% 23.2% 18.9% 16.2% 21.6% 43% 57% 64% 36% Elderly No labor earnings 100% 17.4% 21.2% 18.2% 22.0% 21.2% 39% 61% 66% 34% Female (>60% earnings) 100% 17.0% 14.5% 22.4% 17.3% 28.9% 31% 69% 70% 30% No Dependents, Male (>60% earnings) 100% 13.9% 17.3% 21.1% 25.1% 22.6% 31% 69% 62% 38% no Elderly No labor earnings 100% 26.9% 20.5% 14.9% 16.8% 20.9% 47% 53% 39% 61% Panel B. Distribution of household types across income and location Household typology All Q1 Q2 Q3 Q4 Q5 Bottom-40 Top-60 Urban Rural Gender of household adult members (T1) Members are female majority 30.1% 31.0% 28.2% 30.8% 28.4% 31.8% 29.6% 30.3% 32.1% 27.1% Members are male majority 20.3% 17.5% 21.6% 18.9% 21.6% 21.7% 19.5% 20.7% 17.5% 24.1% Equal number of females and males 49.7% 51.5% 50.1% 50.3% 50.0% 46.5% 50.8% 48.9% 50.3% 48.8% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% Gender of household earners (T2) Earners are female only 16.5% 17.2% 12.3% 19.0% 15.0% 19.0% 14.8% 17.7% 18.5% 13.6% Earners are male only 28.3% 27.2% 31.8% 29.5% 26.6% 26.3% 29.5% 27.5% 25.8% 31.8% Earners are both female and male 34.0% 27.4% 33.0% 34.3% 40.2% 35.3% 30.2% 36.6% 37.7% 28.8% No labor earnings 21.2% 28.3% 22.9% 17.2% 18.2% 19.4% 25.6% 18.3% 18.0% 25.8% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% Gender sustaining household earnings (T3) Female (>60% earnings) 20.1% 20.0% 15.4% 23.2% 19.0% 22.7% 17.7% 21.6% 22.8% 16.2% Male (>60% earnings) 58.7% 51.7% 61.7% 59.6% 62.8% 57.9% 56.7% 60.1% 59.3% 58.0% No labor earnings 21.2% 28.3% 22.9% 17.2% 18.2% 19.4% 25.6% 18.3% 18.0% 25.8% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% Gender sustaining household earnings + presence of children & elderly (T4) Female (>60% earnings) 1.2% 1.3% 0.9% 1.2% 1.6% 1.0% 1.1% 1.2% 1.1% 1.3% Dependents + Male (>60% earnings) 3.4% 4.7% 4.1% 2.7% 2.8% 2.9% 4.4% 2.8% 3.3% 3.7% Elderly No labor earnings 1.5% 3.0% 1.6% 1.1% 1.0% 0.9% 2.3% 1.0% 1.2% 2.0% Female (>60% earnings) 3.9% 4.8% 4.2% 3.4% 3.9% 3.3% 4.5% 3.5% 3.8% 4.1% Dependents, no Male (>60% earnings) 19.7% 20.0% 24.5% 20.3% 18.6% 15.2% 22.2% 18.1% 18.0% 22.1% Elderly No labor earnings 5.6% 9.8% 6.6% 4.4% 3.5% 3.7% 8.2% 3.9% 4.1% 7.7% Female (>60% earnings) 4.4% 5.0% 2.7% 6.8% 4.4% 3.2% 3.8% 4.8% 5.3% 3.2% No Dependents, Male (>60% earnings) 7.4% 7.5% 8.6% 7.0% 6.0% 8.0% 8.0% 7.0% 8.1% 6.4% Elderly No labor earnings 7.3% 6.3% 7.8% 6.7% 8.0% 7.8% 7.0% 7.5% 8.2% 6.1% Female (>60% earnings) 10.5% 8.9% 7.7% 11.8% 9.1% 15.2% 8.3% 12.0% 12.6% 7.6% No Dependents, Male (>60% earnings) 28.1% 19.6% 24.4% 29.6% 35.4% 31.8% 22.0% 32.3% 29.8% 25.8% no Elderly No labor earnings 6.8% 9.1% 6.9% 5.0% 5.7% 7.1% 8.0% 5.9% 4.5% 10.0% Total 100.0% 100% 100% 100% 100% 100% 100% 100% 100% 100% Source: Author’s elaboration based on the ILCS 2020. Note: Calculation applying population sampling weights. Quintiles are based on household consumption per adult equivalent. Elderly refers to stay-at-home household members between 65 and 79 years old. Dependents are children below primary-school age (<6 years old) and elders 80 years and older. Both Panel A and B are based on the same underlying microdata. Panel A percentages add up vertically to 100% within household typology; Panel B percentages add up horizontally to 100%, within the household type. Page 26 Examining the poor population across each household typology does not immediately convey explicit gender disparities. Figure 14 breaks down the pre-fiscal35 poor, non-poor and national populations by household typologies. Gender disparities are not salient when dividing the population by the gender of the adult majority (T1). As expected, households with no labor earners are over overrepresented among the poor; only 21% of households have no labor earners, but these account for 39% of the poor (T2). Similarly, households with female and male earners perform much better accounting for 34% of the population, but only 14% of the poor. On the other hand, one-fifth of households are sustained mostly (>60%) by female- earned labor incomes. These households account for 17% of the poor (T3). Figure 14. Pre-fiscal poverty, by household typologies Panel A. Gender majority (T1) Panel B. Gender of household labor earners (T2) Panel C. Gender share >60% of labor earnings (T3) Panel D. Gender-sustained household earnings (T4) Source: Authors’ calculations based on the ILCS 2020. Note: Poverty rates are based on the national average poverty line and pre-fiscal income (market income plus pensions) per adult equivalent. Analyzing consumption patterns across household types suggests that households with a female vs. male majority may prioritize expenditure decisions differently. All household types seem to allocate around one- half of their consumption to food and one-half to non-food goods and services, and durables are excluded 35Pre-fiscal poverty in Figure 18 is calculated based on market income plus pensions. Robustness checks using pre-fiscal incomes before pensions are included in the analysis. Page 27 (Table 6). Households with a majority of female earners tend to spend smaller shares on tobacco products and higher shares on education than households with a male majority of earners. Nonetheless, this correlation may be driven by the presence of children in the household. Similarly, households with both female and male earners can afford to spend a slightly larger fraction on non-food items; this pattern may be permitted by a larger share of resources when both males and females participate in labor markets. Table 6. Consumption shares across household types By Gender of household earners in labor markets (T2) Female Only Male Only Dual No earners Total Food consumption 49.9% 50.9% 49.3% 51.2% 50.3% Alcohol 0.1% 0.2% 0.2% 0.2% 0.2% Tobacco 2.6% 4.7% 4.8% 2.3% 3.9% Clothing and footwear 2.7% 2.2% 2.5% 1.3% 2.2% Utilities 16.1% 15.2% 14.8% 15.9% 15.3% Furnishings, household equip, maintenance 3.2% 2.8% 3.0% 2.8% 3.0% Health 8.1% 8.5% 7.6% 11.1% 8.7% Transport 3.7% 5.1% 5.5% 7.5% 5.5% Communications 4.1% 3.6% 3.7% 3.0% 3.6% Recreation 1.5% 0.8% 0.4% 0.4% 0.7% Education 2.6% 1.7% 4.1% 1.4% 2.6% Restaurants and accommodation 0.0% 0.0% 0.0% 0.0% 0.0% Miscellaneous 5.4% 4.2% 3.9% 2.8% 4.0% Source: ILCS 2020. Notes: Shares calculated over total food and non-food expenses. Durables and rents are excluded. Female-sustained households and households with no labor earnings depend more heavily on pensions and more volatile income sources, including remittances and agriculture (Figure 15). Labor earnings account for a larger share of incomes among male-sustained households (76% when adding up salaries from hired employment and self-employment incomes). Female-sustained households, on average, obtain a lower 64% of incomes from labor markets, relying more heavily on pensions and remittances. The sale of agricultural net incomes is similar among both types. On average, households with no labor earnings reported in the ILCS, are most dependent on pensions (34%), agricultural net incomes (27%), remittances (15%), and social assistance (14%). Figure 15. Income sources by household type By gender sustaining most household earnings (T3) By gender sustaining household + children & elderly (T4) Page 28 Source: Author’s elaboration based on the ILCS 2020. Note: Calculation applying household sampling weights. IX.Engendered CEQ Results We apply the E-CEQ analysis to the household typologies to identify potential disparities in the interaction of the fiscal system and the gendered characteristics of households. Table 7 and Table 8 present the step- by-step calculation of gender gaps for each income concept of the CEQ and household typologies. In every case, the gender gap is calculated by comparing incomes in households identified with mostly “female traits” and households identified with “male traits”.36 As an attempt to limit the heterogeneity in sociodemographic characteristics of households, we also calculate gender gaps by typology and household (pre-fiscal) quintile. The calculation of gender gaps is presented for T3, as the most disaggregated typology with sufficient observations to perform the analysis by decile. The Appendix contains complementary results using the other typologies.37 Table 7. Income estimates by household type and direct fiscal interventions (Incomes in AMD monthly per adult equivalent. Gender gaps as % of incomes for “male” types ) Population [0] Prefiscal w/o [1] Prefiscal w/ (1a) Prefiscal [1] (1b) Prefiscal [1] [2] share pensions pensions + direct taxes + direct transfers Disposable By Share of household earnings by gender (T3) Any household Female >60% share 20.1 62,173 72,194 58,073 76,346 62,225 Male >60% share 58.7 67,937 74,246 57,838 78,002 61,595 No labor incomes 21.2 36,020 53,615 53,543 64,506 64,433 Gender gap -8.5% -2.8% 0.4% -2.1% 1.0% By urban and rural location Rural: Female >60% share 6.7 53,354 61,912 52,277 67,483 57,847 Rural: Male >60% share 23.9 57,283 63,098 52,423 67,764 57,089 Rural: No labor incomes 10.6 41,106 51,828 51,796 62,403 62,371 Gender gap -6.9% -1.9% -0.3% -0.4% 1.3% Urban: Female >60% share 13.4 66,570 77,320 60,962 80,765 64,408 Urban: Male >60% share 34.9 75,234 81,881 61,548 85,014 64,681 Urban: No labor incomes 10.6 30,897 55,414 55,302 66,623 66,510 Gender gap -11.5% -5.6% -1.0% -5.0% -0.4% By national pre-fiscal quintile* Q1: Female >60% share 3.5 22,989 29,211 24,804 45,866 41,459 Q1: Male >60% share 8.4 25,672 30,194 26,980 47,345 44,131 Gender gap -10.5% -3.3% -8.1% -3.1% -6.1% Q2: Female >60% share 3.7 40,145 50,024 44,105 53,925 48,006 Q2: Male >60% share 10.7 43,537 49,922 43,514 53,018 46,610 Gender gap -7.8% 0.2% 1.4% 1.7% 3.0% Q3: Female >60% share 4.2 52,384 63,252 52,394 64,697 53,839 Q3: Male >60% share 12.6 57,214 63,634 51,053 65,219 52,638 Gender gap -8.4% -0.6% 2.6% -0.8% 2.3% Q4: Female >60% share 4.3 67,456 78,662 64,257 79,530 65,125 Q4: Male >60% share 12.9 72,058 79,238 61,039 80,061 61,863 Gender gap -6.4% -0.7% 5.3% -0.7% 5.3% Q5: Female >60% share 4.4 115,929 127,092 95,510 127,411 95,829 Q5: Male >60% share 14.1 117,332 123,750 90,156 124,651 91,058 Gender gap -1.2% 2.7% 5.9% 2.2% 5.2% 36The difference is expressed as percentage of incomes in “male-identified” households. 37The same income and location analysis will, unfortunately, not be possible for T4 due to limitations in sample size. As seen from Figure 18, several household types are representative of less than 1% of the population in each quintile. Page 29 Source: Authors’ estimation based on ILCS 2020 and E-CEQ (2022). Notes: Gender gaps are calculated as the gap between incomes in households identified with “female” vs. “male” type, as a share of (average) incomes for “male” majority households. * For the quintile disaggregation, population shares add up to people living in households identifiable as either “male” or “female” type. The remaining population lives in households with no reported labor earnings and, hence, is excluded from the calculation of gender gaps. The national pre-fiscal quintile is calculated based on market income plus pensions, per adult equivalent. Table 8. Income estimates by household type and indirect fiscal interventions (Incomes in AMD monthly per adult equivalent. Gender gaps as % of incomes for “male” types ) Population [2] (2a) Disposible (2b) Disposible [2] - [3] Consumable share Disposable [2] - VAT Other indirect taxes By Share of household earnings by gender (T3) Any household - Female >60% share 20.1 62,225 60,699 61,835 60,308 Male >60% share 58.7 61,595 60,038 61,015 59,477 No labor incomes 21.2 64,433 62,602 64,127 62,296 Gender gap 1.0% 1.1% 1.3% 1.4% By urban and rural location - Rural: Female >60% share 6.7 57,847 56,758 57,360 56,271 Rural: Male >60% share 23.9 57,089 55,881 56,497 55,289 Rural: No labor incomes 10.6 62,371 60,050 62,040 59,719 Gender gap 1.3% 1.6% 1.5% 1.8% Urban: Female >60% share 13.4 64,408 62,663 64,065 62,321 Urban: Male >60% share 34.9 64,681 62,885 64,110 62,346 Urban: No labor incomes 10.6 66,510 65,172 66,228 64,890 Gender gap -0.4% -0.4% -0.1% 0.0% By national pre-fiscal quintile* Q1: Female >60% share 3.5 41,459 40,764 41,232 40,537 Q1: Male >60% share 8.4 44,131 43,400 43,819 43,088 Gender gap -6.1% -6.1% -5.9% -5.9% Q2: Female >60% share 3.7 48,006 47,041 47,638 46,674 Q2: Male >60% share 10.7 46,610 45,585 46,215 45,190 Gender gap 3.0% 3.2% 3.1% 3.3% Q3: Female >60% share 4.2 53,839 52,676 53,468 52,305 Q3: Male >60% share 12.6 52,638 51,276 52,137 50,862 Gender gap 2.3% 2.7% 2.6% 2.8% Q4: Female >60% share 4.3 65,125 63,602 64,751 63,228 Q4: Male >60% share 12.9 61,863 60,432 61,257 59,827 Gender gap 5.3% 5.2% 5.7% 5.7% Q5: Female >60% share 4.4 95,829 92,820 95,255 92,247 Q5: Male >60% share 14.1 91,058 88,318 90,134 87,394 Gender gap 5.2% 5.1% 5.7% 5.6% Source: Authors’ estimation based on LCS 2020 and E-CEQ (2022). Notes: Gender gaps are calculated as the gap between incomes in households identified with “female” vs. “male” type, as a share of (average) incomes for “male” majority households. * Incomes are excluded from the table, but gender gaps are calculated under the same method. Population shares add up to people living in households identifiable as either “male” or “female” type. The remaining population lives in households with no reported labor earnings and, hence, is excluded from the calculation of gender gaps. * The national pre-fiscal quintile is calculated based on market income plus pensions, per adult equivalent. Impacts of the fiscal system on gender gaps The overall fiscal system reduces gender disparities across income levels, though reductions are heterogenous. Gender gaps in pre-fiscal income are much larger among the poor. The estimated gender gap in pre-fiscal income is highest among the poorest quintile, at 13%. Moreover, the pre-fiscal income gaps are u-shaped, as they decrease for quintiles 2 and 3, and again increase for the top 40% of the population (quintiles 4 and 5), though never reaching the high magnitude of the bottom-20 (Figure 16, Panel B). Despite observing the largest “equalizing” effect, the first quintile maintains the largest gender gap poorest, at almost 4% at consumable incomes. Page 30 Figure 16. Fiscal incidence on gender gaps, by income level By gender sustaining most household earnings (T3) Panel A. Any household Panel B. By quintile (pre-fiscal income plus pensions) Source: Authors’ estimation based on the ILCS 2020 and the E-CEQ methodology. Notes: Gender gap based on household typology by the majority of labor incomes (females contribute >60% of household labor incomes vs. males contribute >60% of household labor incomes). Quintiles are defined based on pre-fiscal income (market income plus pensions) or the consumption aggregate per adult equivalent. Marginal effects by fiscal intervention Direct taxes have a large marginal effect on improving the relative income position of female-type households, as compared to male-type households. The largest gender “equalizing effect” from pre-fiscal incomes (after pensions) is derived from direct taxation (see the transitions across columns on Table 7). While this result is positive at first sight, it may hide a pervasive longer-term exclusion of women from labor markets. Direct transfers are also “gender-equalizing”, albeit to a smaller extent than direct taxes. On average, direct transfers in Armenia benefit households identified as women-sustained more than male- led households. The analysis does not find significant effects of consumption taxes on the estimated gender gap. Indirect fiscal interventions—which are rather linked to household consumption decisions—lead to only small or negligible changes in the identified gender gaps (Table 8). Gender gaps remain mostly unaffected after applying VAT, while tobacco excise taxes have a more significant but still limited effect. As expected, households with no labor earners capture, on average, the largest relative benefits from direct transfers (as their eligibility is likely high, and their pre-fiscal income likely low). Household work decisions The large impact of PIT on reducing gender gaps may imply that men-sustained households are paying effectively higher taxes on personal incomes. Households with a majority of male labor earners pay a larger share of pre-fiscal incomes (20%) in direct taxes than households with mostly female labor earners (16% of pre-fiscal incomes on average). Among potential explanations for these patterns, women in Armenia could be disproportionally: (a) staying out-of-the labor force, (b) taking on relatively lower-paying employment, or (c) specializing in economic activities that avoid taxes, including informal employment, household Page 31 production for self-consumption, and other forms of work outside labor markets (such as domestic work at own-household, and childcare).38 Some evidence, however, suggests that this difference is not driven by explicit biases in the fiscal system, but rather implicit biases of fiscal policies, along with household preferences for women’s roles. As explained above, Armenia collects PIT at the individual level and, since 2020, at a flat rate. Theoretically, this scheme imposes the same marginal and average tax rate on taxable household incomes, regardless of the gender of employees, the share of household members economically active in labor markets, and the level of labor earnings. Nonetheless, in the data, households with dual earners (both male and female) lose the highest share of their pre-fiscal incomes to income taxes while benefiting the least from direct transfers. This result signals an implicit disincentive for households to allocate all their labor supply to formal labor markets. Available survey microdata does not reflect a higher likelihood of women working informal jobs. According to estimations by the SCRA (based on the LFS), informality accounted for 42% of jobs among men, and 28% among female-held jobs.39 Calculations of informal employment based on the ILCS 2020 are presented in the Appendix. The estimates from these survey data, nonetheless, do not conclude that women workers are more likely to be informal. Household care needs We leverage the T4 typology to test the effects of household care needs on women’s labor decisions. Under T4, we incorporate the typology based on gender with the majority contribution (>60%) to household earnings, as well as the presence of young children and other stay-at-home elders. In particular, we combine the presence of “dependents”—primary school children and elders over 80 years old—who likely require special care with the presence of other elders (ages 65 to 80) and older children (middle school and high school ages) who may require less attention and could assist with household and care tasks. These empirical results support the argument that traditional gender roles and care responsibilities condition women’s time and economic opportunities, impacting their interactions with labor markets and the fiscal system. The gender gap is consistently higher for households with dependents, who most likely need care instead of helping out by providing it. The fiscal system significantly reduces gender gaps for all household types, but the gaps for households with dependents remain negative and largest. On the other hand, the presence of stay-at-home elders (potential grandparents to the young children) reduces the gender gap, relative to a scenario with dependents but no assistance from elders (comparing the blue vs. orange bars in Figure 17). The reduction is most significant when transitioning from pre-fiscal to disposable incomes, highlighting the association with household labor market decisions and the associated PIT 38Domestic tasks, including childcare and care for other household members are generally classified as out-of-the labor-force. 39The SCRA estimates that informality accounted for 38.0% and 35.2% of total jobs in 2019 and 2020, respectively. Restricting to non-agricultural activities, informal employment was 18.1% and 15.3% in 2019 and 2020, respectively (Labor Market in Armenia, 2021). Informal employment includes: “(i) employees holding informal jobs (including paid domestic workers), (ii) employers and own-account workers having informal sector enterprises, (iii) all contributing (unpaid) family workers and (iv) members of informal producers’ cooperatives. Informal employment also included (unpaid) family workers engaged in production of goods exclusively for own final use by their household if their production represents a large share of household consumption. Rates are calculated for workers between 18 and 75 years old, including both main (primary) and additional (secondary) activities.” (Ibid). Page 32 liabilities. Contributory pension incomes also have a significant gender-equalizing effect, particularly among households with elders 65 to 79 years old.40 Nonetheless, the analysis has some limitations. For example, it does not capture access of the household to other sources of assistance, including access to formal and informal care arrangements. Households with no elders could pay for or receive free assistance from friends and extended family to resolve care and domestic responsibilities. On the other hand, analyses considering older children do not yield a clear pattern. This is potential evidence that only younger children and early motherhood have a clear impact on women's incentives and labor market decisions. As children enter primary school and they become more independent, these patterns may no longer hold. Unfortunately, longitudinal data on women's home responsibilities and work decisions are not available for Armenia to further test these hypotheses. An upcoming specialized survey on time use and gender collected by the World Bank could address knowledge gaps around the interaction of care needs and gender disparities. Table 9. Gender gaps across the fiscal system Population [0] Prefiscal w/o [1] Prefiscal w/ [2] Disposable [3] Consumable share pensions pensions Female Majority (%) + dependent + elder <80 1.2 42,475 59,976 57,872 56,113 Male Majority (%) + dependent + elder <80 3.4 51,136 66,529 59,578 57,711 No incomes + dependent + elder <80 1.5 20,337 46,887 57,175 55,802 Gender gap -16.9% -9.8% -2.9% -2.8% Female Majority (%) + dependent, no elder <80 3.9 54,359 63,119 56,490 54,907 Male Majority (%) + dependent, no elder <80 19.7 64,420 68,849 59,321 57,320 No incomes + dependent, no elder <80 5.6 42,526 54,264 66,444 62,518 Gender gap -15.6% -8.3% -4.8% -4.2% Female Majority (%) + elder <80, no dependent 4.4 47,524 68,837 59,107 57,171 Male Majority (%) + elder <80, no dependent 7.4 52,715 69,705 59,923 57,812 No incomes + elder <80, no dependent 7.3 24,227 57,685 65,906 64,391 Gender gap -9.8% -1.2% -1.4% -1.1% Female Majority (%), no dependent, no elder <80 10.5 73,458 78,360 66,163 64,112 Male Majority (%), no dependent, no elder <80 28.1 76,463 80,166 63,875 61,644 No incomes, no dependent, no elder <80 6.8 46,871 50,189 62,802 61,304 Gender gap -3.9% -2.3% 3.6% 4.0% Source: Authors’ estimation based on LCS 2020 and E-CEQ (2022). Notes: Gender gaps are always calculated as the gap between incomes in households identified with “female” vs. “male” type, as a share of (average) incomes for “male” majority households. 40Households that are shown as “no-elders” in T4 could still receive pension incomes, for example, though early retirees (younger than 65 years old), elders 80 years and older (classified as dependents), or contributory pensions that are not linked to retirement, such as survivor or disability pensions. Page 33 Figure 17. Gender gaps across the fiscal system Based on typology (T4) Gender sustaining household earnings + presence of elderly and dependents Source: Authors’ estimation based on the ILCS 2020 and the E-CEQ methodology. Notes: Gender gap based on household typology by the majority of labor incomes (females contribute >60% of household incomes vs. males contribute >60% of household incomes). Elders are defined as stay-at-home members between the ages of 65 and 79. Dependents are elderly members 80 years and older and children younger than 6 years old (the threshold for primary education). Figure 18. Marginal income effects, by fiscal interventions and household type Source: Authors’ estimation based on the ILCS 2020 and the E-CEQ methodology. Gender gap based on household typology by the majority of labor incomes (females contribute >60% of household incomes vs. males contribute >60% of household incomes). Elderly members are defined as stay-at-home elderly between ages 65 and 79. Dependents are elderly members 80 years and older and children younger than 6 years old (the threshold for primary education). Page 34 X. Discussion This application of the E-CEQ pilot in Armenia highlighted several lessons for the empirical application of the E-CEQ methodology in middle-income countries. Identifying household typologies in Armenia helped grasp more nuanced gender-related characteristics of households. Nonetheless, identifying gender disparities and interpreting results from the E-CEQ is seldom straightforward. This E-CEQ leveraged household typologies to understand gender disparities. The typologies sought to capture gender disparities by their proximate effects on households’ needs, eligibility for fiscal benefits, access to economic opportunities in formal labor markets, supplementing income sources and strategies, intra-household bargaining dynamics, etc. In the case of Armenia, the analysis concentrated on typologies defined by demographic and income compositions. The typologies contributed to the understanding that fiscal policies targeted at children are not currently sufficient to offset the high incidence of poverty among households with children. Overall, the fiscal system in Armenia is gender-equalizing, regardless of the definition of pre-fiscal income used and the household typology applied to identify engendered characteristics and relations within the household. The Fiscal system reverses gender gaps in the poorest households most dramatically. Pre-fiscal gender gaps are highest among poor and vulnerable households. The fiscal system reduces gender gaps most significantly for the poor, although those households still suffer the largest negative gaps. Household decisions to participate in labor markets—rather than the provision of social assistance or consumption taxes—have the largest impact on gender gaps across household types in Armenia. PIT (and pensions depending on the pre-fiscal definition) are the most equalizing fiscal policies. Nonetheless, while direct taxes improve the relative incomes of female-led or female-identified households, these effects could reflect more limited opportunities for women. The lower tax liability of female-sustained households may reflect gendered decisions to participate in labor markets, engage in formal or informal work, and allocate time between paid labor in labor markets and household responsibilities. Access to affordable and convenient care for children and elderly family members can therefore influence their labor decisions and interactions with the fiscal system --for women in particular-- and potentially for every household member. Analyzing households with children and elderly members revealed that aid from stay-at-home elders could reduce gender gaps in labor earnings for some households (those with primary school children and above). As women tend to be disproportionally responsible for household responsibilities, it is expected that the presence of children or the elderly requiring care or special attention could widen gender disparities as women’s time and decisions become even more constrained. On the other hand, some children and elders can aid with household chores and care for family members in need. Identifying policy gender biases requires a deep understanding of local norms and gender dynamics, a comprehensive mapping of policies and programs, as well as leveraging diverse sources of quantitative and qualitative information. In Armenia, despite access to a well-established and standardized household budget survey (the ILCS), data gaps to understand gender disparities include intrahousehold variables related to time-use, childcare, and elderly care, among others. An important limitation in applying the E- CEQ methodology is that agricultural sales and net incomes—allegedly more significant among poor and vulnerable households, and the largest sector of employment for women—are not attributable at the Page 35 individual level. Hence, incomes considered to calculate gender gaps necessarily exclude agricultural sector jobs. Going forward, the comprehensive and accurate application of the E-CEQ will most likely require complementary sources of data, or ideally, incorporating more nuanced questions into HBSs in Armenia and other countries. The E-CEQ also opens the door for new questions and knowledge gaps that must be answered in future research. The analysis came across a recurrent yet unresolved policy question in Armenia, regarding potential distortionary incentives of social protection policies on workers’ behavior. Future E-CEQ analyses may provide important contributions to answering this question. The gender effects of tax instruments were more relevant from the side of labor markets, rather than consumption. Nonetheless, further analysis could explore whether the disaggregation of consumption data captured in HBS is adequate to capture gender-specific needs for consumption of goods and services. Further research could disaggregate the potential effects of grandmothers and grandparents in the household. Despite these caveats, the results from the E-CEQ in Armenia confirm the importance of incorporating a gender lens into incidence fiscal analysis. Fiscal policy can be a powerful policy instrument to reduce gender gaps. However, fiscal policies may also inadvertently contribute to widening or perpetuating gender gaps. Ultimately, the E-CEQ provides a powerful tool to answer these questions empirically and based on the specific knowledge and background of each country. References Asian Development Bank. 2019. Armenia Country Gender Assessment. December 2019. Attah, Ramlatu and Elayn M. Sammon. 2020. Integrated Social Protection Systems Country Case Study – Armenia. Oxford Policy Management. March 2020. https://www.unicef.org/eca/media/15966/file. Carrasco Nunez, H., J. Gao, D. Sharma, A. Fuchs, & M. F. Gonzalez Icaza (2021). “Fiscal Policy, Poverty, and Inequality in Armenia.” Draft working paper. March 2021. Carrasco Nunez, H. and C. Cancho. 2019. Assessing the Distributional Impacts of PIT Reform in Armenia. World Bank. Poverty and Equity. April 2019. https://documents1.worldbank.org/curated/en/602441560198556137/pdf/Assessing-the- Distributional-Impacts-of-PIT-Reform-in-Armenia.pdf Cuberes, D. and M. Teignier (2016). “How important are labor market gender gaps in the South Caucasus?” Paper prepared for the FY16 South Caucasus Gender Assessment TA. May 2016. Delgado Coelho, Maria, Aieshwarya Davis, Alexander D Klemm, and Carolina Osorio Buitron. 2022. Gendered Taxes: The Interaction of Tax Policy with Gender Equality. Working Paper No. 2022/026. International Monetary Fund. February 4, 2022. Deloitte. 2021. International Tax. Armenia Highlights. https://www2.deloitte.com/content/dam/Deloitte/ global/Documents/Tax/dttl-tax-armeniahighlights-2021.pdf Garcia-Peña Bersh, Natalia. 2019. “A gendered fiscal incidence analysis for Barbados”, Master’s thesis for the Master in Economics, Departamento de Economia, Universidad de San Andres, Buenos Aires, Argentina. Grown, Caren, Jon Jellema, Sailesh Tiwari, and Mariano Sosa. 2022. ‘XXXX’. World Bank. Page 36 Greenspun, Samantha, 2019 “A gender sensitive fiscal incidence analysis for Latin America: Brazil, Colombia, The Dominican Republic, Mexico, and Uruguay” Grown, Karen and Valodia, Imraan eds. Taxation and gender equity. A comparative analysis of direct and indirect taxes in developing and developed countries. Routledge International Studies in Money and Banking. Abingdon: Routledge, pp. 261–298. Lustig, Nora (Editor). 2018. CEQ Handbook. Estimating the Impact of Fiscal Policy on Inequality and Poverty. CEQ Institute at Tulane University and Brookings Institution Press. Ministry of Finance. Budget data. https://www.minfin.am/hy/page/byujei_operativ_hashvetvutyunner/. Accessed 4th February 2022. Statistical Committee of the Republic of Armenia (SCRA) (2019). Social Snapshot and Poverty in Armenia 2019. ----------. 2020. Social Snapshot and Poverty in Armenia 2020. ----------. 2021. Statistical Data (Database). https://armstat.am/en/. Ministry of Labor and Social Affairs of the Republic of Armenia (MoLSA), World Bank, and UNICEF Armenia. 2020. CORE DIAGNOSTIC OF THE SOCIAL PROTECTION SYSTEM IN ARMENIA. June 2020. https://www.unicef.org/armenia/media/9286/file/Core%20Diagnostic%20of%20the%20Social%20Pr otection%20System%20in%20Armenia.pdf. Rodriguez-Chamussy, Lourdes, Nistha Sinha, and Andrea Atencio. (2018). The Economics of the Gender Wage Gap in Armenia. Policy Research Working Paper No. 8409. World Bank, Washington, DC. Rubiano-Matulevich, Eliana. 2020. Want to celebrate fathers? Let’s talk about paternity leave. Blog posted. June 18th, 2020. https://blogs.worldbank.org/opendata/want-celebrate-fathers-lets-talk- about-paternity-leave. Stotsky, Janet. 1997. How Tax Systems Treat Men and Women Differently. Finance and Development Magazine. International Monetary Fund. USAID. 2019. Armenia Gender Analysis Report. August 2019. on was produced for review by the United States Agency for International Development. It was prepared by Banyan Global. Wai-Poi, Matthew and Jeffrey Woodham. 2022. Fiscal Policy, Equity and Gender: A case study of Jordan. Poverty and Equity Global Practice, World Bank. DRAFT: January 2022. WBL (Women Business and the Law). 2021. Armenia Brief. https://wbl.worldbank.org/content/dam/documents/wbl/2021/snapshots/Armenia.pdf. World Bank. 2017. “Why should we care about care? The role of childcare and eldercare in Armenia.” Poverty and Equity Global Practice. April 2017. ----------. 2020. “Gender related inequalities emerging from COVID-19.” Poverty and Equity Global Practice. Version: April 6th, 2020. World Vision Armenia. 2017. “Caring for Equality Baseline Report.” March 2017. Younger and Khachatryan (2017). Fiscal Incidence in Armenia. CEQ Working Paper 43. Page 37 Appendix Appendix 1. Time Allocation across genders in Armenia, previous evidence Respondent Responsible Male Female Total Usually, me 4% 77% 43% Staying home with child when Equally/Together 37% 19% 27% he/she is sick Usually, partner 60% 4% 30% Usually, me 4% 83% 47% Daily childcare Equally/Together 19% 14% 16% Usually, partner 77% 4% 37% Usually, me 18% 64% 43% Collecting child from school or Equally/Together 35% 22% 28% day care Usually, partner 48% 14% 29% Usually, me 9% 42% 28% Playing or taking child to leisure Equally/Together 69% 53% 60% activities Usually, partner 22% 5% 12% Usually, me 10% 70% 42% Helping child to do homework Equally/Together 38% 23% 30% Usually, partner 53% 7% 29% Source: World Vision 2017. Note: Data collection from July to September 2016. Page 38 Appendix 2. Gender disparities in time allocation, 2022 Average share of time allocation, regular weekend day Work and Elderly Domestic Sleeping Childcare commuting Studying Leisure Selfcare Transport Others care tasks to work Location Female respondent Armenia 34.3 2.1 9.8 5.4 4.5 21.3 15.6 5.3 1.3 0.4 Male respondent Armenia 37.9 1.6 3.5 13.1 2.9 26.3 4.0 5.5 4.1 0.5 Female respondent Yerevan 33.2 1.8 7.9 4.7 6.4 24.2 14.7 5.7 1.5 0.3 Male respondent Yerevan 37.2 1.0 2.9 10.6 4.4 28.9 4.2 5.8 3.7 0.4 Female respondent Other Urban 34.3 1.8 10.6 4.4 4.3 21.9 15.8 5.2 1.2 0.5 Male respondent Other Urban 38.4 1.7 3.4 10.8 2.9 27.6 3.9 5.8 4.0 0.5 Female respondent Rural 35.1 2.5 10.8 6.6 3.3 18.7 16.2 5.0 1.3 0.5 Male respondent Rural 38.0 1.9 4.1 16.3 2.1 23.6 4.0 5.1 4.4 0.6 Individual characteristics Female respondent No Tertiary Education 35.0 2.2 10.9 6.6 3.5 19.8 16.0 5.0 1.3 0.5 Male respondent No Tertiary Education 38.2 1.8 3.7 14.4 2.3 25.0 4.1 5.5 4.2 0.4 Female respondent Tertiary Education 33.7 2.0 9.0 4.4 5.4 22.5 15.3 5.5 1.3 0.4 Male respondent Tertiary Education 37.6 1.4 3.4 11.7 3.6 27.6 4.0 5.5 4.0 0.7 Female respondent 18-24 31.6 1.0 8.6 3.3 12.9 22.5 11.4 7.1 1.6 0.0 Male respondent 18-24 35.3 1.3 1.8 11.0 6.8 27.2 3.6 8.3 4.3 0.5 Female respondent 25-34 30.4 2.0 17.8 4.1 3.8 18.7 16.1 5.6 1.2 0.3 Male respondent 25-34 35.9 1.7 3.6 14.7 3.1 24.1 4.1 5.5 5.1 0.3 Female respondent 35-44 32.4 2.4 13.7 7.2 2.9 17.0 17.4 4.9 1.3 0.6 Male respondent 35-44 36.7 2.5 4.6 15.3 1.8 24.0 3.9 5.8 4.6 0.5 Female respondent 45-64 36.0 2.4 5.5 6.8 3.4 22.1 16.6 5.0 1.5 0.6 Male respondent 45-64 40.4 1.5 3.4 12.1 2.2 27.4 4.1 4.6 3.3 0.8 Female respondent 65+ 40.0 2.0 4.2 3.0 4.2 28.0 13.5 4.7 0.9 0.4 Male respondent 65+ 41.1 0.3 3.5 9.5 2.8 31.5 4.2 4.5 2.4 0.5 Household composition Female respondent Household no children 38.0 2.1 0.7 5.1 5.6 25.4 15.5 5.7 1.3 0.5 Male respondent Household no children 38.3 1.5 0.7 11.8 3.6 29.1 5.0 5.6 3.7 0.6 Female respondent Household with children 31.7 2.0 16.1 5.6 3.8 18.5 15.7 5.0 1.4 0.4 Male respondent Household with children 37.6 1.7 5.7 14.0 2.5 24.1 3.3 5.4 4.3 0.5 Female respondent Household no elderly (65+) 32.8 1.6 11.2 5.9 4.5 20.2 16.0 5.5 1.5 0.5 Male respondent Household no elderly (65+) 37.1 1.2 3.7 13.8 3.1 25.7 4.0 5.7 4.2 0.5 Female respondent Household with elderly (65+) 36.9 2.9 7.5 4.4 4.5 23.2 14.8 5.0 1.0 0.4 Male respondent Household with elderly (65+) 39.5 2.4 3.3 11.5 2.6 27.5 4.0 5.0 3.8 0.5 Female respondent Household no young children (0-6) 36.5 2.3 4.9 5.5 5.2 22.9 15.4 5.5 1.3 0.5 Male respondent Household no young children (0-6) 38.5 1.6 2.2 12.3 3.3 27.6 4.4 5.7 4.0 0.6 Female respondent Household w/ young children (0-6) 29.4 1.5 20.7 5.0 3.0 17.8 16.1 4.7 1.3 0.3 Male respondent Household w/ young children (0-6) 36.7 1.7 6.4 14.7 2.2 23.5 3.3 5.1 4.2 0.3 Female respondent No young children & no elders 34.9 2.0 5.3 6.5 5.5 21.8 16.0 5.8 1.6 0.6 Male respondent No young children & no elders 38.0 1.1 2.2 12.8 3.6 26.9 4.4 6.1 4.2 0.7 Female respondent No young children, elders 38.8 2.9 4.3 4.1 4.9 24.5 14.5 5.1 1.0 0.3 Male respondent No young children, elders 39.4 2.3 2.0 11.3 2.7 28.8 4.3 5.0 3.6 0.5 Female respondent Young children, no elders 29.0 1.0 21.6 4.9 2.9 17.5 16.2 4.8 1.4 0.2 Male respondent Young children, no elders 35.6 1.4 6.4 15.8 2.2 23.3 3.4 5.1 4.2 0.2 Female respondent Young children & elders 30.5 2.9 18.3 5.4 3.2 18.7 16.0 4.4 1.0 0.6 Male respondent Young children & elders 39.7 2.4 6.5 11.9 2.1 24.0 3.3 5.2 4.2 0.4 Female respondent No children & no elders 36.4 1.9 0.8 6.2 5.9 24.0 16.1 6.1 1.5 0.5 Male respondent No children & no elders 37.9 1.0 0.6 12.8 4.0 28.4 5.0 5.9 3.8 0.6 Female respondent No children, elders 40.1 2.4 0.5 3.5 5.2 27.2 14.7 5.2 0.9 0.4 Male respondent No children, elders 39.0 2.5 0.9 9.8 2.8 30.5 5.2 5.2 3.6 0.4 Female respondent Children, no elders 30.7 1.5 17.1 5.8 3.8 18.1 16.0 5.1 1.5 0.4 Male respondent Children, no elders 36.6 1.4 5.8 14.5 2.5 23.7 3.4 5.6 4.5 0.5 Female respondent Children & elders 33.9 3.3 14.0 5.2 3.9 19.4 15.0 4.8 1.0 0.4 Male respondent Children & elders 39.9 2.2 5.3 12.9 2.4 24.9 3.0 4.9 4.0 0.5 Proxy welfare status of the household Female respondent Household no extreme poverty 33.8 2.0 10.2 5.1 4.9 21.3 15.5 5.4 1.3 0.4 Male respondent Household no extreme poverty 37.9 1.5 3.4 13.1 3.1 26.3 3.9 5.5 4.2 0.5 Female respondent Household in extreme poverty 36.4 2.3 8.1 6.5 2.7 21.4 16.3 4.8 1.2 0.4 Male respondent Household in extreme poverty 37.7 2.7 4.7 13.1 2.1 25.6 4.8 5.3 3.3 0.9 Source: Based on data from the survey for “Assessing Gender Disparities in Time Allocation and Household Responsibilities in Armenia ”, 2022, round 1. Notes: Time allocation shares are calculated as percentage of the total time (in hours) reported by each respondent in the survey. This adjustment corrects for respondents reporting less or more than 24 hours of activity in one day, a common issue faced by time-use surveys. Extreme poverty status is proxied by the respondent’s self-assessment that her/his household cannot afford sufficient food. Individual weights applied to represent the national adult population of Armenia. Page 40 Appendix 3. Pre-fiscal poverty by the presence of children and elderly Panel A. Household has children (<6 years old) Panel B. Household has children (<12 years old) Panel C. Household has children (<18 years old) Panel D. Household has any elder member (65+) Panel E. Household has elderly members (65-79) Panel F. Household has elderly members (80+) Source: Authors’ calculations based on the ILCS 2020. Note: Poverty rates are based on the national average poverty line and the pre-fiscal income aggregate per adult equivalent, calculated based on the CEQ methodology (Lustig, 2018). Panels D-H include any household in the survey. Appendix 4. Pre-fiscal poverty, by household typologies and the presence of dependent members Panel A. Typology T3: Gender sustaining Panel B. Typology T4: Gender sustaining household earnings + household earnings presence of children & elderly Source: Authors’ calculations based on the ILCS 2020. Appendix 5. Gender gaps by fiscal intervention, household typologies T1 and T2 Panel A. Direct fiscal interventions Population [2] (2a) Disposible [2] - (2b) Disposible [2] - [3] share Disposable VAT Other indirect taxes Consumable By Gender majority (T1) Members Female majority 30.1 63,365 61,838 63,004 61,477 Members Male majority 20.3 61,275 59,923 60,632 59,280 Members Equal share 49.7 62,120 60,357 61,627 59,886 Gender gap 3.4% 3.2% 3.9% 3.7% By Gender of household labor earners (T2) Female earner only 16.5 62,258 60,728 61,903 60,374 Male earner only 28.3 61,155 59,629 60,588 59,094 Dual eaners 34.0 62,011 60,433 61,423 59,851 No labor incomes 21.2 64,433 62,602 64,127 62,296 Gender gap 1.8% 1.8% 2.2% 2.2% Panel B. Indirect fiscal interventions Population [0] Prefiscal w/o [1] Prefiscal w/ (1a) Prefiscal [1] (1b) Prefiscal [1] [2] share pensions pensions + direct taxes + direct transfers Disposable By Gender majority (T1) Members Female majority 30.1 56,822 68,457 57,479 74,343 63,365 Members Male majority 20.3 64,263 71,994 57,768 75,502 61,275 Members Equal share 49.7 60,217 69,037 56,347 74,810 62,120 Gender gap -11.6% -4.9% -0.5% -1.5% 3.4% By Gender of household labor earners (T2) Female earner only 16.5 58,782 69,168 57,782 73,643 62,258 Male earner only 28.3 60,380 66,429 55,304 72,280 61,155 Dual eaners 34.0 75,247 81,985 60,108 83,888 62,011 No labor incomes 21.2 36,020 53,615 53,543 64,506 64,433 Gender gap -2.6% 4.1% 4.5% 1.9% 1.8% Source: Authors’ estimation based on the ILCS 2020 and the E-CEQ methodology. Page 42 Appendix 6. Incidence of fiscal interventions, by pre-fiscal quintiles excluding pensions Population [0] Prefiscal w/o [1] Prefiscal w/ [2] Disposable [3] Consumable share pensions pensions Share of household earnings by gender (T3), by pre-fiscal quintile (excluding pensions)* Q1: Female >60% share 3.4 16,881 36,006 44,980 43,962 Q1: Male >60% share 5.9 18,369 31,032 48,026 46,926 Gender gap, quintile 1 -8.1% 16.0% -6.3% -6.3% Q2: Female >60% share 4.1 38,266 50,817 47,783 46,401 Q2: Male >60% share 10.8 38,573 47,859 45,115 43,790 Gender gap, quintile 2 -0.8% 6.2% 5.9% 6.0% Q3: Female >60% share 4.6 55,034 63,972 56,228 54,659 Q3: Male >60% share 13.0 54,585 60,813 52,393 50,725 Gender gap, quintile 3 0.8% 5.2% 7.3% 7.8% Q4: Female >60% share 3.9 72,280 78,632 65,791 63,927 Q4: Male >60% share 14.2 71,723 76,156 60,490 58,439 Gender gap, quintile 4 0.8% 3.3% 8.8% 9.4% Q5: Female >60% share 4.1 122,595 127,219 94,690 91,036 Q5: Male >60% share 14.8 117,560 121,006 88,328 84,758 Gender gap, quintile 5 4.3% 5.1% 7.2% 7.4% Source: Authors’ estimation based on the ILCS 2020 and the E-CEQ methodology. The quintiles in this table are calculated based on pre-fiscal w/o pensions income. Appendix 7. Incidence of fiscal interventions on household incomes and gender gaps Individual or household characteristics All Informal worker Formal worker Out of LF Out of WAP By gender Female 52% 44% 52% 62% 47% Male 48% 56% 48% 38% 53% Total 100% 100% 100% 100% 100% By household earnings Female (>60% earnings) 20% 18% 27% 19% 16% Male (>60% earnings) 59% 59% 72% 45% 66% No labor earnings 21% 23% 1% 37% 18% Total 100% 100% 100% 100% 100% By household earnings and member composition Female (>60% earnings) 1% 1% 1% 1% 2% Dependents Male (>60% earnings) 3% 2% 3% 3% 5% + Elderly No labor earnings 2% 1% 0% 3% 1% Female (>60% earnings) 4% 3% 4% 3% 6% Dependents, Male (>60% earnings) 20% 17% 20% 13% 35% no Elderly No labor earnings 6% 5% 0% 8% 9% No Female (>60% earnings) 4% 3% 5% 6% 2% Dependents, Male (>60% earnings) 7% 8% 8% 8% 6% Elderly No labor earnings 7% 6% 0% 16% 1% No Female (>60% earnings) 11% 11% 17% 8% 6% Dependents, Male (>60% earnings) 28% 32% 41% 20% 21% no Elderly No labor earnings 7% 10% 1% 10% 6% Total 100% 100% 100% 100% 100% By household quintile (pre-fiscal) Quintile 1 20% 22% 13% 21% 26% Quintile 2 20% 21% 18% 20% 22% Quintile 3 20% 19% 20% 20% 21% Quintile 4 20% 20% 23% 19% 17% Quintile 5 20% 18% 27% 20% 14% Total 100% 100% 100% 100% 100% Source: Authors’ calculations based on the ILCS 2020. Page 43