Policy Research Working Paper 10547 Fiscal Policy as a Tool for Gender Equity in El Salvador Monica Robayo-Abril Ana Maria Tribin José Andrés Oliva Poverty and Equity Global Practice August 2023 Policy Research Working Paper 10547 Abstract This paper analyzes fiscal incidence in El Salvador through an alarming rate of 42.7 percent. An increasing gender a gender lens using the Commitment to Equity model. The gap in poverty rates is also observed among households study aims to identify fiscal policies that promote gender where women are the sole providers. The results show that equality and facilitates evidence-based policy recommenda- the net fiscal system can increase the incidence of poverty tions aimed at reducing gender disparities and promoting among this group by 4.3 percentage points. In compari- more inclusive fiscal policies. The analysis shows that fiscal son, it increases by only 2.3 percentage points among their policy is not pro-poor, as it can lead to a 3.1 percentage male counterparts. A microsimulation exercise of poten- point increase in overall poverty using the US$6.85 2017 tial fiscal reforms to improve the welfare position of these purchasing power parity poverty line, disproportionately households reveals that a fiscal package eliminating indirect impacting particular groups. Households headed by single subsidies, social security exemptions for vulnerable groups, women with at least one child under six years old experience and conditional cash transfers to households that meet cer- a poverty rate increase of 4.3 percentage points, reaching tain conditions could reverse these unfavorable outcomes. This paper is a product of the Poverty and Equity Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at mrobayo@worldbank.org, atribinuribe@worldbank.org, and joliva@fusades.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Fiscal Policy as a Tool for Gender Equity in El Salvador1 Monica Robayo-Abril Ana Maria Tribin José Andrés Oliva JEL classification: H22, I38, D31 Keywords: gender inequality, poverty, fiscal policy, fiscal incidence, social spending, taxes 1This analysis was prepared as part of a collaboration between the World Bank and UNDP to inform the gender agenda in the country. The document was part of the Poverty Program for El Salvador in the World Bank’s Global Poverty and Equity Practice and the gender team for Regional Bureau for Latin America and the Caribbean at the United Nations Development Program. This work was also supported by Global Solutions Group on Fiscal and Social Policy of the World Bank. The study was carried out by a team composed of Monica Robayo-Abril (Senior Economist, World Bank), Ana María Tribin (Senior Economist, World Bank) and José Andrés Oliva (Researcher, Fusades). The authors are grateful for the valuable comments given by Hugo Ñopo (World Bank), Luis Dasilva Depaiva, Yolanda Villar, Gonzalo Aguilar and Rodrigo Barraza (UNDP) and research assistance from Daniela Dos Santos (UNDP) and those attending the workshop of tax incidence with a focus on gender, organized by the Global Solutions Group on Fiscal and Social Policy of the World Bank in December 2021. I. Introduction Disparities in access to opportunities for women in the labor market may contribute to an increase in the poverty of households headed by women or where women are the breadwinners. While fiscal policies should strive to promote economic well-being for vulnerable households, their effectiveness in closing gender gaps is usually insufficient. Fiscal policy can influence gender equality through several channels. Gender inequalities within labor markets significantly shape the behaviors and opportunities of both women and men. Moreover, direct taxes and social security contributions represent primary instances where gender biases are manifested in regulations, subsequently influencing women’s incentives to engage in the labo r market and choose between formal and informal sector employment. Furthermore, the monetary value of public services, such as education and health care, affects women as they often assume the role of caregiver for children and elderly family members in many households. This role can impact their reservation wages and labor-force participation decisions. The decisions men and women make regarding the labor market and the barriers they face in accessing opportunities can have significant implications for poverty and vulnerability, as labor income is usually a fundamental source of income for households living in poverty or at risk of falling into poverty. Fiscal policies can, therefore, have unified consequences that reinforce gender gaps. This study employs the CEQ (Commitment to Equity2) methodology, expanded with a gender lens, to explore the role of fiscal policies in widening or reducing existing gender gaps in El Salvador. The CEQ methodology determines the impact of the overall fiscal system on different poverty and inequality measures by comparing the market income scenario, i.e., the pre-fiscal scenario, with the post-fiscal scenario, where households have already paid and received direct and indirect taxes and transfers, respectively. As presented in Grown and Valodia (2010) and Greenspun (2019), the method can be expanded to include a gender perspective by disaggregating households according to their socio-demographic characteristics as ‘female’ and ‘male’ households, allowing the identification of the impact of fiscal policy on gender gaps. Thus, this methodology provides a tool for evidence-based recommendations that employ a gender approach. The inclusion of a gender perspective in analyzing the impacts of fiscal policies requires several steps: (i) identifying the typology of ‘female’ and ‘male’ households, (ii) gaining a detailed understanding of the country’s tax and benefits system, (iii) collecting relevant macro and fiscal data and (iv) 2The World Bank’s Poverty and Equity Group partnered with Tulane University’s Department of Economics to implement the CEQ in a set of countries in different regions. This approach is one of the first efforts to comprehensively assess the system of taxes and benefits in developing countries (including subsidies and indirect taxes and in-kind benefits in the form of free education and health care) and be able to compare countries over time. This type of analysis sheds light on the main constraints that impede an effective reduction of poverty and inequality through tax policies and social benefits. 2 performing a meticulous microsimulation analysis using administrative and household survey data. The initial step involves categorizing households based on gender composition or the presence of specific gender-related factors. It acknowledges that households can differ significantly in terms of their structure, roles, and socio-economic characteristics, and it is crucial to capture these distinctions when conducting gender-focused analyses. Second, it needs a detailed examination of the tax policies, regulations and benefit schemes implemented in the country, and how different types of taxes (e.g. income tax, consumption tax) and social benefits (e.g. child allowances, social security benefits) interact with individuals and households. Third, macro and fiscal data, including information on government revenues, expenditures and budget allocations, provide a broader context for understanding the fiscal landscape and is critical to assess the allocation of public resources and their potential gender implications. Fourth, microsimulation analysis using administrative and household survey data allows us to quantify the potential gendered impacts of various fiscal policy scenarios on individuals or households. This new evidence is crucial for identifying fiscal reforms that can contribute to gender equality and inform the country’s dialogue on egalitarian and pro-poor tax reforms. While evidence on the gender dimensions of fiscal incidence holds significant relevance for informing policy guidance on tax, transfer and expenditure reforms, it is surprisingly scarce. There are no previous studies for El Salvador, and while there is limited evidence for other countries, there are very few studies that look comprehensively at the incidence of the full fiscal system with a gender lens. Greenspun (2013) reviewed the existing literature, including 16 gendered fiscal incidence studies, delving into their research scope, methodologies and principal discoveries. Seven studies scrutinize gender equity in government spending on education and health, five studies focus on tax incidence, and the remaining four studies center on estimating the effect of taxes and transfers. While these studies yield valuable insights into gender equity implications of specific taxes and transfers or some combinations, none examine the incidence of taxes and cash and in-kind transfers in unison. Greenspun (2019) conducted fiscal incidence analysis for several Latin America Countries to fill this knowledge gap. When looking only at direct taxes and transfers, evidence for other countries is mixed. For example, according to Garcia-Peña Bersh (2019), direct transfers mainly benefit men in Barbados. For New Zealand, Aziz (2013) finds that the burden of direct taxation falls mainly on men, with an equalizing effect on final income. Bakker (2017) indicates that personal income tax (PIT) deductions in Canada benefit men. Grown and Valodia (2010) reveal that, in terms of indirect taxes and subsidies, male-type households bear the heaviest VAT burden in Argentina, Barbados, Ghana, Mexico, South Africa, Uganda and the United Kingdom, but not in India or Nicaragua.3 3All these results sound sensitive to the presence or absence of zero rates, exemptions and preferential rates for domestic public goods and essential goods and services, therefore some changes can correct some of these inequalities. 3 We analyzed El Salvador’s tax and transfer system using the latest available multipurpose household survey (EHPM) from 2019, collected by the General Directorate of Statistics and Censuses (DIGESTYC), and administrative data and macro data on fiscal accounts. We used the 2019 household survey as the most recent record of a situation without the impact of the COVID-19 crisis that temporarily affected household income and the reliability of household surveys collected in the country during the pandemic. Since data collection stopped in the quarantine months and the full theoretical sample was not covered, the 2020 survey data are not fully comparable to previous years. Therefore, it is not advisable to use data from an atypical crisis year like 2020 to evaluate the tax system’s impact.4 The findings of this study suggest that the tax and transfer system in El Salvador disproportionately affects households with women as the sole providers, leading to increased poverty rates. Specifically, the poverty rate among households with female breadwinners was 22.3 percent before fiscal policy, which is higher than that among households with male breadwinners (15.5 percent). However, after the deductions for tax payments and the receipt of transfers, poverty rates increased disproportionately more for female households, with households with female breadwinners reaching a poverty rate of 26.6 percent and their male counterparts, 17.8 percent.5 As a result, the gender gap expanded. Furthermore, the study finds that the poverty rates of women who are the sole providers for households with young children experienced the most significant increase. These households had among the highest pre-fiscal poverty rates, and fiscal policy failed to target this vulnerable group, increasing poverty by 4.3 percentage points, reaching a rate of 42.7 percent, 1.5 times the country’s national poverty rate. To address these issues, the paper simulates a potential fiscal reform through a microsimulation exercise. This reform includes an exemption from social security contributions, a conditional transfer to female-headed households and an elimination of indirect subsidies to the most affluent households to avoid fiscal policy unintended consequences in exacerbating gender gaps. The study also recommends investing in public childcare and care services in the medium term to prevent adverse effects on households with women as the breadwinners. The paper is organized as follows. Section II provides a description of the labor-market situation for women in El Salvador. Section III explains the CEQ methodology with a gender approach. Section IV presents the main results of the impact of the tax and transfer system on the poverty of households with different economic and demographic profiles. Section V evaluates measures of progressive and 4 The gender analysis for El Salvador was developed from a partnership between the World Bank and the United Nations Development Program, based on the CEQ evaluation conducted by the World Bank. 5 The following are used in this study regarding lines of international poverty measured in 2017 Parity of Power Acquisitive (PPA) dollars, which are USD$ 2.15 (extreme poverty line), USD$ 3.65 (lower-middle-income line) and USD$ 6.85 (upper-middle-income line). 4 horizontal equity. Section VI analyzes the marginal impacts of each intervention of the tax and transfer system on poverty and inequality for different types of households and its coverage. Section VII presents microsimulations of potential reforms to fiscal policy to increase gender equality. Finally, Section VIII summarizes the main conclusions of the study. II. Description of the state of the labor market for women In El Salvador, there are significant disparities between men and women regarding employment status, job opportunities and income inequality. In the past 20 years, women’s participation in the labor market has been low, with less than 50 percent of working-age women (16 years and older) participating in the labor market. In contrast, around 80 percent of men participate in the labor market (Figure 1). According to recent benchmarking exercises, women’s labor-force participation rates in El Salvador are among the lowest in the Latin America and Caribbean (LAC) region.6 The COVID-19 pandemic has further exacerbated the employment situation of women, who already faced unfavorable circumstances compared to men. According to Gutiérrez, Martin and Ñopo (2021), the pandemic has led to an increase in care work for women with minor children, decreasing their labor supply in Latin America. In El Salvador, the employment gap between men and women in households with children under six years old is the second largest among 16 Latin American countries, with a gap of 43 percent. Furthermore, El Salvador is among the three countries with the highest percentage of employed women unable to work during the pandemic (53 percent), alongside Honduras and Paraguay. Although women’s unemployment rates have been lower than men’s, they have steadily increased since 2000. Men’s unemployment rate was around 8.2 percent between 1998 and 2020, while women’s unemployment rate has averaged 4.6 percent, but with a noticeable upward trend for women. As of 2020, the unemployment gap has decreased to less than 1 percentage point, whereas in 2008, it was 3 percentage points (Figure 2). 6 Source: Robayo-Abril, Monica; Barroso, Rafael. 2022. 5 Figure 1. Labor participation rate, percentage of the Figure 2. Unemployment rate, percentage of the labor working-age population (16+), women versus men, force, women versus men, 1998–2020 1998–2020 100 10 9 80 8 Percentage (%) Percentage (%) 7 60 6 5 40 4 3 20 2 1 0 0 Women Men Women Men Source: Own estimates based on EHPM (1997–2020). Note: Working-age population is defined by international standards (ILO) as suitable in terms of age to perform productive functions (1 years and older). The 2020 labor market estimates are based on official data, which may underestimate job losses due to COVID-19 due to methodological changes in the household survey. Labor informality is more frequent among women and has been accentuated in recent years.7 The prevalence of labor informality is more common among women and has been increasing in recent years. Proxying informality with social security measures commonly used in the labor literature (lack of access to social security), survey data show that the percentage of women in the economically active population without social security has been rising since 2008, leading to a widening gender gap (Figure 3). In 2008, both men and women had the same informality rate, but by 2020, the gender gap had reached 6 percentage points. Moreover, women’s inactivity is more influenced by demographic factors and caring for dependents within the household than men’s. For example, among inactive working-age women, 65 percent cite domestic and care work within the home as the primary reason for being out of the labor market, compared to only 2 percent of men (Figure 4). In contrast, men cite other reasons for their inactivity, 7 Informality is calculated using the social security contribution question in the EHPM household survey. Any worker who answered “no” to the question “Are you affiliated or covered by any private or public social security system?” is considered informal. Other concepts of informality, estimated using firm size or in occupation characteristics, are not used in this study. 6 such as attending school (31.4 percent), disability (23.7 percent) and retirement (11 percent), while these percentages are lower for women (13 percent, 11 percent and 2.7 percent, respectively). Figure 3. Informality rate, percentage of total employment, Figure 4. Reasons for inactivity, percentage of women versus men, 1998–2020 inactive women and men, 2019 77 Housework and care 65.27 1.99 75 Attend a formal education 13.01 center 31.41 Percentage (%) 73 11.41 Cannot work (disabled) 23.69 71 Illness or accident 2.73 12.35 69 Retired or pensioner 2.69 11.07 67 For family or personal 2.21 obligations 5.22 65 Others 1.2 4.7 Thought there was no job 0.47 available 1.75 Women Men 0 10 20 30 40 50 60 70 Women Men Source: Own estimates are based on EHPM (1997–2020). The definition of informality based on social security is applied, using the question “Are you affiliated or covered by any private or public social security system? ”. The COVID-19 pandemic significantly impacted women in Latin America due to the social division of labor by sex. Bergallo et al. (2021) found that a significant proportion of female employment in the region is concentrated in economic sectors considered non-essential, such as education and tourism, which were slowed down by the pandemic. In addition, confinement measures prevented domestic workers, a sector that employs between 11 million and 18 million people in the region, with high participation of women, from carrying out their work (UN-Women, ILO, and ECLAC, 2020). In El Salvador, historical averages show that up to 42.9 percent of the employed female population is engaged in commerce, composed of retail, wholesale trade, restaurants, hotels, repairs, etc., and approximately 30 percent is employed in services. About 12.7 percent are employed in the education sector, and 11.5 percent in domestic work (Figure 5). In contrast, men’s employment is distributed more evenly among the different sectors, with 21 percent in commerce, 24 percent in agriculture, 10.8 percent in construction and 7.3 percent in industry (Figure 6). This gendered distribution of employment puts women at a disadvantage during times of crisis, as seen in the case of the COVID-19 pandemic. 7 Figure 5. Distribution of employment by sector Figure 6. Distribution of employment by sector of of economic activity, women economic activity, men 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% 2010 2011 2012 2013 2014 2015 2016 2017 2018 2020 2010 2011 2012 2013 2014 2015 2016 2017 2018 2020 Domestic service Domestic service Education, health and personal services Education, health and personal services Public administration and defense Public administration and defense Banking, finance, insurance, profesional services Banking, finance, insurance, profesional services Electricity, gas, water, transport, coumunication Electricity, gas, water, transport, coumunication Retail, wholesale trade, restaurants, hotels, repairs Retail, wholesale trade, restaurants, hotels, repairs Construction Construction Other manufacturing Other manufacturing Manufacturing (food and beverages, tobacco, apparel and Manufacturing (food and beverages, tobacco, apparel and clothing) clothing) Agriculture, primary activities Agriculture, primary activities Source: Own estimates based on EHPM (1997–2020). Note: The employed population was used as a reference in both cases. According to the concepts used regarding employees in the Household Survey of El Salvador, ‘employed’ refers to those who perform work for which they receive remuneration or profit or work without cash payment in a family establishment. The 2020 estimates are based on official data, which may underestimate job losses due to COVID-19 due to methodological changes from the household survey. Women’s employment is more heavily concentrated in domestic work, while men ’s employment is more prevalent in permanent employment. Permanent employment and other types of contracts are dominated by men, with 63.6 percent and 64.2 percent of these positions, respectively. Men also 8 account for 86.5 percent of temporary employees and almost all apprentices/interns at 99.4 percent. On the other hand, women represent 91.9 percent of those employed in domestic service and 65.7 percent of those who are self-employed-business owners (Table 1, Panel a). Gender gaps in labor earnings are also observed along the wage distribution. As Table 1, Panel b, shows, 68 percent of people earning between 20 percent and 40 percent above the minimum wage are men. Also, among higher income earners, men’s earnings are also significantly higher. Women also have higher participation among the poor. When using the international poverty line of US$ 6.85 dollars per day at purchasing power parity (PPP), taking 2017 as the base year, the poverty rate reached 28.8 percent in 2019; more than half of the poor are women (53.4 percent women and 46.6 percent men). Likewise, women’s poverty rate is slightly higher than the men’s poverty rate (29 percent versus 28.5 percent, respectively). Table 1. Distribution of employment status by gender Panel a. Women (%) Men (%) Total (%) Permanent employment 457,358.0 36.4% 798,107.0 63.6% 1,255,465.0 100.0% 35.9% 43.6% Self-employed (non-business owner) 378,564.0 50.4% 372,509.0 49.6% 751,073.0 100.0% 29.7% 20.3% Employed in domestic service 124,998.00 91.9% 10,982.0 8.1% 135,980.0 100.0% 9.8% 0.6% Others 69,952.0 35.3% 127,942.0 64.7% 197,894.0 100.0% 5.5% 7.0% Unpaid family member 87,427.0 52.8% 78,254.0 47.2% 165,681.0 100.0% 6.9% 4.3% Self-employed (business owner) 58,800.0 65.7% 30,639.0 34.3% 89,439.0 100.0% 4.6% 1.7% Employer 47,039.0 35.8% 84,527.0 64.2% 131,566.0 100.0% 3.7% 4.6% Temporary salaried 50,028.0 13.5% 321,676.0 86.5% 371,704.0 100.0% 3.9% 17.6% Apprentices 35.0 0.6% 6,016.0 99.4% 6,051.0 100.0% 0.0% 0.3% TOTAL 1,274,201.0 41.0% 1,830,652.0 59.0% 3,104,853.0 100.0% 9 Panel b. Women (%) Men (%) Total (%) Below SM (20%) 285,879 35.1% 527,777 64.9% 813,656 100.0% 22.4% 28.8% In SM (+-20%) 632,281 49.6% 642,611 50.4% 1,274,892 100.0% 49.6% 35.1% Above SM, between 20% and 40% 63,045 31.7% 135,701 68.3% 198,746 100.0% 4.9% 7.4% Above SM, between 40% and 80% 81,515 32.5% 169,343 67.5% 250,858 100.0% 6.4% 9.3% Above SM, between 80% and double 20,805 35.0% 38,706 65.0% 59,511 100.0% 1.6% 2.1% Above the SM, between double and 2.50 times 57,683 40.4% 84,975 59.6% 142,658 100.0% 4.5% 4.6% Above the SM, 2.5 and 3 times 26,779 41.8% 37,325 58.2% 64,104 100.0% 2.1% 2.0% Above the SM, between 4 and 5 times 27,824 37.5% 46,297 62.5% 74,121 100.0% 2.2% 2.5% Above SM, above 5 times 78,390 34.6% 147,917 65.4% 226,307 100.0% 6.2% 8.1% 1,274,201. TOTAL 0 41.0% 1,830,652.0 59.0% 3,104,853.0 100.0% Source: Own estimates based on EHPM 2019. Note: MW stands for minimum wage III. Methodology: Expanding the Equity Commitment Assessment to understand fiscal policy with a gender lens This section aims to present a gender-sensitive adaptation of the methodology of fiscal incidence. The first important step is to identify any explicit and implicit gender biases that may exist in laws and regulations (Stotsky, 1996). In terms of explicit gender bias, which means treating men and women differently in legislation, El Salvador’s institutional framework has made significant progress towards 10 gender equality, although there are still some exceptions. The country’s global ranking on the World Bank’s Women, Business and the Law Index has improved to 45th place with a score of 88.8 in 2021, compared to 78.1 20 years ago.8 While El Salvador performs relatively well in areas such as mobility, workplace, entrepreneurship and assets, it falls behind in terms of salary, marital conditions, paternity, and pensions.9 However, implicit gender bias may still be at play, as women primarily shoulder care responsibilities in El Salvador. Implicit gender bias refers to social norms and economic behavior that affect men and women differently. For example, El Salvador’s inflexible labor law, which includes restrictions on part- time work, disproportionately affects women with children with significant caregiving responsibilities. The tax advocacy approach aims to comprehensively assess the tax/benefit system using a diagnostic tool called the Equity Commitment Assessment (CEQ). With some modifications, the CEQ methodology allows for assessing the gender impacts of current fiscal policies. The CEQ analysis in this paper includes seven definitions of income, each of which is described in Diagram 1. In all cases, the per capita value was used.10 In conceptual terms, each income is defined as follows: ➢ Market income comprises household income and wages, and other non-labor income, such as private transfers, excluding any element of fiscal policy. ➢ Market income plus pensions refers to the previous income with the addition of pensions. ➢ Gross income is the sum of market income plus pensions and direct transfers. ➢ Net income is calculated by subtracting income tax and social security contributions from market income plus pensions. ➢ Disposable income can be calculated either by adding direct transfers to net income or by subtracting direct taxes and social security contributions from gross income. ➢ Consumable income is obtained by adding indirect subsidies and subtracting indirect taxes from disposable income. ➢ Final income is the consumable income plus the monetary quantification of public services received in education and health.11 8 Source: https://wbl.worldbank.org/content/dam/documents/wbl/2021/snapshots/El-salvador.pdf. 9 Despite the existence of Article 25(g) of the ‘Law on Equality, Equity and the Eradication of Discrimination against Women’, which mandates equal treatment of male and female employees, differential treatment is still prevalent. For instance, women are entitled to 18 weeks of paid maternity leave, calculated at 75 percent of their salary, whereas men are only entitled to 3 days. These disparities in paternity and maternity leave directly impact recruitment costs and contribute to women being perceived as more expensive employees than men. Additionally, women and men do not retire at the same age, with women retiring at 55 and men at 60, and differentiated mortality tables are used for pension calculation, which means women require more years of contribution to attain their pensions. 10 On this occasion, the taxable income included in the CEQ methodology was not used. For the amount per person, the income was divided by the number of household members. The CEQ market entry and the measurement of DIGESTYC differ in that the former does not include domestic workers as household members when performing the division. Additionally, the CEQ market entry includes the value of the employer’s contribution for health. No adult-equivalent measures were used at market entry. For a detailed explanation, please refer to the Commitment to Equity Handbook: Estimating the Impact of Fiscal Policy on Inequality and Poverty by Nora Lustig, 2018. 11 For a detailed explanation refer to Appendix 1. 11 Diagram 1. Definitions of income concepts Market Income (reference) = Wages and salaries, capital income, private transfers (remittances, private pensions, alimony, etc.); the imputed income from the value of the dwelling house (DIGESTYC does not include it), before taxes, social security and public transfers. Does not include contributory pensions. Contributory pensions + − Contributions − Market income plus pensions = + + − Conditional or unconditional direct transfers (in cash or Personal income taxes and non- goods), non-contributory pensions, cash-substitutable pension social security contributions. programs. − Gross income = I b Net income = I n Conditional or unconditional direct Personal income taxes and non- transfers (in cash or goods), non- pension social security contributions. contributory pensions, cash- substitutable programs. − + Disposable income = I d − Indirect consumption subsidies: + − Consumption taxes and other indirect energy, water, transport and gas. taxes. Consumable income = I c Monetary value of services received in + − Co-payments in education and health. education and health. − Final income = I f Source: Lustig and Higgins (2019). 12 The gendered tax incidence analysis acknowledges that fiscal policy can have differentiated impacts on men and women, as well as on ‘female’ versus ‘male’ households. To understand how fiscal policies affect household activities, it is important to define household types or ‘typologies’ that are appropriate for the country context. As recommended by Grown and Valodia (2010), the analysis should move beyond the traditional approach of grouping households by sex of the head of household and also consider other characteristics, such as employment and demographic structure, using a richer typology for tax incidence (Figure 7). In this study, we follow some of the definitions of Grown and Valodia (2010) and Greenspun (2019) based on demographics and labor income. Specifically, we differentiate households based on whether they are headed by a woman or a man, or from a labor perspective, we specify whether a woman is the sole provider of income or contributes with more than 50 percent of the budget. According to Grown and Valodia (2010), households can be classified according to gender. Typically, the initial classification considered is by the head of the household, although it is often conceptually confusing and empirically disordered. Among the different categories they include are the employment categories often used in the literature because employment is used as an indicator of bargaining power. This typology distinguishes between households with female family providers (without employed men), households with male family providers (without employed women), households with dual income and households without employed adults (Figure 7). Figure 7. Household typologies based on their gender relations Classification Comment By head of household Female-headed households Commonly used. Male-headed households Definitions vary by country. By adult composition Female majority Explore differences in spending patterns. Male majority By employment Female-headed who contributes over half of the household budget Explore bargaining power. Male-headed who contributes over half of the household budget Equal contribution Not employed Source: Grown and Valodia (2010). However, Greenspun (2019) introduces a different definition of female provider. According to Greenspun (2019), households with female breadwinners are those where a woman has the highest 13 labor income or where women contribute more than 50 percent of household income. For this document, we use the two definitions, i.e., households where a woman is the sole provider (only the woman is employed) or where the woman contributes more than 50 percent to the family budget. In the case of El Salvador, we compared the relevant indicators between the following types of households with the male counterparts: 1. Households headed by a woman (self-reported). 2. Households headed by a woman with dependents (children under six years of age or older adults). 3. Households headed by single female heads with at least one child under six years of age. 4. Households where the woman is the sole provider. 5. Households where labor income of a woman represents 50 percent or more of the household income. 6. Households where the woman is the sole provider of income in the household and who does not receive remittances from another country. This last classification is very relevant, given the importance of remittances for Salvadoran households. Finally, the indicators used in the analysis are defined as follows: To measure inequality, we use the Gini coefficient. The Gini index for market income versus final income was calculated for each type of household to determine whether fiscal policy increases or decreases inequality. For example, if the Gini index increased more among households headed by women when comparing market income with consumables, it is possible that fiscal policy unequally affects these households due to the gender profile of their head. Additionally, since the Gini index measures income concentration or deconcentration within each group, the Theil index was also utilized to disaggregate inequality measurement across groups. To measure the gender gap, we compare the average income of households with relevant profiles for women, such as those where they are the sole providers or heads of the household, with the average income of households with a male profile. This comparison is conducted for each type of income according to the CEQ, with a higher value indicating greater equity between groups. To measure the incidence of poverty, the headcount index is calculated for each group of households identified as relevant from a gender perspective. It compares the change in the poverty rate using market income with consumable income, thus establishing the impact of fiscal policy on poverty. Furthermore, the disparity in the poverty rate between ‘female’ households and ‘male’ households was also calculated, using both market income and consumable income. Additionally, indicators of progressivity, horizontal equity and marginal contributions to the reduction of poverty and inequality were included for each type of household considered. 14 IV. What is the impact of fiscal policy on gender gaps in El Salvador? This section examines the characteristics of households based on gender. It then analyzes the fiscal’s impact on poverty and gender gaps. Characterization of male and female households before fiscal policy A significant proportion of the poor (at least one-third of the poor) rely on market incomes that come from women (Table 2). The incidence of female-headed households is high among poor households. According to the EHPM 2019, out of the 1,938,396 households in El Salvador, 37.1 percent are headed by women, and 62.9 percent are headed by men. Similarly, using the new poverty lines established by the World Bank and using purchasing power parity (PPP) conversion factors for 2017, female-headed households represent 31.2 percent of poor households with incomes below US$ 3.65 per day at PPP and 35.8 percent of poor households living with incomes of less than US$ 6.85 per day (PPP). Many households in poverty have dependents, whether headed by men or women. Data indicate that for households in poverty headed by women, 28.7 percent have dependents (children or elderly individuals of either gender) when using a threshold of US$ 3.65 per day. In comparison, this figure rises to 32.3 percent when using a threshold of US$ 6.85 per day. Among male-headed households in poverty, these figures reach 63.7 percent and 57.4 percent, respectively. One noteworthy finding is that one in five poor households is headed by a woman with no partner or small children. Specifically, 20.6 percent of households living under US$ 3.65 per day (PPP) and 26.0 percent of those living in poverty using a line of US$ 6.85 per day (PPP) are headed by a woman who does not have a partner and does not live with young children (under six years of age). The corresponding percentages for male-headed households are lower, at 6.2 percent and 6.4 percent, respectively. Additionally, 20.1 percent of households with incomes of less than US$ 3.65 per day to PPP and 24.7 percent of those with incomes of less than US$ 6.85 per day to PPP have a woman who contributes over half of the household budget. Women are the sole breadwinners for 10 percent of households in poverty using a threshold of US$ 3.65 at PPP or 13.3 percent of households in poverty using US$ 6.85 at PPP. Finally, a high percentage of poor households in El Salvador do not receive remittances. At least 83.2 percent of households living on less than US$ 3.65 per day at PPP and 79.4 percent in poverty living on less than US$ 6.85 at PPP receive no remittances. 15 Table 2. Composition of households before fiscal policy, based on market income % of total % of total Total poor Total poor % of total poor Total poor household 3.65 6.85 households households Households 1,938,396 100.0% 157,190 100.0% 565,499 100.0% Panel "A" classification by demographics Female-headed households 719,294 37.1% 49,103 31.2% 202,206 35.8% Men-headed households 1,219,102 62.9% 108,087 68.8% 363,293 64.2% Female-headed with no dependents 118,909 6.1% 3,951 2.5% 19,615 3.5% Female-headed with dependents 600,385 31.0% 45,152 28.7% 182,591 32.3% Male-headed with no dependents 259,562 13.4% 7,952 5.1% 38,512 6.8% Male-headed with dependents 959,540 49.5% 100,135 63.7% 324,781 57.4% Female-headed household, sole provider with no young children 562,383 29.0% 32,317 20.6% 146,937 26.0% Female-headed household, sole provider with at least one young child 37,895 2.0% 6,416 4.1% 18,709 3.3% Female-headed household with a partner and no young children 100,193 5.2% 7,528 4.8% 27,945 4.9% Female-headed household with a partner and at least one young child 18,823 1.0% 2,842 1.8% 8,615 1.5% Male-headed household, sole provider with no young children 219,399 11.3% 9,753 6.2% 35,977 6.4% Male-headed household, sole provider with at least one young child 2,726 0.1% 0 0.0% 598 0.1% Male-headed household with a partner and no young children 760,246 39.2% 64,137 40.8% 224,980 39.8% Male-headed household with a partner and at least one young child 236,731 12.2% 34,197 21.8% 101,738 18.0% Panel "B" classification by labor income Female-headed household and main breadwinner 279,286 14.4% 15,704 10.0% 75,290 13.3% Male-headed household and main breadwinner 197,111 10.2% 5,722 3.6% 31,629 5.6% Both men and women are providers 1,217,141 62.8% 96,686 61.5% 356,716 63.1% Unemployed 244,858 12.6% 39,078 24.9% 101,864 18.0% Female-headed household who contributes over half of the household's income 578,475 29.8% 31,658 20.1% 139,681 24.7% Male-headed household who contributes over half of the household's income 986,675 50.9% 60,415 38.4% 257,569 45.5% Men and women contribute equally to household income 128,388 6.6% 26,039 16.6% 67,015 11.9% Unemployed 244,858 12.6% 39,079 24.9% 101,234 17.9% Female-headed household and main breadwinner, with remittances 72,143 3.7% 2,488 1.6% 14,535 2.6% Male-headed household and main breadwinner, with remittances 48,031 2.5% 520 0.3% 5,581 1.0% Both men and women are providers and receive remittances 222,796 11.5% 14,569 9.3% 56,853 10.1% Unemployed and receiving remittances 125,612 6.5% 8,782 5.6% 39,673 7.0% Female-headed household and main breadwinner, with no remittances 207,143 10.7% 13,216 8.4% 61,385 10.9% Male-headed household and main breadwinner, with no remittances 149,080 7.7% 5,202 3.3% 26,048 4.6% Both men and women are providers and do not receive remittances 994,345 51.3% 82,117 52.2% 299,863 53.0% Unemployed and not receiving remittances 119,246 6.2% 30,296 19.3% 61,561 10.9% Source: EHPM 2019. Notes: Children and the elderly are considered dependent household members. Young children are considered as younger than 6 years old. Impact of fiscal policy on inequality between ‘female’ and ‘male’ households The Gini index is commonly used to measure income inequality, but it only shows inequality within groups, not between them. So, we can assess inequality separately for female and male households, but not between them. Transitioning from market income to final income, we observed a reduction in inequality indices for each group, indicating that fiscal policy is associated with decreased income concentration within these groups. Fiscal policy contributes to reducing inequality for most female and male households, as shown in Table 3. 16 Table 3 shows that the group with the highest income inequality is male-headed households, with a Gini index of 0.425. This group has a higher concentration of income than the entire population. The group with the second-highest inequality is households where a woman is the only provider, with a Gini index of 0.402, which is slightly below the Gini index for the entire population of 0.411. The Theil index breakdowns presented in Table 4 demonstrate that the majority of inequality arises from within-group differences rather than differences between groups, both before and after the implementation of fiscal policy. The Gini index, which is designed to measure inequality within each group, cannot be used to decompose income inequality between groups. To address this second measure of inequality, we employ the Theil index decomposition method as outlined in Liao (2016). Our results indicate that the primary source of inequality stems from income disparities within each group rather than differences between groups (Table 4). The Gini index and Theil index both show a decline in income inequality for most groups when moving from market income to final income, according to the CEQ application used in this research (Table 3, Panels a and b). Table Table3. Inequality 3 EvolutionIndexes for different of the household designation types by type ofand income concepts household and income -2019- Panel "a" Gini Index (Ordered from highest to lowest, by market income gini) Market Plus pension Net Disposable Consumable Final Male-headed households 0.425 0.415 0.407 0.403 0.398 0.356 Male-headed household and main breadwinner with no remittances 0.412 0.412 0.404 0.402 0.398 0.375 Total population 0.411 0.403 0.395 0.391 0.385 0.344 Female-headed household and main breadwinner with no remittances 0.403 0.411 0.401 0.396 0.392 0.343 Female-headed household and main breadwinner 0.402 0.402 0.392 0.388 0.383 0.334 Male-headed household who contributes over half of the household's income 0.400 0.401 0.393 0.390 0.386 0.342 Male-headed household with a partner and at least one young child 0.395 0.394 0.383 0.378 0.371 0.315 Female-headed household who contributes over half of the household's income 0.394 0.392 0.382 0.378 0.373 0.331 Male-headed household and main breadwinner 0.394 0.389 0.380 0.378 0.376 0.352 Female-headed households 0.387 0.381 0.373 0.369 0.363 0.321 Female-headed household, sole provider with at least one young child 0.379 0.374 0.366 0.361 0.355 0.310 Female-headed with dependents 0.341 0.341 0.331 0.323 0.316 0.243 Male-headed with dependents 0.264 0.266 0.251 0.247 0.244 0.204 Panel "b" Theil Index (Ordered from highest to lowest, based on Theil Index of market income) Male-headed households 0.329 0.332 0.324 0.319 0.311 0.252 Male-headed with dependents 0.307 0.315 0.308 0.302 0.292 0.228 Male-headed household and main breadwinner with no remittances 0.312 0.310 0.299 0.296 0.289 0.256 Total population 0.304 0.310 0.301 0.296 0.288 0.232 Male-headed household with a partner and at least one young child 0.298 0.297 0.286 0.281 0.270 0.202 Male-headed household who contributes over half of the household's income 0.296 0.302 0.294 0.29 0.28 0.23 Female-headed household and main breadwinner with no remittances 0.288 0.299 0.284 0.278 0.270 0.210 Male-headed household and main breadwinner 0.283 0.281 0.272 0.269 0.264 0.233 Female-headed household and main breadwinner 0.281 0.288 0.275 0.270 0.263 0.206 Female-headed household who contributes over half of the household's income 0.270 0.275 0.263 0.258 0.252 0.201 Female-headed households 0.261 0.270 0.262 0.257 0.250 0.199 Female-headed with dependents 0.248 0.258 0.249 0.244 0.236 0.184 Female-headed household, sole provider with at least one young child 0.197 0.197 0.184 0.176 0.168 0.109 Source: Own estimates using EHPM of El Salvador 2019 Notes: Children and the elderly are considered dependent household members. Young children are considered as younger than 6 years old. 17 TableTable 4 Evolution 4. Theil Index the Theil of Market Decomposition, vs Final Decomposition Index Income Income Market income Final income Panel "A" classification according to demographics Total 0.317832 0.239548 Female-headed households 0.265702 0.198579 Male-headed households 0.347742 0.263466 Between groups 0.000018 0.01% 0.000002 0.00% Within each group 0.317813 99.99% 0.239546 100.00% Total 0.285957 0.208148 Female-headed with dependents 0.24305 0.175139 Male-headed with dependents 0.312808 0.228768 Between groups 0.00017 0.06% 0.000223 0.11% Within each group 0.285787 99.94% 0.207925 99.89% Total 0.315578 0.209895 Female-headed household, sole provider with at least one young child 0.193896 0.109877 Male-headed household, sole provider with at least one young child 0.332106 0.225352 Between groups 0.000813 0.26% 0.000089 0.04% Within each group 0.314765 99.74% 0.209806 99.96% Panel "B" classification according to labor income Total 0.304071 0.235783 Female-headed household and main breadwinner 0.286452 0.208434 Male-headed household and main breadwinner 0.309598 0.258453 Between groups 0.00658 2.16% 0.004096 1.74% Within each group 0.297491 97.84% 0.231687 98.26% Total 0.306248 0.231987 Female-headed household who contributes over half of the household's income 0.285527 0.212678 Male-headed household who contributes over half of the household's income 0.317916 0.242644 Between groups 0.000784 0.26% 0.000907 0.39% Within each group 0.305464 99.74% 0.23108 99.61% Total 0.33 0.255599 Female-headed household and main breadwinner, with no remittances 0.292895 0.212168 Male-headed household and main breadwinner, with no remittances 0.34007 0.282963 Between groups 0.013011 3.94% 0.008395 3.28% Within each group 0.316989 96.06% 0.247204 96.72% Source: Own estimates using EHPM of El Salvador 2019 Note: The total Theil index of each pair of groups was estimated by considering only those included in each definition. In addition to inequality indicators, we examined the differences in average per capita income of female versus male households, both before and after fiscal policy. To assess the discrepancy in average per capita income between female and male households, Figure 8 illustrates the evolution of the ratio between household income (average or median) per capita for female and male households, where a ratio of 1 would imply general equality. Panel a shows the ratio of the averages, while Panel b shows the ratio of the medians, as the mean is more sensitive to extreme values. 18 In some groups, female households receive more per capita income on average, while in others, they receive less than male households. Both the mean and median indicate that in households where women contribute more than men to the household budget, as well as in households where there are women heads with dependents, the average income of women is above that of comparable men, as the indicator or ratio is located above the unit line. However, in households where women are the sole providers and do not receive remittances, women’s average income is below men’s. Fiscal policy does not seem to have an effect on relative incomes when comparing male-headed and female-headed households. Both the mean and median show that there is no significant influence of fiscal policy on income differences by gender, as the ratio of the average or median income of female- headed households and male-headed households remains approximately the same and very close to unity. Despite the implementation of fiscal policies, significant income gaps persist between male and female households in certain groups, both before and after policy interventions. Specifically, households where the woman is the sole breadwinner (with or without remittances) and households headed by single women with children under six years of age exhibit notable income gaps against women. Analyzing the ratios between median income showed that these three cases exhibit gender income gaps, as the ratios were below the unit line before fiscal policy. Furthermore, fiscal policy does not contribute substantially to reducing or reversing these gaps. 19 Figure 8. Ratio of per capita income (average or median) of female households among male households (=1 implies equality of income between women and men in that type of household; less than 1 implies gender gap) Panel "A" Female-to-male mean income ratio, by CEQ income concepts and 1.2 household typologies 1.0 0.8 0.6 0.4 0.2 0.0 Market Income Plus pensions Net Disposable Consumable Female / Male Head of household Head of household with dependents Head of household with children under 6 Sole provider Greater contribution to the budget Only provider without remittances Equality Panel "B" Female-to-male median income ratio, by CEQ income 1.2 concepts and household typologies 1.1 1.0 0.9 0.8 0.7 0.6 Market Income Plus pensions Net Disposable Consumable Female / Male Head of household Head of household with dependents Head of household with children under 6 Sole provider Greater contribution to the budget Only provider without remittances Equality Source: Own estimates based on the EHPM 2019. 20 Impact of fiscal policy on the gender gap between ‘female’ and ‘male’ households In El Salvador, while fiscal policy reduces inequality, it contributes to an increase in poverty, making it not pro-poor. To assess the impact of fiscal policy on poverty in El Salvador, we calculate poverty rates based on market income using various poverty lines and compare them with poverty rates based on consumable income, which takes into account taxes paid and direct and indirect transfers received. We exclude monetized expenditures on education and health from our analysis, as they are not close cash substitutes. Our findings show that using the US$ 6.85 2017 PPP line, poverty increases from 23.1 percent to 26.2 percent due to fiscal policy. However, the impact is slightly different when a poverty line of US$ 3.65 per day 2017 PPP is used, as the poverty rate decreases slightly from 7.2 percent to 7.4 percent.12 These results indicate that fiscal policy is not pro-poor, which is consistent with previous research (Oliva 2020, 2015; Robayo-Abril and Barroso, 2022). Appendix 2 of this paper provides poverty measurements for each group analyzed, including indicators for poverty rate and gap and severity of poverty, along with statistical significance indicators for all changes in the poverty rates. Fiscal policy leads to a similarly significant increase in poverty for both female-headed households and those with male heads. Using the poverty line of US$ 6.85 per day at PPP, poverty increases for female-headed households by 3.1 percentage points, from 21.1 percent to 24.2 percent, comparable to the overall increase and the one observed among male-headed households (Figure 9). This change is statistically significant. In addition, fiscal policy also has a significant impact on the poverty of households headed by women with dependents (elderly or children). As shown in Figure 10, poverty for female-headed households with dependents increases from 21.7 percent to 25.1 percent with fiscal policy. According to Table 2, this group of households is particularly relevant as they comprise about one-third of households in poverty, with incomes below US$ 6.85 per day at PPP. Meanwhile, there is also an increase in households with a male head and dependents, with poverty increasing from 25.7 percent to 29.2 percent (Figure 10). According to Table 2, this group represents 57.4 percent of households in poverty using a poverty line of US$ 6.85 per day at PPP. 12This methodological aspect is related to the consideration of education and health as goods that cannot be replaced by cash, and therefore not comparable to income. 21 Figure 9. Poverty rate using a poverty line of US$ 6.85 PPP Figure 10. Poverty rate using a poverty line of US$ 6.85 PPP (percentage), by gender of the head of household (percentage), by gender of the head of household with dependents (older adults and children) Poverty Rate Poverty 35 35 Rate 30 29.2 27.29 30 24.13 24.25 24.22 25.7 24.9 25.9 25.1 25.1 25 23.36 23.51 21.12 20.96 25 21.7 20.37 20.10 21.0 21.6 20.8 20 20 15 15 10 10 5 5 0 Market Plus Net Disposable Consumable 0 Income pensions Market Plus pensions Net Disposable Consumable Income Female-headed with dependents (older adults and children) Female-headed households Men-headed households Men-headed with dependents (older adults and children) Source: Own estimates with data from the EHPM 2019. By examining different groups of female-headed households, it becomes apparent that lone female- headed households with young children under six years of age experience a much higher poverty rate than the general population; moreover, they also experience a significant increase in poverty after fiscal policy. The poverty rate for households headed by females with at least one child under six years of age increased by 4.3 percentage points after fiscal policy, from 38.4 percent to 42.7 percent (Figure 11). These changes are statistically significant. This group represents 3.3 percent of households in poverty living on less than US$ 6.85 per day. It is important to note that the group of male-headed households without a partner with at least one son or daughter is found to be insignificant based on the data presented in Table 2 (0.1 percent). The results reveal that fiscal policy has a differential impact on poverty among households where women provide most of the family budget compared to those where men do. Specifically, for households with incomes less than US$ 6.85 per day, where women are the main contributors to the budget, poverty increases by 3.9 percentage points, rising from 26.3 percent to 30.2 percent. Meanwhile, in households where men are the primary contributors, poverty increases by 4.2 percentage points, from 23 percent to 27.2 percent (Figure 12). These increases are both statistically significant. 22 Figure 11. Poverty rate using a poverty line of Figure 12. Poverty rate using a poverty line of US$ 6.85 PPP; US$ 6.85 PPP; female-headed household of women contributions to family income by gender of the head of who are sole providers with at least one young child household Source: Own estimates with data from the EHPM 2019. Male-headed households who are sole providers with young children are omitted, as they only represent 0.1 percent of households in the country. A critical finding of this study is that fiscal policy has a greater impact on increasing poverty among households with a sole female breadwinner than among households with a sole male provider, which widens gender gaps. Among households with a sole female breadwinner, the poverty rate increases by 4.3 percentage points (from 22.3 percent to 26.6 percent), whereas households with the man as the sole provider experience a smaller increase of 2.3 percentage points (from 15.5 percent to 17.8 percent) (Figure 13). These changes are statistically significant, and the poverty gap between the two household types increases from around 6.8 percentage points to 8.8 percentage points. These results suggest that women in households with a female sole breadwinner are disadvantaged both in the labor market and in terms of fiscal policy. In terms of fiscal policy, the inequalities arising from the labor market are not being compensated; as shown in Figure 8, the income from the labor market of this group of women is lower before fiscal policy. Furthermore, receiving transfers and making tax payments leads to an increase in poverty. It is also important to consider the role of remittances. While some households with female sole breadwinners receive remittances, which does not fully conform to the definition of a sole breadwinner, it is important to analyze what happens when they do not receive remittances. Our results show that households with a female sole breadwinner without remittances also experienced a substantial increase in poverty (4.1 percentage points), 23 especially compared to corresponding male-headed households (2.4 percentage points). In this case, the poverty gap between the two types of households widens (Figure 14).13 Figure 13. Poverty rate using poverty line of US$ 6.85 PPP; Figure 14. Poverty rate using a poverty line of US$ 6.85 female and male heads who are the sole provider of the PPP; female and male heads who are the sole providers household of the household and do not receive remittances Source: Own estimates with data from the EHPM 2019. In conclusion, our study finds that fiscal policy increases the incidence of poverty and expands gender gaps for those who are the sole providers of the household and those who do not have a partner but have children under six years of age. In addition, there is a more significant increase in the poverty rate of women who are the sole providers, with a rise of 4.3 percentage points, which widens the poverty gap between the two groups. In addition, the poverty rate of the group of single mothers with small children under six years of age is already high (38.4 percent), and fiscal policy increases it by 4.3 percentage points. 13For a more detailed analysis of poverty rates, gaps and severity, please refer to Appendix 2 of this document. Additionally, Appendix 3 presents cumulative income curves for household types, where the labor market plays a gender-differentiating role. In these cases, the consumable income curve, or the situation after fiscal policy in the case of female-headed households, is in a more unfavourable position compared to the market income position or before fiscal policy. This quantifiable fact illustrates that fiscal policy has a counterproductive effect in these groups, increasing poverty and the gender gap, particularly using a poverty line of US$ 6.85 per day at PPP. A more accurate category to describe a specific group of women is those who live in households where they are the only ones who contribute income to the household and also do not receive remittances. At least 8.4 percent of poor households below the US$ 3.65 PPP line and 10.9 percent with incomes below the US$ 6.85 PPP line have women who are the only providers and do not receive remittances. 24 Impact of fiscal policy on mobility between ‘female’ and ‘male’ household income groups Another relevant aspect of the analysis is the influence of fiscal policy on mobility between income groups. Up to this point, fiscal policy increases poverty in net terms, considering an income threshold of US$ 6.85 per day PPP; that is, the net increase in poverty reflects inflows and outflows of poverty since some households experience income increases and others experience income decreases. Despite household surveys being instruments that usually suffer from underreporting information from extremely high-income strata, it is interesting to determine mobility to and from other income groups, particularly transitions from and to the vulnerable stratum and the middle class. The World Bank classifies households in the Latin America region living with incomes between US$ 6.85 per day PPP and US$ 14 per day in 2017 PPP as vulnerable, and those between US$ 14 per day PPP and US$ 81 per day, as middle-class. In particular, for El Salvador, the category of ‘vulnerability’ is relevant, given that, according to World Bank statistics, 41.1 percent of the population in 2019 was within these thresholds, which constituted the highest percentage in Latin America, followed by the Dominican Republic. Our results show that fiscal policy led to an expansion of the ‘vulnerable’ and a reduction of the ‘middle class’. From a broader perspective, taking the transition matrix of the five strata and comparing the situation before and after fiscal policy, we found that the vulnerable stratum stands at 39.4 percent after fiscal policy. At the same time, this percentage reaches 38.4 percent before government action. On the other hand, the middle class represents 28.4 percent after fiscal policy, while it represents 32.3 percent before quantifying transfers and tax payments. Likewise, in the different panels of Table 5, in most cases, there is a permanence effect in the same stratum because the highest percentages correspond to the same strata of the transition matrix. Fiscal policy does not contribute significantly to upward income mobility. The transition matrix shows that some households experience upward mobility in the income distribution and others downwards. From the perspective of the former, although fiscal policy increases the incomes of some households, in most cases it is not enough to lift them out of poverty. Among groups that rise in income distribution due to fiscal policy, the most relevant change is the transition from poverty below US$ 3.65 a day to poverty below US$ 6.85 per day; however, there is minimal movement from poverty to the vulnerability stratum. For example, 12.4 percent of households headed by women with children under six years of age living in extreme poverty (income below US$ 2.15 per day PPP) transition to the higher income stratum (between US$ 2.15 and US$ 3.65 per day); that is, they move upward in the income distribution, but not enough to move to the vulnerable group. Similar results are observed among other female groups. Among households headed by women with children under six years of age in poverty (between US$ 2.15 and US$ 3.65 daily PPP), only 9.5 percent moved to a higher stratum (between US$ 3.65 and US$ 6.85 per day), but none to the vulnerable group. Finally, among households headed by women with children under six years of age in poverty (US$ 3.65–US$ 6.85 per day), only 1 percent moved to the vulnerable group. Among the other groups of female households, movements from poverty to vulnerability tend to be more significant but still limited. The most common transition for households moving downward in the income distribution is the movement from middle class to vulnerability, followed by the one from vulnerability to poverty . A higher percentage of households with women heads and women as sole providers move from the middle class to the vulnerable class 25 than from the vulnerable class to poverty (Table 5). However, among households with women and children under six, transitions are more significant from the vulnerable class to poverty, than from the middle class to vulnerability (14.5 percent and 13.7 percent, respectively). Up to 11 percent of the poor living with less than US$ 6.85 a day came from the vulnerable stratum. Table 5. Transition matrices before and after fiscal policy Panel "a" Transitions, all households Panel "b" Female-headed household, sole provider and with at least one young child After fiscal policy After fiscal policy <$2.15 $2.15- $3.65- Vulnerable Middle class Total <$2.15 $2.15- $3.65- Vulnerable Middle class Total per day $3.65 $6.85 ($6.85-$14 ($14-$81 per per day $3.65 $6.85 ($6.85-$14 ($14-$81 PPP per day per day per day PPP) day PPP) PPP per day per day per day per day PPP PPP PPP PPP PPP) PPP) <$2.15 per day PPP 74.3 12.3 6.3 5.3 1.7 100.0 <$2.15 per day PPP 87.6 12.4 0.0 0.0 0.0 100.0 $2.15-$3.65 per day PPP 2.5 87.0 7.2 2.7 0.6 100.0 $2.15-$3.65 per day PPP 3.5 87.0 9.5 0.0 0.0 100.0 Before fiscal policy Before fiscal policy $3.65-$6.85 per day PPP 0.0 4.5 90.5 3.9 1.1 100.0 $3.65-$6.85 per day PPP 0.0 1.5 97.5 1.0 0.0 100.0 Vulnerable ($6.85-$14 per Vulnerable ($6.85-$14 per day day PPP) 0.0 0.0 11.0 85.7 3.3 100.0 PPP) 0.0 0.0 14.5 85.5 0.0 100.0 Middle class ($14-$81 per day Middle class ($14-$81 per PPP) 0.0 0.0 0.0 13.7 86.3 100.0 day PPP) 0.0 0.0 0.0 16.9 83.1 100.0 Panel "c" Female-headed households Panel "d" Male-headed households After fiscal policy After fiscal policy <$2.15 $2.15- $3.65- Vulnerable Middle class Total <$2.15 $2.15- $3.65- Vulnerable Middle class Total per day $3.65 $6.85 ($6.85-$14 ($14-$81 per day $3.65 $6.85 ($6.85-$14 ($14-$81 per PPP per day per day per day per day PPP per day per day per day PPP) day PPP) PPP PPP PPP) PPP) PPP PPP <$2.15 per day PPP 77.9 10.4 5.7 4.4 1.6 100.0 <$2.15 per day PPP 64.9 17.4 8.0 7.7 1.9 100.0 $2.15-$3.65 per day PPP 2.3 87.6 6.1 3.4 0.6 100.0 Before fiscal policy $2.15-$3.65 per day PPP 3.0 85.8 9.4 1.2 0.7 100.0 $3.65-$6.85 per day PPP 0.0 5.0 90.8 3.5 0.8 100.0 Before fiscal policy Vulnerable ($6.85-$14 per day $3.65-$6.85 per day PPP 0.0 3.7 89.9 4.7 1.8 100.0 PPP) 0.0 0.0 12.4 84.6 3.0 100.0 Vulnerable ($6.85-$14 per Middle class ($14-$81 per day PPP) 0.0 0.0 0.0 18.0 82.0 100.0 day PPP) 0.0 0.0 8.8 87.5 3.8 100.0 Middle class ($14-$81 per day PPP) 0.0 0.0 0.0 14.9 85.1 100.0 Panel "f" Female-headed, sole provider Panel "g" Male-headed, sole provider After fiscal policy After fiscal policy <$2.15 $2.15- $3.65- Vulnerable Middle class Total <$2.15 $2.15- $3.65- Vulnerable Middle class Total per day $3.65 $6.85 ($6.85-$14 ($14-$81 per PPP per day per day per day PPP) day PPP) per day $3.65 $6.85 ($6.85-$14 ($14-$81 PPP PPP PPP per day per day per day per day PPP PPP PPP) PPP) <$2.15 per day PPP 69.9 21.4 8.7 0.0 0.0 100.0 $2.15-$3.65 per day PPP 5.6 83.7 9.9 0.9 0.0 100.0 <$2.15 per day PPP 69.1 28.8 2.1 0.0 0.0 100.0 $3.65-$6.85 per day PPP 0.0 2.3 94.9 2.4 0.3 100.0 $2.15-$3.65 per day PPP 1.4 93.1 4.6 1.0 0.0 100.0 Before fiscal policy Vulnerable ($6.85-$14 per $3.65-$6.85 per day PPP 0.0 4.6 86.8 8.6 0.0 100.0 Before fiscal policy day PPP) 0.0 0.0 8.8 88.0 3.2 100.0 Vulnerable ($6.85-$14 per day Middle class ($14-$81 per PPP) 0.0 0.0 6.3 87.0 6.7 100.0 day PPP) 0.0 0.0 0.0 11.5 88.5 100.0 Middle class ($14-$81 per day PPP) 0.0 0.0 0.0 9.8 90.2 100.0 Source: Own estimates based on the EHPM 2019. Poverty lines are based on the 2017 PPP poverty lines. Female-headed households are more likely to move from extreme to moderate poverty as a result of fiscal policy, but few of these households move out of poverty. According to Panel c of Table 5, 9.4 percent of female-headed households living with incomes between US$ 2.15 and US$ 3.65 per day 26 moved into a stratum with incomes between US$ 3.65 and US$ 6.85 per day; this share is more significant compared to the corresponding share among male-headed households (6.1 percent). Sole- provider mothers also experienced this more than single male providers and the general population (9.9 percent, 4.6 percent and 7.2 percent, respectively). V. What are the key aspects of progressive fiscal policy and horizontal equity in El Salvador from a gender perspective? This section describes the measures of progressivity and horizontal equity used for gender analysis and presents the progressive features of the different fiscal policy interventions, as well as their horizontal equity. Progressiveness The notions of progressivity or regressivity of taxes or transfers are not exempt from methodological discussions. A wide range of indicators summarizes the desired property of tax system progressivity. For example, transfers’ and tax payments’ progressivity can be measured in absolute or relative terms (a pro-poor characteristic). In absolute terms, a progressive expenditure (tax) is when the amount of the benefit (tax) declines (increases) with pre-transfer income. An expenditure (tax) is progressive in relative terms if it benefits (taxes) poorer households more (less) than wealthy ones relative to their income. All social programs are progressive in relative terms among all household groups, given that the share of benefits to income decreases with pre-transfer income. Figures 15 and 16 show the benefit share relative to income decreases when moving from poorer to wealthier income deciles. However, despite the relative progressivity, the limited fiscal resources allocated to such programs, low generosity and low coverage among the poor limit their ability to reduce poverty or inequality significantly. The share of social security contributions to income follows a progressive pattern. Figures 17 and 18 show that these contributions increase when moving from lower to higher income deciles. On the other hand, they seem more progressive for male-headed households since the shares show a more significant decrease compared to female-headed households. However, it is crucial to remember that the highest percentages of the working population not affiliated with social security occur in the lowest income deciles, which also explains this effect. On the other hand, indirect subsidies are more progressive in relative terms for female-headed households than for their male counterparts. For example, subsidies are more progressive in relative terms among female-headed households, given that the benefit share in relation to income is higher 27 for the poorest deciles and lower for the wealthiest deciles. However, progressivity is not preserved by looking at another typology. For example, in households where men contribute more than 50 percent of the household budget, the curve is flat and increases for the high deciles, suggesting less progressivity.14 The value added tax (VAT) tax burden on households is high throughout the income distribution; the results show that its incidence is high, reaching up to 10 percent of income. This is consistent with other studies (Oliva, 2020). This rate is also equated with the effective rate calculated through administrative records when obtaining the tax paid in relation to gross national income; however, a decrease in the percentage for low deciles is not notable, showing little regressivity in relative terms (Figures 19 and 20). Finally, spending on education and health is progressive overall. Still, they are even more progressive for households where women are the sole providers and contribute more than 50 percent of the household budget. Spending on education is more progressive in relative terms for women than for their male peers. For health spending, it is also slightly more progressive for women who contribute mainly to the household. Figure 15. Incidence of social programs, households with Figure 16. Incidence of social programs, female heads woman versus man as the sole provider providing more than 50 percent of household income versus male counterparts Source: Own estimates based on the EHPM 2019. 14 The fiscal incidence curves for all key fiscal interventions, for male and female households are presented in the Appendix. 28 Figure 17. Incidence of social security contributions, Figure 18. Incidence of social security contributions, male versus female-headed households households with woman versus man as the sole provider Source: Own estimates based on the EHPM 2019. Figure 19. Incidence of VAT, male- versus female- Figure 20. Incidence of VAT, households with women or headed households men as main providers Source: Own estimates based on the EHPM 2019. Relative progressivity and horizontal equity However, there are also more elements around the definition of progressivity to pay attention to , for example, the notions of horizontal equity (equal income, equal treatment) or vertical equity (different income, different treatment). Horizontal equity is defined as an equal treatment where a direct transfer is granted to women and men on the same terms or in identical amounts. 29 According to Grown and Valodia (2010), horizontal equity is insufficient to compensate women for previous conditions of inequality. Grown and Valodia (2010) note that, with horizontal equity, sufficient differentiation would not be achieved to pay for or compensate women for the underlying conditions that cause gender inequity before transfers. Greenspun (2019) defines “perfect” horizontal equity as program benefits and tax amounts that are allocated to the groups proportional to their shares in the population. For example, if 30 percent of households are headed by a woman and 70 percent by males, 30 percent of the transfer amount is allocated to female-headed households and 70 percent to male-headed households. However, Greenspun (2019) also establishes relative progressivity concerning gender if the household with a gender characteristic has a lower income and receives a higher percentage of the expenditure or transfer. Greenspun (2019) indicates that to fulfill this property, the percentage of direct transfer must be higher than the percentage that such a group receives of total income. For example, suppose female-headed households receive 40 percent of the transfer but receive 25 percent of total income. In that case, it could be said that there is relative progressivity from a gender perspective. For example, if the proportion of policy transfer is concentrated more among women or in some female-categorized households, and if, in that case, that group receives a smaller percentage of total income, the program could be considered relatively progressive in an unambiguous manner. The property of relative progressivity is satisfied in most social transfers in El Salvador, namely in- kind and monetary subsidies; however, the situation is different regarding VAT and health expenditure. Table 6 presents the proportion of the groups in the country’s total households and the proportion of each group in the total market income in general. Both indicators are located in the first rows. Each of the panels in Table 6 shows the situation of three groups in particular, female-headed households, households where women are the sole providers and households where they contribute more than 50 percent to the household budget. The proportion received from the amount allocated by the government in all fiscal interventions is higher than 36.2 percent of the total market income received by female-headed households. However, female-headed households contribute 37.7 percent of VAT and receive 35.2 percent of the overall resources allocated to health but receive only 36.2 percent of income (Panel a of Table 6). This differs from social security contributions, where these households contribute only 31.3 percent but receive 36.2 percent of all market income. Contributory pensions do not present horizontal equity among poor households (living below the US$ 3.65 line) headed by women; education benefits and social transfers results are similar. About 24.9 percent of the contributory pension benefits reach households headed by women, less than the share of poor households headed by a female (31.2 percent). On the other hand, poor households (using the US$ 3.65 line) headed by women receive 37.9 percent of indirect subsidies, slightly higher than the 31.2 percent that the group represents in the population, indicating a lack of horizontal equity. 30 Likewise, female-headed households receive 34.9 percent of overall education benefits, which exceeds the share of this group among the poor (using the line of US$ 3.65), 31.2 percent. Similarly, poor households with female heads receive 34.6 percent of social transfers, which exceeds the percentage of this poverty group (Table 6, Panel a). In the case of the VAT, there is no horizontal equity among the poor (using the US$ 3.65 line). Female- headed households contribute 38 percent of overall VAT collection among the poor (using the US$ 3.65 line), higher than 31.2 percent of poor households (living on less than US$ 3.65 per day) headed by women. There is no horizontal equity for contributory pensions among households with women as the only providers (Table 6, Panel b). Poor households (using the US$ 3.65 line) with women as the only providers receive 66.4 percent of the overall benefits, but they only represent 73.3 percent of the poor, so horizontal equity does not hold. When looking at indirect subsidies, social transfers and health, these households receive 78 percent, 74 percent and 72.4 percent of the benefits among the poor, respectively, but they represent 73.3 percent of the poor. Among households supported by women, most of the fiscal interventions are progressive in relative terms. Table 6, Panel b shows that the percentage of the resources devoted to the programs is higher than the percentage that households supported by women receive of total (relative) income. They receive 67.8 percent of transfers for programs, 67.5 percent of indirect subsidies, 78.7 percent of education and 55.6 percent of health, which exceeds 53 percent of the total market income the group receives. In the case of poor households (with incomes of less than US$ 3.65 per day) with women as the sole provider, transfer programs are practically horizontally equitable because the percentages of expenditure allocate similar proportions of market income. On the one hand, for this group, women providers receive 74.5 percent of the income and absorb up to 74.7 percent of the amount spent in the programs. On the other hand, this percentage exceeds what they received in the case of male households (25.3 percent). Table 6. Progressivity and Horizontal equity 31 Table 6 Progressivity and horizontal equity Panel "a" Panel "b" Panel "c" Female-headed with Male-headed with Female-headed with Male-headed with women contributing men contributing women as sole men as sole over half of the over half of the Female- headed Male- headed Total breadwinner breadwinner Total household budget household budget Total % over Population - Market income US$6.85 ppi 37.6% 62.4% 100.0% 55.0% 45.0% 100.0% 37.5% 62.5% 100.0% Total 37.1% 62.9% 100.0% 58.5% 41.5% 100.0% 36.9% 63.1% 100.0% % over- Market Revenue US$6.85 ppi 36.1% 63.9% 100.0% 51.8% 48.2% 100.0% 38.8% 61.2% 100.0% Total 36.2% 63.8% 100.0% 53.0% 47.0% 100.0% 38.5% 61.5% 100.0% % on the amount of the program Contributory pensions (treated as deferred income) US$6.85 ppi 41.3% 58.7% 100.0% 52.9% 47.1% 100.0% 46.4% 53.6% 100.0% Total 40.3% 59.7% 100.0% 53.6% 46.4% 100.0% 46.8% 53.2% 100.0% direct taxes US$6.85 ppi 27.4% 72.6% 100.0% 57.1% 42.9% 100.0% 39.7% 60.3% 100.0% Total 27.4% 72.6% 100.0% 57.1% 42.9% 100.0% 39.7% 60.3% 100.0% Social security contributions US$6.85 ppi 31.5% 68.5% 100.0% 49.2% 50.8% 100.0% 37.1% 62.9% 100.0% Total 31.3% 68.7% 100.0% 49.3% 50.7% 100.0% 36.6% 63.4% 100.0% 32 33 In the case of households where women are the sole providers, the relative progressivity in transfers channeled to the elderly does not seem to be evident. For example, 47.3 percent of non-contributory pensions channeled to households where the only provider is a woman when they receive 53% of the total income. When separating households between women and men who are the sole breadwinners, social security contributions look progressive in a relative sense, while VAT appears regressive. Social security contributions are progressive, with 49.3 percent of the overall contributions made by households where women are the sole providers. This is less than these household shares of income (53 percent). In comparison, households supported only by men contribute up to 50.7 percent. While households supported by women concentrate 53 percent of income, they contribute a larger share (59 percent) of the VAT collected (Table 6, Panel b). Finally, there is relative progressivity in non-contributory pensions, social transfers and education spending among households where women contribute more than 50 percent; this is not the case when looking at other fiscal interventions. These households receive 38.5 percent of total income, 46.8 percent of non-contributory pensions, 39.3 percent of programs and 40.4 percent of education (Table 6, Panel c). Among households where women contribute more than 50 percent, health and non- contributory pension expenses are relatively regressive since they receive 37.1 percent and 37.3 percent, respectively, lower than the 38.5 percent they concentrate on total income. Regarding subsidies, they appear with a similar percentage (38.4 percent). This type of household contributes less to social security but more to VAT. Among households where women contribute more than 50 percent to the budget, social security taxes reached 37.3 percent, less than the 38.5 percent they represent of total income; however, their payments represent 40.6 percent of the total VAT collected (Table 6, Panel c). In conclusion, relative progressivity or horizontal equity is a necessary, but not sufficient, property for the fiscal system to also be pro-poor. As noted, although there is relative progressivity among households where women are the sole provider, fiscal policy led to a rise in the poverty rates among this group. VI. What are the marginal impacts of the fiscal interventions on the gender gaps? In this section, we review the marginal contribution of each fiscal intervention to poverty and inequality reductions and the respective coverage in the population. 34 Marginal contributions The coverage, progressivity and generosity of benefits are intimately linked to their ability to reduce poverty. In practice, the ability of the program or fiscal policy in reducing poverty depends on three factors: (1) good targeting of the poorest, i.e. that the benefit actually reached the poor; (2) good coverage of the poor, i.e. that the beneficiaries of the resources represent a significant proportion of the target population; and (3) the adequacy of benefits, i.e. or how generous is the benefit relative to household income. For example, if the benefit reaches the target poor population, resources are widely concentrated among the poor, and the amount of benefit is equal to or greater than the depth or severity of poverty among those groups of households, then the program or transfers will be effective in reducing poverty. Direct transfers may be designed to reach the poor in a very targeted manner, with high coverage among poor or most vulnerable groups; however, if generosity or resources for this group are limited relative to the depth of poverty, the impact on poverty reduction will be small. On the other hand, it is possible to allocate a large amount of resources to various segments of the population. Still, if they are weakly targeted, the effect will also be small. This is more complex when households receive several types of transfers, some targeted and others weakly targeted. We use the concept of marginal contribution to poverty and inequality developed in the CEQ methodology, which summarizes all the previous aspects in a single measure. The definition of marginal contribution is equivalent to distinguishing whether poverty is greater, equal or lesser compared to the respective level in absence of intervention, taking the system as a whole (Lustig and Higgins, 2018). The marginal contribution is the difference between, for example, the poverty rate (or another measure of inequality) of income without fiscal policy or fiscal intervention of interest but with all other interventions, minus the poverty rate of post-fiscal income, including all interventions. This concept establishes the influence of each intervention on the income distribution by implicitly incorporating both the concentration or deconcentration of the resources allocated – which can be measured with concentration indices – and the amount of the resources granted, that is, the size of the program itself. Education and health expenditures present the highest marginal contribution to poverty among all household typologies. Figures 21, 22 and 23 present the results of the marginal contributions to poverty for the different types of households. The results include the contributions of in-kind transfers (education and health) and help identify which components of fiscal policy may be relevant to reducing or widening gender gaps. However, the effect of health and education on poverty may be overestimated; they are also not substitutable for cash. Education and health present the most remarkable marginal contributions among the household types. Also, at the other extreme, VAT and 35 social security contributions increase poverty significantly, as reflected in negative marginal contributions, largely due to their large size and heavy burden on the poor. Marginal contributions of poverty reduction programs are higher among households where women are the sole providers than households where men are the sole providers. It is relevant how the marginal contributions of most poverty reduction programs are similar between female-headed and male-headed households (Figure 21); however, poverty reduction contributions are lower for households with men as sole providers than for households with only female providers (Figure 22). However, indirect taxes increase poverty among households where women are the sole providers compared to their male counterparts. In other words, poverty decreases overall because indirect taxes exceed the poverty-reducing effect of other programs. Contributory pensions stand out as a fiscal intervention with a significant marginal effect on poverty reduction, both for female- and male-headed households (Figure 21). This effect is related to the coverage of the poorest quintiles of the income distribution due to the minimum pension guarantee defined in the law. Since 2020, the minimum old-age and partial disability pensions have increased by 46.5 percent and 44.5 percent, respectively. With this increase, the monthly benefits increased from US$ 207.6 to US$ 304.1 and US$ 145.32 to US$ 210.00, respectively. Direct transfers also reduce poverty, but their impact is limited; health and education help reduce gender gaps, while indirect taxes place a larger weight on female households. These transfers include the uniform and food packages, the agricultural package and non-contributory pensions. Its effect on reducing poverty is related to poor households’ access to, for example, education and the quality of information used in programs to benefit poor households. As noted below, although transfers have reasonably high coverage, the size of the programs is small, resulting in a small poverty reduction effect. Health and education expenditures have an equalizing effect between female and male households. On the contrary, indirect taxes such as VAT tend to represent a heavier burden on female-headed households; this unintended effect must always be taken into account in the design of public policy. 36 Figure 21. Marginal contributions to poverty reduction (US$ 6.85 a day) by gender of the household head (US$ 6.85 PPP) Marginal effect to poverty (line US$ 6.85) Education and health Education Basic education Health Ministry of Health secondary education Preschool Education Salvadoran Social Security Institute contributory pensions Water Subsidy Uniform and utility pack Gas Subsidy school feeding Subsidy for public transport Lower threshold electricity subsidy Non-contributory pensions Health system for the armed forces Solidarity Communities Education Voucher Tertiary education Solidarity Communities health voucher Bonds (Both) of Solidarity Communities bean packet corn package Direct Taxes Higher Threshold Electricity Subsidy gasoline tax magisterial welfare Contributions to Social Security (Employees) Contributions to Social Security (Employers) Value Added Tax -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1 0.12 Female-headed households Men-headed households Source: El Salvador’s 2019 EHPM estimates. 37 Figure 22. Marginal contributions to poverty reduction (US$ 6.85 per day) by the gender of the household ’s main provider Marginal effect to poverty (line US$ 6.85) Education and health Education Basic education Health Ministry of Health secondary education Preschool Education Salvadoran Social Security Institute Uniform and utility pack Water Subsidy contributory pensions school feeding Subsidy for public transport Gas Subsidy Lower threshold electricity subsidy Bonds (Both) of Solidarity Communities Solidarity Communities Education Voucher Health system for the armed forces Solidarity Communities health voucher Non-contributory pensions bean packet corn package Direct Taxes Higher Threshold Electricity Subsidy gasoline tax magisterial welfare Tertiary education Contributions to Social Security (Employees) Contributions to Social Security (Employers) Value Added Tax -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1 0.12 Female-headed where she is the sole breadwinner Men-headed where he is the sole breadwinner Source: El Salvador’s 2019 EHPM estimates. 38 Figure 23. Marginal contributions to poverty reduction (US$ 6.85 PPP) by gender of the primary income earner Marginal effect to poverty (line US$ 6.85) Education and health Education Basic education Health Ministry of Health secondary education Preschool education Salvadoran Social Security Institute Uniform and utility pack water subsidy Public transport subsidy school feeding Gas Subsidy Lower threshold electricity subsidy Bonds (Both) of Solidarity Communities Solidarity Communities Education Bonus Health system for the armed forces Tertiary education corn package Non-contributory pensions Solidarity Communities health bonus bean packet direct taxes Higher Threshold Electricity Subsidy gas tax magisterial welfare Contributions to Social Security (Employees) Contributions to Social Security (Employers) Value Added Tax -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1 0.12 Female-headed who contributes over half of the household budget Men-headed who contributes over half of the household budget Source: El Salvador’s 2019 EHPM estimates. These marginal contributions to poverty reduction of each element of the tax system are consistent with the changes in the poverty rate observed among the different income types considered in the CEQ methodology. Figures 24, 25 and 26 show the change between market plus pensions, net, available and consumable income for different types of households that result after the net fiscal system is considered. In the case of the population living in female-headed households and where females are the only providers, a decrease of 0.7 and 0.5 percentage points in the transition from 39 market income to income plus contributory pensions is observed (Figures 24 and 25). Among the population where the head is a woman and of women as sole providers, there is an increase in poverty of 0.6 and 0.4 percentage points between market income plus pensions and net income (Figures 24 and 25). However, the increase for women with minor children is greater, reaching 1.7 percentage points, reflecting mostly the action of social security contributions (Figure 26). In the relevant cases from the gender perspective, a poverty reduction effect is again obtained, with the accounting of direct transfers between net income and disposable income; however, it stands out that the effect is smaller – reaching 1.2 percent – for the group of households where women are the only providers (Figure 25). Finally, a rise in poverty during the passage between disposable income and consumable income is registered, which implies that indirect subsidies are not enough, both in terms of depth and the amount, to compensate for the payment of indirect taxes where the VAT stands out; up to 4.5 percentage points for female-headed households with children under six years of age and 4.1 percentage points for households headed by women (Figures 24 and 26). In this sense, a more significant increase in poverty stands out for the population with households with mothers without a partner who are the only providers (5.6 percentage points) (Figure 25). Figure 24. Changes in the poverty rate among female-headed households (US$ 6.85 PPP) 5 4.1 4 Percentage points 3 2 0.6 1 0 -1 -2 -0.7 -0.9 Plus pensions Net Disposable Consumable Change from market income to each of the income types Figure 25. Changes in poverty among households with women as the sole providers (US$ 6.85 PPP) 5.6 6 Percentage points 5 4 3 2 1 0.4 0 -1 -2 -0.5 -1.2 Plus pensions Net Disposable Consumable Change from market income to each of the income types 40 Figure 26. Changes in poverty rate (US$ 6.85 PPP) among households with women without a partner and at least one child under six 5 4.5 4 Percentage points 3 1.7 2 1 0 -1 -0.1 -0.5 Plus pensions Net Disposable Consumable Change from market income to each of the income types Source: Population-based estimates using El Salvador’s EHPM 2019. Coverage Direct transfers reduce poverty more than other fiscal interventions mostly due to their high coverage among the poor rather than their size. Table 7 (Panels a, b and c) show that the coverage of transfers from social programs among households living on less than US$ 6.85 per day is high, around 62 percent for female-headed households, 69 percent for households with women as the sole provider and 72.6 percent for households with women contributing more than 50 percent to household income. On the other hand, the percentages of coverage are also high for their male counterparts, where coverage of the poor is close to 70 percent. However, the main poverty-targeted conditional cash transfers have structurally low coverage of the poor, both for female and male households. Table 7 (Panels a, b and c) show that coverage of the poor (US$ 3.65) is minimal, close to 2.6 percent among female-headed households, 1.6 percent among female-headed households, 1.6 percent among women as the sole provider and up to 3.4 percent among households with women contributing more than 50 percent to household income. Likewise, this coverage is low, although slightly higher, among the respective male households, reaching 3.6 percent and 3.8 percent, respectively (see Table 7). Finally, there is high coverage of indirect subsidies among households living in poverty (US$ 6.85). Coverage of the poor is close to 82.5 percent among female-headed households, 84.8 percent among households with women as the sole provider and 85.2 percent among households with women who contribute more than 50 percent to household income. This is related to the intensity of the use of indirect services subsidized by the State across the income distribution. Secondary-level dropouts are also notable. The decrease in education coverage is remarkable in the transition from basic education to secondary education; among female-headed households, coverage decreases from 50 percent to 9.6 percent; in households with a woman as the sole provider, it falls 41 from 57 percent to 11 percent, and in households where women contribute more than 50 percent of the household budget, coverage decreases from 59 percent to 11 percent. In general, conditional cash transfers, such as school-feeding programs, have positive but small impacts on poverty reduction because the resources allocated in these areas are scarce (Figures 21, 22 and 23). Table 7. Coverage Rates, Tax and transfers (direct and in-kind) (Percentage of households that receive the benefit or pay the tax) Panel "a" Difference Female- headed Male- headed 50% of the household income Female-headed household Male-headed household Women sole provider Men sole provider Female head of household provides > 50% of the household income Women sole provider Men sole provider Male head of household provides > 50% of the household income Female-headed household Male-headed household Male head of household provides > 50% of the household income Fig. 4 Programs or transfers Fig. 5 Programs or transfers Fig. 6 Programs or transfers Fig. 4 Programs or transfers Fig. 5 Programs or transfers Fig. 6 Programs or transfers 6.0% 6.0% 6.0% of income of income of income 6.0% 6.0% 5.0% 6.0% of income of income 5.0% of income 5.0% 4.0% 4.0% 5.0% 4.0% 4.0% 4.0% 4.0% 3.0% Percentage Percentage 3.0% 2.0% Percentage 3.0% 2.0% Percentage Percentage 3.0% 2.0% 2.0% Percentage 2.0% 2.0% 1.0% 1.0% 0.0% 1.0% 1.0% 0.0% 0.0% 1 2 3 4 5 6 7 8 9 10 0.0% 0.0% 1 2 3 4 5 6 7 8 9 10 0.0% 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 Deciles 6 7 8 9 10 1 2 3 4 5 Deciles 6 7 8 9 10 1 2 3 4 5 Deciles 6 7 8 9 10 Deciles Deciles Deciles Female head of household provides > 50% of the household income Female-headed household Male-headed household Women sole provider Men sole provider Female Male head head of household of household provides provides > 50% > 50% of the of the household household income income Female-headed household Male-headed household Women sole provider Men sole provider Male head of household provides > 50% of the household income Fig. 7 Social security contributions Fig. 8 Social security contributions Fig. 9 Social security contributions Fig. 7 Social security contributions 0.0% Fig. 8 Social security contributions 0.0% Fig. 9 Social security contributions 0.0% of income 0.0% 1 2 3 4 5 6 7 8 9 10 0.0% 1 2 3 4 5 6 7 8 9 10 of income 0.0% 1 2 3 4 5 6 7 8 9 10 -1.0% of income -1.0% of income -1.0% 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 of income 1 2 3 4 5 6 7 8 9 10 -1.0% of income -1.0% -1.0% -2.0% -2.0% -2.0% -2.0% Percentage -2.0% -3.0% Percentage -2.0% Percentage Percentage -3.0% -3.0% -3.0% Percentage Percentage -3.0% -3.0% -4.0% -4.0% -4.0% -4.0% -4.0% -5.0% -4.0% Deciles -5.0% -5.0% -5.0% Deciles Deciles Deciles -5.0% -5.0% Deciles Deciles Female head of household provides > 50% of the household income Female-headed household Male-headed household Women sole provider Men sole provider Female Male head head of household of household provides provides > 50% > 50% of the of the household household income income Female-headed household Male-headed household Women sole provider Men sole provider Male head of household provides > 50% of the household income Fig.10 Fig. 11 Fig. 12 Fig.10 Indirect subsidies Fig. Indirect 11 subsidies Fig. Indirect 12 subsidies Indirect subsidies Indirect subsidies Indirect subsidies 6.0% 6.0% 6.0% of income of income of income 6.0% 6.0% 6.0% of income of income 4.0% of income 4.0% 4.0% 4.0% 1.0% 2.0% 2.0% 1.0% Percentage Percentage Percentage 2.0% 2.0% 1 2 3 4 5 6 7 8 9 10 Percentage Percentage 0.0% 0.0% Percentage -4.0% 1 2 3 4 5 6 7 8 9 10 0.0% 1 2 3 4 5 6 7 8 9 10 0.0% 1 2 3 4 5 6 7 8 9 10 Deciles -4.0% 1 2 3 4 5 Deciles 6 7 8 9 10 1 2 3 4 5 Deciles 6 7 8 9 10 Deciles Deciles Deciles Female head of household provides > 50% of the household income Female-headed household Male-headed household Women sole provider Men sole provider Female Male head head of household of household provides provides > 50% > 50% of the of the household household income income Female-headed household Male-headed household Women sole provider Men sole provider Male head of household provides > 50% of the household income 60 Fig. 13 Value Added Tax Fig. 14 Value Added Tax Fig. 15 Value Added Tax Fig. 13 Value Added Tax 0.0% Fig. 14 Value Added Tax 0.0% Fig. 15 Value Added Tax 0.0% of income 0.0% -2.0% 1 2 3 4 5 6 7 8 9 10 0.0% 1 2 3 4 5 6 7 8 9 10 of income 0.0% -2.0% 1 2 3 4 5 6 7 8 9 10 of income of income -2.0% 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 of income -2.0% 1 2 3 4 5 6 7 8 9 10 -4.0% of income -4.0% -5.0% -4.0% -4.0% -6.0% -5.0% -6.0% Percentage Percentage -6.0% Percentage -6.0% -8.0% Percentage -8.0% -10.0% Percentage Percentage -8.0% -8.0% -10.0% -10.0% -10.0% -10.0% -10.0% -12.0% -12.0% -15.0% -12.0% -12.0% -15.0% Deciles -14.0% -14.0% Deciles Deciles Deciles -14.0% -14.0% Deciles Deciles Female head of household provides > 50% of the household income Female-headed household Male-headed household Women sole provider Men sole provider Female head of household provides > 50% of the household income Male head of household provides > 50% of the household income Female-headed household Male-headed household Women sole provider Men sole provider Male head of household provides > 50% of the household income Fig. 16 Education Fig. 17 Education Fig. 18 Education Fig. 16 Education Fig. 17 Education Fig. 18 Education 40.0% 40.0% 40.0% of income of income 40.0% 40.0% of income 40.0% 30.0% of income of income 30.0% of income 30.0% 30.0% 30.0% 30.0% 20.0% 20.0% 20.0% 20.0% 10.0% Percentage Percentage 20.0% 20.0% Percentage Percentage 10.0% 10.0% Percentage 10.0% 0.0% Percentage 10.0% 10.0% 0.0% 0.0% 1 2 3 4 5 6 7 8 9 10 0.0% 0.0% 0.0% 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 Deciles 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 Deciles 6 7 8 9 10 1 2 3 4 5 Deciles 6 7 8 9 10 Deciles Deciles Deciles Female head of household provides > 50% of the household income Women sole provider Men sole provider Female head of household provides > 50% of the household income Female-headed household Male-headed household Male head of household provides > 50% of the household income Female-headed household Male-headed household Women sole provider Men sole provider Male head of household provides > 50% of the household income Fig. 19 Health Fig. 20 Health Fig. 21 Health Fig. 19 Health Fig. 20 Health Fig. 21 Health 40.0% 40.0% 40.0% income income 40.0% 40.0% income 40.0% income income 30.0% income 30.0% 30.0% 30.0% 30.0% 30.0% of of of of 20.0% of of 20.0% Porcentage Percentage 20.0% 20.0% Percentage 20.0% Porcentage Percentage 20.0% 10.0% Percentage 10.0% 10.0% 10.0% 10.0% 10.0% 0.0% 0.0% 0.0% 1 2 3 4 5 6 7 8 9 10 0.0% 0.0% 0.0% 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Deciles 1 2 3 4 5 6 7 8 9 10 Deciles Deciles Deciles Deciles Deciles Female head of household provides > 50% of the household income Women sole provider Men sole provider Female head of household provides > 50% of the household income Female-headed household Male-headed household Women sole provider Men sole provider Male head of household provides > 50% of the household income Female-headed household Male-headed household Male head of household provides > 50% of the household income Source: Own estimates based on EHPM 2019. 61