JOBS WORKING PAPER Issue No. 28 Stubborn Gender Gaps in Paraguay’s Labor Market Elizabeth Ruppert Bulmer, Raquel Scarpari, and Adrian Garlati STUBBORN GENDER GAPS IN PARAGUAY’S LABOR MARKET Elizabeth Ruppert Bulmer, Raquel Scarpari, and Adrian Garlati Activities under the Let’s Work Partnership are supported by grants under the Jobs Umbrella Multidonor Trust Fund and/or IFC Let’s Work Multidonor Trust Fund. © 2019 International Bank for Reconstruction and Development / The World Bank. 1818 H Street NW, Washington, DC 20433, USA. Telephone: 202-473-1000; Internet: www.worldbank.org. Some rights reserved This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: Elizabeth Ruppert Bulmer, Raquel Scarpari, and Adrian Garlati. “Stubborn Gender Gaps in Paraguay’s Labor Market.” World Bank, Washington, DC. License: Creative Commons Attribution CC BY 3.0 IGO. Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third-party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Images: © World Bank. Further permission required for reuse. 2 ACKNOWLEDGEMENTS This report was written by Elizabeth Ruppert Bulmer (Lead Economist), Raquel Scarpari and Adrian Garlati (Consultants) of the Jobs Group. It was prepared under the leadership of Ian Walker (Manager, Jobs Group), Pablo Gottret (Practice Manager, Social Protection and Jobs Global Practice), and Michal Rutkowski (Senior Director, Social Protection and Jobs Global Practice). The team is grateful for comments provided by World Bank peer reviewers Bénédicte Leroy de la Brière (Lead Economist, Social Protection and Jobs Global Practice) and Aline Coudouel (Lead Economist, Social Protection and Jobs Global Practice), and by Gerhard Reinecke (International Labour Organisation, Santiago Office for the Southern Cone of Latin America). The team wishes to express its gratitude to the Government of Paraguay’s statistical agency Dirección General de Estadísticas, Encuestas y Censos (DGEEC) for its close collaboration and partnership on the firm census and survey data. The authors also owe thanks to the Secretaría Técnica de Planificación del Desarrollo Económico y Social and the Ministerio de Hacienda for their technical support, and to the Ministerio de Trabajo, Empleo y Seguridad Social for its effective partnership. This assessment of gender gaps in labor outcomes is part of the Let’s Work Paraguay program, and builds on two preceding and related reports: “Paraguay Jobs Diagnostic: The Dynamic Transformation of Employment in Paraguay”, Jobs Series Issue No. 10, Jobs Group, World Bank; and “Firm Productivity and Employment in Paraguay 2010-2014”, Jobs Working Paper Issue No. 19, by Elizabeth Ruppert Bulmer and Adrian Scutaru, Jobs Group, World Bank. Let’s Work Paraguay was made possible through a grant from the World Bank’s Jobs Umbrella Trust Fund, supported by the Department for International Development/UK AID, and the Governments of Norway, Germany, and Austria, as well as the Austrian Development Agency and the Swedish Development Agency (SIDA). Let’s Work partners include the African Development Bank Group (AfDB), Asian Development Bank Group (ADB), Austrian Federal Ministry of Finance (BMF), Department for International Development (DfID), European Investment Bank (EIB), European Development Finance Institutions (EDFIs), Inter-American Development Bank (IADB), International Labor Organization (ILO), International Youth Foundation (IYF), Islamic Corporation for Development of Private Sector (ICD), Ministry of Foreign Affairs of Netherlands, Overseas Development Institute (ODI), Private Infrastructure Development Group (PIDG), Swiss Secretariat for Economic Affairs (SECO), World Bank Group (WBG), and World Business Council for Sustainable Development (WBCSD). 3 CONTENTS ACKNOWLEDGEMENTS ....................................................................................................................... 3 1. OVERVIEW ................................................................................................................................... 5 2. SOCIO-DEMOGRAPHIC TRANSFORMATION AND FEMALE LABOR FORCE PARTICIPATION ............... 8 2.1. DEMOGRAPHIC TRANSFORMATION.................................................................................................... 8 2.2. FEMALE LABOR FORCE PARTICIPATION ............................................................................................ 10 3. GENDER GAPS IN LABOR MARKET OUTCOMES ............................................................................ 14 3.1. GENDER GAPS IN UNEMPLOYMENT.................................................................................................. 15 3.2. GENDER GAPS IN WORK STATUS....................................................................................................... 17 3.3. GENDER DIFFERENCES IN THE DISTRIBUTION ACROSS SECTORS...................................................... 21 3.4. GENDER GAPS IN JOB QUALITY AND EARNINGS ............................................................................... 25 4. LABOR DEMAND ........................................................................................................................ 33 4.1. SNAPSHOT OF PRIVATE SECTOR FIRMS............................................................................................. 33 4.2. GENDER DIFFERENCES IN FIRM-TYPE, FIRM-SIZE AND WAGES ........................................................ 34 4.3. GENDER DIFFERENCES IN SECTOR OF WORK .................................................................................... 38 4.4. DYNAMIC IMPLICATIONS FOR CLOSING GENDER GAPS IN JOB QUALITY ......................................... 42 5. CHALLENGES FOR THE FUTURE AND POLICY OPTIONS ................................................................. 44 5.1. ENHANCING WOMEN’S CAPACITY TO OBTAIN BETTER JOBS ........................................................... 45 5.2. REDUCING BARRIERS FACING WOMEN IN ACCESSING QUALITY EMPLOYMENT.............................. 49 5.3. INCREASING THE PRODUCTIVITY OF INFORMALLY EMPLOYED WOMEN ......................................... 54 5.4. STIMULATING JOB CREATION ATTRACTIVE TO WOMEN .................................................................. 57 REFERENCES ..................................................................................................................................... 59 ANNEX A – DATA SOURCES AND VARIABLE DEFINITIONS.................................................................... 63 ANNEX B – LABOR SUPPLY ANALYSIS DETAILS .................................................................................... 66 ANNEX C – LABOR DEMAND REGRESSIONS ........................................................................................ 92 4 1. OVERVIEW Paraguay has experienced robust economic growth in recent decades, leading to significant advances in development. Paraguay is a small, open economy of seven million inhabitants, bordered by the much larger economies of Brazil, Argentina and Uruguay. Paraguay is landlocked, and has abundant and rich land resources and a river system that provides hydroelectric power. Over the past 45 years, Paraguay’s GDP grew almost eightfold, and GDP per capita nearly tripled in real terms, reaching US$3,825 in 2015 (upper- middle-income country status). This strong economic growth was accompanied by significant improvements in living standards, although Paraguay’s rate of improvement lagged that of some of its neighbors in the region. Extreme poverty fell from 13 percent in 2003 to 5 percent in 2015, and moderate poverty fell from 51 percent to 27 percent in the same period (Figure 1). Real incomes of the bottom 40 percent increased steadily, and the Gini coefficient declined from 0.55 in 2003 to 0.48 in 2015. Important structural and demographic transformations have taken place in the past 15 years. The labor force expanded by 2.6 percent annually over the past decade, and economic growth was robust enough to create adequate employment to absorb new labor force entrants. The employment-to-GDP elasticity of 0.58 – in line with global norms – translated into an average 63,000 net new jobs annually between 2008 and 2015. Paraguay’s economy is in the midst of a structural transformation from an agricultural to a services-based economy. The majority of recent job growth was concentrated in retail (accounting for 45 percent of net new jobs) and government services (over 20 percent), followed by manufacturing, construction, finance and real estate and other services. Job creation was accompanied by strong labor productivity growth averaging 2.1 percent per year in real terms, most of which reflected within-sector productivity gains. Although average worker welfare has increased, not all workers benefited equally, with large observed differences between men and women. This note analyzes household survey data and firm-level data to measure gender gaps in employment outcomes over the past 15 years and shed light on the degree to which economic growth has translated into more and better jobs for men and women, and the relative impact on each group. The analysis relies primarily on micro-level data from the annual Encuesta Permanente de Hogares for 2001 through 2016, the Encuesta Continua de Empleo for 2010-2014, the Censo Economico 2011, a census of firms, and the 2015-16 Encuesta de Empresas, a follow-up firm survey1. Patterns in labor supply and its correlates will be examined using household-level data, and the analysis will consider how gender and other worker characteristics are related to labor market outcomes. In addition, this note explores the degree to which private sector labor demand and firm productivity differ by gender; this is done using firm-level data to examine the drivers of firm performance and employment growth. Gender gaps in Paraguay’s labor market are estimated to be large. The analysis below finds that women have much weaker attachment to the labor force and are more likely to be neither in employment nor in education or training – one quarter of women aged 15-29, compared to 5 percent of men. Among women who do participate, their labor outcomes are inferior on average to men’s. Informality is widespread, and women are significantly more likely to work informally, whether in self-employment, as informal wage workers, or as unpaid family workers. Women are also more likely to be employed by informal firms compared to men. Gender gaps in earnings are very high, driven mostly by differences in employment status and sector. Being informal carries a larger monthly wage penalty for women (63 percent) than for men (47 percent). And women tend to be concentrated in highly informal sectors like retail, hotels and 1 These surveys were carried out by Paraguay’s Dirección General de Estadística, Encuestas y Censos (DGEEC). See Annex A for a description of data sources. 5 restaurants, agriculture, and other services, which are characterized by low productivity and low earnings. Controlling for observable characteristics, the unexplained gender wage gap exceeds 20 percent when measured using hourly wages, and 40 percent when measured using monthly wages. The female share of employment in firms is highest in micro-sized firms, which tend to pay lower wages than larger firms and have lower levels of productivity. The firm census data also reveals that average wages are lower in firms employing more women. Given that micro-sized firms struggle to grow, these firm-size effects impact women more than men, intensifying the gender divide in job quality due to a combination of dynamic and static effects. The large gaps in labor outcomes for men and women are highly persistent, showing only modest improvement during this period of robust economic growth. Female labor force participation rose markedly, from 48 percent in 1997 to 57 in 2016, as did the share of working-age women in school, especially those combining studying with work. On the other hand, female unemployment rates are relatively unchanged since 2005. The level of female employment increased more than male employment between 2008 and 2015, but rising formality rates disproportionately benefited men. Of the 226,000 net jobs added for women during this period, 58 percent were formal; of the 218,000 net jobs added for men, over three-fourths were formal. The share of female earnings in total annual earnings has not deviated much from the 30 percent mark, despite robust average wage growth and employment gains for women. Women’s inferior labor outcomes are compounded by, and help to reinforce, the observed sectoral segregation of men and women. By selecting into more accessible but less productive sectors such as retail, hotels and restaurants, and other services – sectors that accommodate self-employment when wage work is unavailable – women in effect limit their output and earnings. The fact that men earn higher returns to experience compared to women also serves to magnify the wage gap over time, due to women’s shorter work tenure as a result of childbearing and other family-care responsibilities. The structural transformation underway has clearly benefited female workers, but the concentration of economic activity in low-productivity services – and notably retail – does not bode well for creating more productive jobs in the future. The potential gains from structural transformation come from a shift out of low-productivity primary activities into more productive industrial and service-sector activities that have higher value-added content. But in Paraguay, most service sectors are characterized by low skill content and low earnings. The high degree of informality that pervades the economy is problematic, not only for job quality, due to the associated lower productivity and earnings of informal work compared to formal employment2, but also because it impedes robust and sustainable economic growth that is grounded in productive employment and income gains, which in turn can generate a positive feedback loop. Nevertheless, it is important to acknowledge that the expansion of labor-intensive service sector activities helped reduce inequality by absorbing more and more women into paid work. The challenge going forward will be sustaining further improvements in job quality while at the same ensuring more inclusive growth. Improving gender equity is critical to enhancing the potential economic and social gains of the ongoing demographic and economic transitions. Paraguay’s jobs and development challenges cannot be fully overcome without addressing gender gaps in the labor market. The degree to which Paraguay will achieve its development objectives will depend on how men and women can benefit from improved labor outcomes – for example, through greater labor force participation, increased labor demand, and better 2 Informal work may also be deemed of low quality due to its lack of social insurance and worker protections. Brummund et al. (2016) utilize four criteria for measuring job quality: income, benefits, stability, and job satisfaction. 6 jobs. Expanding economic opportunities for women and girls can have large payoffs for economic growth, diversification, and poverty reduction3. Given the large and persistent differences between women’s and men’s employment outcomes, further policy effort is needed to increase women’s access to productive economic opportunities. Policies aimed at fostering productivity and the competitiveness of the private sector for job creation and increasing women’s access to non-traditional sectors are key for addressing the low female participation rate, the higher unemployment rates among women and, especially, the occupational segregation that channels women toward low-quality jobs. The findings of this note can inform policy design to help Paraguayans reach their development potential through more, better, and inclusive jobs for men and women. The remainder of this note is structured as follows. Section 2 examines recent socio-demographic trends that have affected the number of women entering the labor market in Paraguay. Section 3 looks at gender differentials in labor market outcomes relating to work status, sector of employment and earnings, inter alia. Section 4 considers the gender composition of labor demand by private sector firms, and section 5 concludes with a discussion of policy options for the future. 3 IMF Managing Director Christine Lagarde and the Prime Minister of Norway, Erna Solberg, World Economic Forum Annual Meeting 2018 “Why 2018 must be the year for women to thrive”, https://www.weforum.org/agenda/2018/01/the-time-has-come-for-women-to-thrive-heres-how/ 7 2. SOCIO-DEMOGRAPHIC TRANSFORMATION AND FEMALE LABOR FORCE PARTICIPATION 2.1. DEMOGRAPHIC TRANSFORMATION Paraguay is currently undergoing a demographic transition that has social and economic implications. Paraguay’s population is growing at a high but slowing rate, and the country’s age composition is changing, driven by two offsetting demographic factors: declining fertility rates and rising life expectancy. During the last century, countries around the world experienced similar transitions, resulting in larger shares of prime- age and elderly populations. These demographic trends are more advanced in OECD countries and in the rest of Latin America, but Paraguay is well on its way (Figures 1 and 2). Generally speaking, falling fertility rates lead to a decrease in the dependency ratio and a rise in the working age population share. Paraguay’s working-age population share has already risen from 63 percent in 2002 to 69 percent in 2015, and is projected to reach 73 percent by 2025. Figure 1. Fertility Rates and Life Expectancy, International Comparison (2014) Source: World Bank, WDI The growth of the working-age population has put considerable pressure on the labor market, creating a need for faster job creation. As women are having fewer children, the resulting smaller family sizes are less taxing on women’s time and result in more women entering work, although the effects are tempered by increasing elder care demands. The labor force expansion witnessed over the past decades has already contributed to GDP growth, and implies the potential to derive an additional demographic dividend4 from 4 The demographic dividend concept refers to the positive economic gains generated by a significant increase in the working age population sustained over a 20- to 30-year horizon in countries entering the second phase of the demographic transition, which is characterized by low fertility rates, low mortality, and rapidly falling dependency ratios (Ahmed et al., 2014). There is evidence in the literature of more rapid economic growth when the working age population growth rate exceeds the total population growth rate (Bloom and Williamson 1998; Bloom et al. 2000). 8 increased human capital, labor earnings, and household consumption and savings. The pace of job creation over the past 15 years was fast enough to fully absorb new labor force entrants without exacerbating unemployment. Paraguay experienced very strong output per capita growth during this period (3.1 percent per year), one third of which was driven by the increase in the working age population5. Figure 2. Paraguay’s Declining Fertility, 1990-2050 4.5 4 Children per Woman 3.5 3 2.5 2 Replacement rate 1.5 1 Source: UN, DESA The projected increase in the labor supply could be enhanced in the long term by a larger share of women entering the workforce. The ultimate size of the demographic dividend will depend on Paraguay´s ability to create productive work opportunities for all, and to absorb large numbers of youth and women into productive jobs. As the demographic transition slows, promoting gender equality and increasing women’s labor force participation will be progressively important for enhancing otherwise slow-growth dynamics, and for reducing poverty and improving welfare for both women and men. Shifts in the broader social, economic, and cultural contexts also play a role in attracting women into different types of work. The strength of these “pull” factors will affect the economic and social dividends associated with more women in the labor force The key channels through which demographic change can enhance economic growth include productivity improvements and scale economies due to higher population density, the dilution of reproducible capital (i.e. investment does not keep pace with labor force growth), and age structure effects. This last channel has been the focus of much of the demographic dividend literature (Eastwood and Lipton 2011). 5 Ruppert Bulmer et al. (2017) 9 2.2. FEMALE LABOR FORCE PARTICIPATION Despite significant gains in the last two decades, female labor force participation has stagnated at levels far below the rate for men. Female participation has increased almost 10 percentage points since 1997, hovering around 55-57 percent in recent years (Figure 3). This compares favorably to the LAC regional average of 53 percent (Figure 4). But the participation gap vis-à-vis men is very large, equivalent to 27 percentage points. This gap has narrowed over time, mainly due to the increase in women’s participation, although declining participation by men – from nearly 87 percent in 1997 to 84.2 percent in 2016 – also contributed. Figure 3. Female and Male Labor Force Participation (%) 90.0 85.0 80.0 75.0 70.0 65.0 60.0 55.0 50.0 45.0 Male Female Source: Encuesta Permanente de Hogares Figure 4. Labor Force Participation Rates for LAC Region (%, 2016) a. Men b. Women Sources: SEDLAC, Encuesta Permanente de Hogares 10 Consistent with evidence from the economic literature, the main factors affecting labor market entry are gender, education level, and to some extent age. This is confirmed by findings from two different regression tests performed on household survey data6. The first logit regression estimates the likelihood of workers with given characteristics to participate in the labor market, or to remain inactive. The results indicate that women are 28 percent less likely than men to participate in the labor market. This statistical gap is similar to the probability advantage of a tertiary graduate over someone lacking primary school completion for labor market entry (Figure 5; Annex A, Table A1). The second test uses a logit regression to estimate the factors affecting the likelihood of workers to transition from inactivity to activity, conditional on being inactive in the initial period. This analysis, which uses a panel dataset from the Encuesta Continua de Empleo spanning 2010-2014, finds that men are 12 percent more likely to transition compared to women (Annex A, Table A2). Figure 5. Correlates of Being Active versus Inactive (probabilities expressed as average marginal effects, 2015) Note: Only statistically significant variables are shown. Full regression results are in Annex A Table A1. Source: Authors’ calculations based on Encuesta Permanente de Hogares data. Women in Paraguay are much more likely than men to be neither in employment nor in education or training (NEET) (Figure 6)7. In 2016, one-fourth of women aged 15-29 were neither studying nor working, compared to only 5 percent of men. Since 2001, some improvement was observed among women, but the gap is closing very slowly. Women’s labor force participation decisions are affected by their fertility and time constraints related to women’s time allocation to childcare and other household demands (see Das et al. 2018 for a comprehensive treatment of female labor force participation). Male fertility or time use does not have a comparable effect on male labor supply decisions. The high NEET rate for women is especially worrying because of the difficulty of transitioning into productive work. If youth and women are economically inactive because they are in education or training, they are investing in skills that may 6 Logit regressions allow for bilateral comparisons between two work states, whereas multinomial regressions compare worker characteristics across the entire set of work states. Neither regression tests for causality, only correlation. 7 Known by the acronym NEET in many countries, “disconnected youth” in the United States and “ni-ni” (neither-nor) in Spanish-speaking countries. 11 improve their future employability. The NEET population, by contrast, risks degrading its human capital, reducing the chances of finding skilled work, and increasing the prospects for social exclusion. Figure 6. Gender Differences in those Neither in Employment nor Education (NEET), Ages 15-29 (a) Urban (a.1) Female (a.2) Male (b) Rural (b.1) Female (b.2) Male Source: Encuesta Permanente de Hogares Rural residents have much higher NEET rates, especially rural women. Women in rural areas are more likely to be out of school and out of work compared to women in urban areas and compared to men in rural areas. The NEET rate among rural women aged 15-29 is 38 percent, significantly higher than the rates of 18 percent among urban women, 4.8 percent among rural men, and 4.3 percent among urban men. Further analysis is necessary to investigate the underlying causes of this high rate of female inactivity in rural areas. Given the scarce formal job opportunities in rural settings, as well as the low incomes typical of rural employment, women may not have the incentive to invest in education or participate in the labor force. Local social norms in rural communities may also be reinforcing traditional gender roles, e.g., men as breadwinners, women as care-providers in the household. Another potential factor is the relatively lower access to education for rural populations compared to their urban counterparts. Rural women may 12 also face large mobility constraints (e.g., lack of safe, good quality transportation to commute to work and school) or household income constraints (e.g., rural poverty rates are very high8), which may restrict access to training and economic opportunities. NEET rates are starting to decline as more women are electing to study. Given the gains in educational attainment observed over the last decade and a half, youth are staying in school longer. The increases are observed in both rural and urban settings, and for both genders. Furthermore, there is a positive trend in the shares of youth combining work and study, especially for urban females, which in particular coincides with the jump in labor force participation observed in 2012 and 2013. 8 Rural moderate poverty was 36 percent in 2017, and rural extreme poverty was 9 percent (World Bank 2018a). 13 3. GENDER GAPS IN LABOR MARKET OUTCOMES In the last two decades, Paraguay’s robust economic growth performance translated into significant job creation and improved labor market outcomes; improvements were felt across the labor market, by women as well as by men. Paraguay’s labor market is characterized by large disparities along gender lines. Despite increases in the demand for labor and rising incomes, women continue to lag men in nearly every labor market indicator (Figure 7). The gaps in labor market outcomes between women and men are analyzed in detail below. Figure 7. Snapshot of the Working Age Population by Gender (2016) Working Age Pop: 4.7 mi F: 2.4 mi, 50.3% M: 2.3 mi, 49.7% Active (Labor Force): Inactive: 3.3 mi 1.4 mi F: 1.3 mi, 41.1% F: 1 mi, 73.4% M: 2 mi, 58.9% Employed: Unemployed: M: 0.4 mi, 26.6% 3.1 mi 0.2 mi F: 1.2 mi, 39.7% F: 0.1 mi, 51.2% M: 1.9 mi, 60.3 % M: 0.1 mi, 48.8% Wage Employee: Employer: Unpaid: Self-Employed Family Farming 1.7 mi 0.1 mi 0.2 mi 0.6 mi 0.4 mi F: 0.7 mi, 41.6 % F: 0.04 mi, 23.4 % F: 0.1, 52 % F: 0.3 mi, 50.3 % F: 0.1 mi, 27.1 % M: 1 mi, 58.4 % M: 0.1 mi, 76.6 % M: 0.1 mi, 48 % M: 0.3 mi, 49.7 % M: 0.3 mi, 72.9 % Public Sector Formal Private Informal Private 0.2 mi Sector: 0.4 Sector: 1 mi F: 0.11 mi, 51.5 % F: 0.14 mi, 34.5 % F: 0.4 mi, 38.7 % M: 0.1 mi, 48.5 % M: 0.27 mi, 65.5 % M: 0.6 mi, 61.3 % Source: Encuesta Permanente de Hogares 14 3.1. GENDER GAPS IN UNEMPLOYMENT The gender gap in the unemployment rate is significant, despite low unemployment in general. In 2016, the unemployment rate was 5 percent for men and 7.6 percent for women. This gap has changed little in the last two decades. Although the gender unemployment gap has shown some fluctuation since 2012, this was driven by changes in female unemployment (Figure 8). Unemployment rates for both men and women in urban areas are higher than in rural areas, but the rural gender gap is larger (Figure 9). Urban unemployment tends to be more prevalent among more educated workers who may be queuing for a formal job. Rural unemployment, by contrast, appears to be less voluntary. Access to formal jobs is more limited outside of cities, and rural poverty is higher; both factors contribute to lower rural unemployment. The gap between men’s and women’s unemployment is around 2 percentage points in urban localities, and 3-4 percentage points in rural settings. Figure 8. Unemployment Rates for Men and Women (%) 16.0 14.0 12.0 10.0 8.0 6.0 4.0 2.0 0.0 Male Female Source: Encuesta Permanente de Hogares 15 Figure 9. Urban and Rural Unemployment by Gender (%) Source: Encuesta Permanente de Hogares Regression analysis sheds light on the characteristics of unemployed workers. Using a logit methodology, we find that women are 2.6 percent more likely than men to be unemployed (Annex A, Table A1). Women are also 14.7 percent less likely than men to transition from unemployment into work, and the women who do move tend to enter informal wage work or self-employment (Annex A, Table A2). But an even higher share of unemployed women exits the labor force altogether (one third of unemployed women, compared to one fifth of unemployed men). Education is the strongest determinant for being unemployed versus inactive for women and men, but is stronger for women. Having a secondary education increases the probability of being employed versus being inactive by 11 percent for women and by 5 percent for men; entering university increases the probability by 19 percent for women and only by 7 percent for men, and having a bachelor’s degree increases the probability by 31 percent for women and only 6 percent for men (Annex A, Table A3). Another determinant factor for women is being a mother. Having children younger than 13 years-old in the household decreases the probability of women being employed rather than inactive by 1.5 percent, but increases the probability of men by 2 percent. 16 3.2. GENDER GAPS IN WORK STATUS Even though twice as many formal jobs were created since 2008 compared to informal jobs, informal employment remains predominant for both men and women. According to the recent Jobs Diagnostic report, the informal share of employment fell from 78 percent in 2008 to 71 percent by 2014, based on workers’ main occupation (informal jobs lack social insurance coverage or take place in an unregistered firm9). One third of women are in informal wage work, slightly lower than the share for men (Figure 10). Nearly a quarter of employed women are self-employed and 11 percent are in unpaid family work, both considerably higher than the rates for men (15 and 7 percent, respectively). Although the formality rates for men and women are similar, men tend toward the private sector, whereas a larger share of women hold jobs in public administration. Figure 10. Comparing Female and Male Work Status for Primary and Secondary Occupations10 (%, 2016) Source: Encuesta Permanente de Hogares The gender gap among employers is stark. Only 3 percent of women are employers as a main occupation, half the percentage for men. Further analysis is needed to understand the reasons for the small presence of female employers in Paraguay. It may be related to gender differences in access to capital or property titles, low savings, or a lack of decision power over household earnings. Although employers can be informal, two-thirds were employers of formally registered firms in 201511. 9 A formal job is defined to include (i) wage employees contributing to Social Security, (ii) employers of a registered firm (RUC), and (iii) self-employed workers with a registered firm (RUC); informal employment includes (i) farmers/herders/fisherman (self-employed or employer of firm with no RUC), (ii) unpaid family worker, (iii) self- employed, employee or employer of firm with no RUC, (iv) wage employees not contributing to Social Security. 10 Primary and secondary occupation are designated by the survey respondent rather than defined based on work hours or income. 11 The share of informal employers in total employers fell from over 50 percent in 2008–10 to about 40 percent in 2011–14, and subsequently to 35 percent in 2015. 17 Because informal work is pervasive and pays relatively poorly, a large share of workers engages in more than one job. Men are more likely than women to be engaged in small-scale farming, especially in their secondary occupation. Women, on the other hand, are over-represented in self-employment for their secondary occupation, followed by informal wage work. Work status has implications for the number of hours worked. Women in Paraguay are more likely to hold multiple jobs and work fewer hours in each. In fact, most women have at least two jobs. On average, women work five hours less than men in their main occupation, and one hour less in their secondary occupation (Figure 11). Looking only at their main job, women in farm-related activities work only half the hours that male farmers work, and are likely to have an additional job to supplement their farming income (Figure 12). In each work status category, women work fewer hours, which greatly reduces female earnings compared to their male counterparts (gender gaps in earnings are discussed further below). Figure 11. Average Hours Worked per Week (2016) Source: Encuesta Permanente de Hogares Figure 12. Comparing Average Hours by Work Status (Main Occupation) (a) Female (b) Male Source: Encuesta Permanente de Hogares 18 Women are statistically less likely to be in formal jobs, and multiple factors feed into this lower probability. Probabilistic regressions comparing those in informal and formal work indicate that men are 4.2 percent more likely to be in formal work than women (see Annex A, Table A1). Young people, less educated workers, and rural workers also struggle to access formal employment. Work status is driven by a combination of labor supply and labor demand factors, and the underlying drivers vary by gender. With respect to labor demand, or lack thereof, the scarcity of formal jobs likely pushes job-seekers into informal work options. Gender or ethnic discrimination (discussed below) may also affect labor demand. The likely drivers of labor supply decisions include workers’ preferences for full-time vs. part-time work, education and skill level, and sectoral preferences, inter alia. Regressions carried out separately for men and women highlight gender-differentiated effects of several key explanatory variables: household structure, education, ethnicity and location. One important determinant of formality status for women is being a mother. Women with children, especially young children, have greater home-based demands on their time, leading women to seek informal work requiring fewer or more flexible work hours. Having children younger than 13 years-old in the household decreases women’s probability of formal work by 2 percentage points, while the presence of youth age 13 or older has a smaller but still significant negative effect (Figure 13; Table A3, in Annex A). The presence of an elder person in the household also decreases the probability of formal employment for women (by 3 percentage points). Higher informality rates among women can be interpreted as a coping strategy to balance home care demands and work. Parenthood has the opposite effect for men, for whom the presence of children in the household modestly increases their likelihood of formality. Education is a powerful determinant of being formally employed for both men and women, but it plays a larger role for women. For example, having a secondary education increases the probability of being formal by 20 percent for women and 17 percent for men; entering college increases the probability by 30 percent for women and 26 percent for men; and having a bachelor’s degree increases the probability by 32 percent for women but only by 26 percent for men (Table A3 in Annex A). Because women tend to struggle to access high quality employment, facing steeper odds due to, for example, shorter or less relevant work experience or gender-related implicit bias, educational attainment provides an important signal of worker capacity and productivity. 19 Figure 13. Logit Estimates Comparing Formal vs. Informal Work Status for Men and Women Average marginal effects for the probability of being formal, age 15+ (2016) Source: Authors’ calculations based on Encuesta Permanente de Hogares data Language and location also matter for work status. Although only 2 percent of Paraguay’s population is indigenous, seventy percent of the population speaks Guarani 12 . Two-thirds of monolingual Guarani- speaking households live in rural areas. With respect to work status, women from strictly Guarani-speaking households are 3 percent less likely to be in formal work, and men are 6 percent less likely. Urban residents not only have a higher probability of entering the labor force, they are in closer proximity to a wide range of job opportunities, especially to wage work in formal firms. These effects are most pronounced in Asuncion and Central, but also prevail around Itapúa and Alto Paraná (Figure 14). It is therefore not surprising that urban residents have a higher likelihood of being in formal employment. Note that the largest number of paid female jobs in Paraguay’s private sector are in Asunción. Despite department-level differences in proximity to large urban centers, the female share of paid employment is fairly consistent across departments. 12 Encuesta Permanente de Hogares 2017 20 Figure 14. Map of Paid Employment by Department Source: Authors’ calculations based on 2011 Economic Census (Censo Economico) on formal firms with at least one paid employee 3.3. GENDER DIFFERENCES IN THE DISTRIBUTION ACROSS SECTORS Distinct patterns emerge with respect to the sectoral distributions of men and women. Women are concentrated in the commerce sector13, other services (three-fifths of which in domestic work), and to a lesser extent agriculture, all sectors with relatively low productivity (Figure 15). These three sectors together employ nearly three-quarters of total employed women, compared to just over half of men. Men are more evenly distributed across sectors, but most prominently in agriculture, commerce, manufacturing and construction. 13 In this report, commerce sector includes wholesale and retail trade, hotels and restaurants, unless otherwise indicated. 21 Figure 15. Sector Distribution of Employment (2016) Source: Encuesta Permanente de Hogares Paraguay’s ongoing structural transformation from agriculture to a services-based economy is reflected in shifting employment patterns. GDP grew by 80 percent in real terms between 2001 and 2015, more than half coming from expanded services. At the same time, agricultural employment contracted, especially for informal farmers, while manufacturing grew slightly, and service sector employment expanded rapidly. The majority of recent job growth between 2008 and 2016 was in retail, hotels and restaurants (accounting for 45 percent of net new jobs) and government services (over 20 percent), followed by manufacturing (13 percent), construction (11 percent), finance and real estate (10 percent) and other services (9 percent). 22 Employment growth was more sectorally diversified for men than for women, although commerce sector growth was important for both. Looking at the changes in employment for the period 2005 to 2015, the largest gains for both men and women were in the retail, hotel and restaurant sector (Figure 16): the retail, hotel and restaurant share of male employment rose 4.5 percentage points, while the female share rose 4.9 percentage points. It is important to note that retail, hotel and restaurant jobs tend to be informal and have low productivity (appearing below the horizontal axis in Figure 16). The fact that job growth was fastest in this sector at least partly reflects the absence of better jobs in other segments of the economy. For men – including the large number leaving farm work – more jobs were added in the relatively higher productivity sectors of government, manufacturing, construction and financial services sectors. For women, after retail, many entered government jobs (relatively high productivity) followed by financial services and manufacturing. Even though other services lost some female employment share, nearly a fifth of new jobs for women were in other services, the least productive sector. Figure 16. Change in the Employment Share Across More and Less Productive Sectors (2005-2015) (a) Male employment 23 (b) Female employment Source: Authors’ calculations based on Encuesta Permanente de Hogares data Rising formality rates have improved job quality on average, but men have benefited more than women. More formal jobs were added compared to informal jobs between 2008 and 2015, and even though total female employment rose more than male employment during this period, fewer of the added jobs for women were formal (Table 1). Of the 218,000 net jobs added for men since 2008, 170,000 were formal, but of the 226,000 net jobs added for women, 130,000 were formal (Table 1). It is notable that most formal (and informal) jobs were added in commerce, but the government sector added 84,000 formal jobs during the period, more than half of which were taken up by women. After public administration, women found the largest proportion of formal jobs in commerce, manufacturing and other services, but found much more informal work in commerce and other services. 24 Table 1. Net Job Creation by Gender 2008-2015 Male Female Total Share in Total Share in total Sector Type of job Number Employment Number total change Employment change (%) 2015 (%) 2015 Agriculture, cattle, and fishing Formal 3,197 1% 16,082 409 0% 926 Informal -50,905 -23% 412,374 -326 0% 179,396 Manufacture and mining Formal 22,688 10% 91,315 19,726 9% 32,317 Informal 14,589 7% 175,174 -1,524 -1% 77,463 Electricity, gas, and water Formal 2,063 1% 9,414 1,944 1% 3,495 Informal 1,502 1% 2,793 677 0% 1,307 Construction Formal 11,950 5% 24,500 777 0% 777 Informal 34,008 16% 183,005 734 0% 1,496 Retail, restaurants, and hotels Formal 63,267 29% 146,676 30,947 14% 84,951 Informal 50,929 23% 295,830 53,893 24% 283,350 Transport and communications Formal 7,589 3% 42,679 1,423 1% 10,043 Informal -7,968 -4% 57,530 -1,941 -1% 4,921 Finance and real estate Formal 22,574 10% 53,824 8,075 4% 32,871 Informal -1,302 -1% 45,833 14,240 6% 31,689 a Formal 36,991 17% 47,048 21% Government and public administration 118,509 142,237 Informal 5,704 3% 27,976 9,450 4% 26,589 Other services Formal -109 0% 22,668 19,716 9% 39,608 Informal 1,344 1% 85,995 20,618 9% 270,169 Total net job creation Formal 170,210 78% 525,667 130,065 58% 347,225 Informal 47,901 22% 1,286,510 95,821 42% 876,380 a Note: Sector is self-reported in the household survey, and data on government employment differ from the public administration records. Informal workers in government are mostly full-time, fixed-term contractors, concentrated in more skilled occupational categories (e.g., scientists, technicians, clerks) as well as in manual labor. Source: Encuesta Permanente de Hogares Besides having low productivity, the sectors that employ the largest share of women have higher rates of informality. Among women in commerce (i.e., retail, hotels and restaurants), half are self-employed, 22 percent are in informal wage work and another 11 percent are unpaid. And for women working in other services, 72 percent are in informal wage work, and another 16 percent are self-employed. 3.4. GENDER GAPS IN JOB QUALITY AND EARNINGS Women earn significantly less than men, and multiple factors drive this large earnings gap. In 2016, women working full-time earned a median monthly wage of US$435, compared to US$611 for men (40 percent higher). The female share of total annual earnings in Paraguay was around 30 percent in 2015, a figure that has changed little over time (Figure 17). The gender gaps in labor force participation, work status, job quality, work hours and sector described and quantified above each affect the earnings of women, and as such contribute to the observed large wage gap between men and women. Other less observable factors further widen the wedge between male and female earnings. The discussion that follows examines these factors in detail. 25 Figure 17. Total Annual Earnings by Gender (billions, 2005 USD PPP) 30.0 25.0 20.0 15.0 10.0 5.0 0.0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Male Female Source: Authors’ calculations based on Encuesta Permanente de Hogares data Differences in work status are a primary factor behind the gender wage gap. Earnings differences across employment types are significant. Employers and public sector employees have the highest earnings on average, while farmers, self-employed, and informal wage workers earn significantly less (Figure 18). It is notable that even within these high-wage and low-wage work status categories, women earn less than men, with the exception of the public sector (see Annex B, Table B4). These differences compound the wage gap, especially given that women are concentrated in self-employment, farming and informal wage work. When we compare the distribution of monthly earnings, the percentage of women earning less than the minimum wage is much higher than the percentage of men (Figure 19, left panel). When non-wage workers are excluded from the comparison, we find that the distribution of wages for men and women is much closer, although women’s wages are still lower, and a large share of both men and women earn less than the minimum wage (Figure 19, right panel). 26 Figure 18. Average Wage by Work Status and Gender (Full-time Workers in 2016, 2005 USD PPP) (a) Monthly earnings (b) Hourly wage Note: Full-time defined as at least 35 hours per week. See Annex B Table B4 for details. Source: Encuesta Permanente de Hogares 27 Figure 19. Distribution of Monthly Earnings for Full-time Workers (2016, 2005 USD PPP) (a) Monthly wage (all workers) (b) Monthly wage (wage workers only) Note: Earnings from primary occupation only. Source: Authors’ calculations based on Encuesta Permanente de Hogares data Sectoral segregation explains part of the gender wage gap, although even within sectors, women earn less. Commerce, other services and agriculture, which employ three out of every four working women, have the lowest average female earnings (Figure 20; Annex B Table B4). Even within these sectors, men earn much more on average, partly linked to the fact that men engage in slightly different types of work within these sectors. In agriculture, women’s monthly average earnings were just over half of those for men (US$250 in 2005 US dollar PPP for full-time female agricultural workers, compared with US$414 for men). In other services, which for women are dominated by domestic work, women’s monthly average earnings were only 45 percent of those of men, due to poor remuneration of domestic workers14. 14 In 2016, female domestic worker wages averaged US$304 per month and US$2 per hour (2005 US dollar PPP). 28 Figure 20. Average Earnings by Sector and Gender (Full-time Workers in 2016, 2005 USD PPP) (a) Monthly earnings (b) Hourly wage Note: Full-time defined as at least 35 hours per week. See Annex B Table B4 for details. Source: Encuesta Permanente de Hogares Regression analysis that controls for individual and job characteristics estimates a very large gender wage gap. Using Mincer-type regressions to control for individual characteristics, we estimate the effects of work status, education level, and sector, among other factors, on workers’ monthly earnings (Annex B, 29 Table B5).15 16 The gender wage gap is estimated at 43 percent17, indicating a very large earnings advantage in favor of men, other things being equal. Other factors shown to be significant to earnings are formality status, education, experience, sector of work, region and ethnicity (Figure 21). Being in a formal job raises monthly earnings by over 50 percent, other things being equal. Returns to education increase monotonically with education level: tertiary graduates earn a significant wage premium—equivalent to 93 percent—over similar workers lacking a primary degree (and controlling for formality), and even the relative returns to completing university compared with only some tertiary schooling is large at 30 percent. Workers from Guaraní-speaking households earn significantly less than those from non-Guarani-speaking households. And there is a large wage premium for Asunción-based workers. Figure 21. Correlates of Earnings (Real 2005 Monthly Wages, 2015) Note: Reflects specification 1 in Table B5, Annex B. Source: Authors’ calculations based on Encuesta Permanente de Hogares data To control for differences in employment status between men and women, we estimate the correlates of monthly wages for wage workers only, but find a still large gender wage gap. One key factor driving the gap in earnings is the fact that women work fewer hours in their primary occupation compared to men, on average, and that women are more likely to be in non-wage informal work, which pays less. Looking 15 The unit of measure used here—monthly wage earnings in each worker’s main occupation—reflects actual hours worked. 16 Note that after controlling for selection bias using a Heckman correction, the results are nearly identical to the ordinary least squares (OLS) estimates reported in Annex B, Table B5. 17 Specification 1 in Table B5 in Annex B. 30 only at wage workers (excluding self-employed and employers), the gender gap narrows to 33 percent, and limiting the comparison to full-time workers only shows that the gap narrows still further, to 27 percent. Comparing the correlates of monthly and hourly earnings provides important insight into gender-based welfare differences and implicit discrimination. By using monthly earnings as the unit of measure, the estimates reflect the relative wellbeing of male and female workers with respect to their primary occupation. But in a more specific comparison of wages for the same hour of work (i.e., regressing hourly wages of full-time wage workers on the same set of explanatory variables considered above), the gender wage gap is still large at 19 percent (Annex B, Table B6)18. Sector of work is a significant determinant of earnings, and the magnitude of sectoral returns varies by gender. We observe large variations in male and female earnings across sectors, controlling for individual characteristics, formality and region. For example, women in agriculture earn a small fraction of the monthly income of their female counterparts in commerce. In contrast, men in agriculture earn 47 percent less than their male counterparts in commerce (Annex B, Table B7). Manufacturing employment for women is not well-paid compared to men, likely due to the very low wages prevalent in apparel manufacturing. Women in manufacturing earn 17 percent less than women in commerce, whereas men in manufacturing (and mining) earn 10 percent more than men in commerce. Jobs in the electricity, gas, and water sector pay the highest wages for both men and women (56 and 53 percent more, respectively, than commerce). The next most remunerative sectors for men are transport and communications, construction, and finance and real estate (the latter returns are not significant for women). Certain worker and job characteristics matter more for women’s earnings; others matter more for men’s earnings. Notable differences arise with respect to experience, formality status and education (Annex B, Table B8). Running Mincer-type regressions separately for male and female full-time wage workers, we find that monthly earnings are more strongly correlated with potential experience for men than for women, and likely reflect differences in actual experience due to women’s more frequent transitions into and out of the labor force. Being informal makes a bigger pay difference for women compared to men: the informality penalty (or formality premium) for women is estimated at 30 percent of the monthly wage, compared to 24 percent for men.19 With respect to education, attainment level is more strongly correlated with female earnings than male earnings. For example, having a secondary degree increases women’s and men’s wages by 44 percent and 26 percent respectively, while having a tertiary degree increases women’s wages by 95 percent and men’s wages by 75 percent. These patterns, largely consistent with international findings in the literature20, are slightly biased by sector selection. When we control for sector of work, female educational returns decline somewhat and the gender gap in education premia narrows. Government work has generated significantly better employment opportunities in recent years, especially for educated women, but is not free from gender-based discrimination. Most (net) job creation observed in the government sector since 2008 was in formal positions, and most added posts were filled by workers with at least some tertiary education. In fact, women with a tertiary degree filled 60 percent of the 47,000 added government sector jobs taken up by women, while men with a tertiary degree filled about 40 percent of the 37,000 added government jobs taken up by men. Government jobs pay relatively 18 We find no consistent time trend in the size of the gender wage gap. When we repeat the regression analysis for each year between 2008 and 2015, we observe sizable year-to-year variations in the gender wage gap, both positive and negative. 19 The informality penalty is even larger when all paid workers are considered (i.e., full-time and part-time wage and self-employed workers): 63 percent for women, and 47 percent for men (Annex A, Table A7). 20 Dougherty (2005) and Montenegro and Patrinos (2014) provide evidence of higher female educational returns, while Hubbard (2011) finds no difference in tertiary education returns for men and women. 31 more than private sector work: without controlling for sector, the monthly wage premium for public sector employment is 13 percent for men and 19 percent for women (comparing full-time wage employees). The premium is even higher with respect to hourly wages (18 and 32 percent, respectively), which reflects the lower work hours required in public employment. Better pay and other non-wage benefits make public employment attractive, especially for skilled women struggling to find alternative formal job opportunities. But public sector work is not a panacea for bridging gender differences in job quality. On the contrary, government employment offers better pay to educated men compared to educated women with similar characteristics. Male returns to education are in fact markedly higher than female returns in the public administration sector (Annex B, Tables B9 and B10)21. This includes at the tertiary level, precisely where we might expect less gender-based discrimination. This result implies that although the government sector is an important source of formal employment for women, there are large unexplained differences in the compensation of skilled men and women that may be the result of discrimination. The large unobservable earnings gaps estimated above point to both sectoral misallocation of female workers as well as likely discrimination. Because men and women have different individual characteristics on average (e.g., different education levels, different length of experience) and their jobs also have different characteristics (e.g., different sectors of work, formality status), part of the observed gender wage gap stems from these allocative differences, but part stems from unobservable factors that likely include discrimination. The Blinder-Oaxaca decomposition technique compares the size of the effect coming from differences in endowments to that coming from differences in returns. Whereas the Blinder-Oaxaca gender gap with respect to monthly wages is estimated at 31 percent, if women were assigned the same characteristics – or “endowments” – as men on average but there was no change in the returns to these endowments, the gender wage gap would narrow. If, on the other hand, women received the same average returns to their endowments as male workers, they would in fact earn more the men (Annex B, Table B11). This indicates that the largest part of the observable wage gap is driven by the fact that men and women are rewarded differently for their endowment characteristics (whether individual or job-related). Although we cannot decisively conclude this is the result of explicit discrimination, since there may be other unobservable factors that create a wedge between men’s and women’s productivity, unequal treatment of men and women seems a likely culprit. The gaps are smaller with respect to hourly wages but move in the same direction. The analysis in the following section considers the labor demand side of jobs outcomes in Paraguay as a complement to – and to validate – the labor supply side findings from the household survey data presented above. 21 Female returns to education in fact lose significance within the public sector. 32 4. LABOR DEMAND 4.1. SNAPSHOT OF PRIVATE SECTOR FIRMS The analysis below provides a sex-disaggregated snapshot of Paraguay’s private sector firms, their formality status and their demand for labor. The analysis describes the types of firms that are the main source of employment for women compared to those that disproportionately employ men. Gender- differentiated patterns with respect to firm size, age, formality status, sector, region, level of output and average wage level are examined using three main sources of firm-level data covering the period 2010- 2014 (see Annex A for a detailed description). In complement to these descriptive findings, regression analysis tests for the correlates of firm size, growth, productivity and wages that also test for gender effects. Informality is widespread among firms captured in the economic census, but formal firms generate significantly more employment. According to the 2011 census of firms in Paraguay (a national dataset of formal and informal firms excluding self-employed farmers and household-based enterprises), the private sector was comprised of 211,042 firms employing 799,153 employees in 2010 (Table 2)22 . Thirty-eight percent of identified firms were informal, and even among registered firms, 60 percent were self-employed entrepreneurs without any paid employees, and thus were essentially informal (from the perspective that they lack access to social insurance, consistent with the definition used in the labor supply analysis above). The 51,457 formal firms identified as having at least one paid employee accounted for two-thirds of all jobs23. Table 2. Breakdown of Firms Captured in the Economic Census (2010 data) Firms Employment Level (#) Share (%) Level (#) Share (%) Informal (unregistered, no RUC) 79,373 37.6% 123,397 15% Formal (registered with RUC) 131,669 62.4% 675,756 85% Registered Self-Employed 80,212 38.0% 138,731 17% Firms with at least 1 paid employee 51,457 24.4% 537,025 67% TOTAL 211,042 100% 799,153 100% Note: RUC is the registration number Registro Único de Contribuyente. Source: Economic Census 2011 22 Given the high degree of informality in Paraguay’s economy, the firm census does not capture about two-thirds of total employment. It does, however, capture formal private wage employment very well, and partly captures informal wage employment and registered self-employment. See Ruppert Bulmer and Scutaru (2018) for a more detailed analysis of Paraguay’s firm-level datasets. 23 Note that jobs in large registered firms are not necessarily formal, as employers may underreport or employ workers on temporary or informal contracts that do not offer social insurance coverage. 33 4.2. GENDER DIFFERENCES IN FIRM-TYPE, FIRM-SIZE AND WAGES Women are more likely to be employed by informal firms or engaged in self-employment compared to men (Figure 22). Considering the distribution of workers across firm type, we find that nearly 22 percent of women are employed by informal firms, almost twice the share for men. Self-employment is also more prevalent among women, one fifth of whom are in registered self-employment, compared to 15 percent of men. And within formal firms, employment is dominated by men by a ratio of two-to-one (Figure 23). Women, by contrast, comprise 56 percent of employment within informal firms. These findings are similar to the patterns observed in the labor supply analysis above. Higher informality rates among women translate into lower productivity (Figure 24), as well as lower earnings. Figure 22. Female and Male Employment Shares in Informal and Formal Firms and Registered Self- Employment 100% 90% 20.7% 15.1% 80% 70% 60% 50% 57.7% 73.5% 40% 30% 20% 10% 21.6% 11.3% 0% Female Male Informal Firms Formal Firms Registered Self-Employment Source: Economic Census 2011 Figure 23. Share of Male and Female Employees, by Type of Firm 70.0% 65.6% 60.0% 55.9% 52.2% 47.8% 50.0% 44.1% 40.0% 34.4% 30.0% 20.0% 10.0% 0.0% Informal Formal Registered Self- Employment Male Female Source: Economic Census 2011 34 Figure 24. Comparing Productivity of Informal and Formal Firms Note: Productivity is measured as a firm’s value added divided by its total employment. Source: Authors’ calculations based on Economic Census 2011. The female employment share is highest in micro-sized firms, and declines as firms get larger. Among informal firms with fewer than 10 employees, women outweigh men: 68 percent are majority female, compared to 32 percent majority male (Figure 25). The share of majority female firms is considerably lower among informal firms with 10-19 employees (31 percent) and even lower among informal firms with 20- 49 employees (15 percent). In the formal sector, the gender gaps are narrower, but majority male firms are by far the norm. Regression analysis confirms a negative correlation between female employment share and firm size24. Among formal firms with at least one paid employee, a 10 percent increase in the share of paid female employees is associated with a 1.2 percent decline in overall firm size (i.e., total number of paid employees, controlling for other factors (Annex C, Table C1). Moreover, firms with a majority of female employees are 11 percent smaller than those with a majority male workforce, other things being equal. 24 The regression analysis of the correlates of firm size and performance throughout this chapter is limited to formal firms with at least one paid employee. 35 Figure 25. Comparing Gender Share of Employment by Firm Size (a) Informal firms 100.0% 100.0% 90.0% 85.7% 80.0% 67.7% 69.2% 70.0% 60.0% 50.0% 40.0% 32.3% 30.8% 30.0% 20.0% 14.3% 10.0% 0.0% 0.0% 1 to 9 10 to 19 20 to 49 50 to 99 100 to 499 500+ Majority of Male Employees Majority of Female Employees (b) Formal firms 100.0% 81.3% 80.0% 75.2% 75.2% 69.3% 67.4% 59.2% 60.0% 40.8% 40.0% 30.7% 32.6% 24.8% 24.8% 18.7% 20.0% 0.0% 1 to 9 10 to 19 20 to 49 50 to 99 100 to 499 500+ Majority of Male Employees Majority of Female Employees Source: Economic Census 2011 Most firms in Paraguay are micro-sized and suffer from low productivity25 26. The vast majority of firms has fewer than 10 employees (Figure 26). This is a pattern common in developing economies, even among formal firms. Whereas micro-firms outnumber all other firm-size categories, they generate only half of all jobs, while large firms (100+ employees) account for a quarter of all firm-based jobs (Figure 27). Regression analysis shows a positive, significant correlation between productivity and firm size (Annex C, Table C2). This correlation is comparatively strong in Paraguay, relative to other countries27. 25 Ruppert Bulmer and Scutaru (2018) 26 Note that firm productivity is defined as value added per paid employee. Other factors that contribute to productivity, such as physical and human capital, are not addressed due to data limitations. As a result, no analysis of total factor productivity is possible. 27 Ruppert Bulmer and Scutaru (2018) 36 Figure 26. Distribution of Firms by Size Category (number of formal and informal firms) 200000 150000 100000 N. 50,000 Firm size 0 1-9 10-19 20-99 100+ Source: Economic Census 2011 Figure 27. Share of Employment by Firm-size Category 11.35% 13.86% 52.36% 6.71% 8.73% 6.98% 1-9 10-19 20-49 50-99 100-499 500+ Source: Economic Census 2011 The relationship between firm wages and firm size is mostly positive. Average labor cost per paid worker rises with firm size when size is defined as a continuous variable. When we control for sector and sales level, firms with more than 10 paid employees pay lower wages, potentially reflecting lower productivity, but when we add controls for productivity level, larger firms pay more. 37 Average wages are lower in firms employing more women. The average wage in firms that are majority female is 23 percent less than in majority male firms (Figures 28; and Annex C, Table C3). This is consistent with the lower earnings for women compared to men in the preceding labor supply analysis. Controlling for sector, size, age and region, average firm wages decline by 0.28 percent for every 1 percentage point increase in the female employment share. Figure 28. Kernel Distribution of Average Firm Wages, by Majority of Employees Source: Source: Authors’ calculations based on Economic Census 2011 4.3. GENDER DIFFERENCES IN SECTOR OF WORK Paraguay’s private sector exhibits large differences in the sectoral employment patterns of men and women. Consistent with the household survey data, the economic census data shows that women are more likely to hold jobs in less productive sectors such as commerce and certain types of services, which are also highly informal. In the mining, utilities and construction sector and in manufacturing, most jobs are held by men, and most of these are formal (Figure 29). 38 Figure 29. Unequal Gender Distribution Across Sectors and Formality Note: MinUtilConstr denotes mining, utilities and construction. Source: Economic Census 2011 Regression analysis finds that textile and apparel manufacturing, hotels and restaurants, and other services – all sectors with a high concentration of female employment – have relatively high labor- intensity and therefore low labor productivity, controlling for other factors such as firm size, age and region (Annex C, Table C2)28 29. Commerce, which is dominated by small firms, is estimated to be more productive only when controlling for firm size. Textile and apparel manufacturing has the highest share of female employees (65 percent) and is among the least productive subsectors when comparing average productivity across subsectors (Figures 30 and 31). But the presence of large textile and apparel manufacturing firms creates opportunities for women to transition into formal wage employment. Construction, on the other hand, is one the most productive subsectors, but less than 8 percent of the construction sector’s paid workforce is female. Other sectors with a positive productivity link and a low female share of employees include transportation, metals & machinery, and financial services. 28 The negative coefficient for utilities is an anomaly driven by firm size. 29 Note that the relatively higher productivity estimates may result from the simple regression specification which cannot account for input price variation or quality of outputs. Another potential source of bias is the fact that productivity is measured by value added per reported paid worker, independent of how many unpaid or contract or undeclared employees a firm may have. 39 Figure 30. Gender Breakdown of Employment within Subsectors Source: Authors’ calculations based on Economic Census 2011 Figure 31. Comparing Value-added per Worker across 2-digit Industries Note: Reference sector is apparel. Regression on pooled 2010 and 2014 firm survey data. Data labels are the ISIC Rev. 4 two-digit codes, as follows: 10: food manufacturing; 11: beverage manufacturing; 12: tobacco manufacturing; 13: textiles; 14: apparel; 15: leather; 16-18; wood, paper, printing; 20-21: chemicals; 22: rubber and plastics; 23: other non-metallic mineral products; 24-30: metal and machines; 31-33: other manufacturing; 35-39: utilities; 41-43: construction; 45-47: wholesale and retail trade (i.e., commerce); 49-53: transport and storage; 55-56: hotels and restaurants; 58-63: information and communications; 64-68: finance, insurance, real estate; 69-75: professional and technical; 77-82: administrative and support services; 84: public administration; 85: education; 86-88: health and social work; 90-93: arts, entertainment, recreation; 94-96: other services. Source: Authors’ calculations based on Economic Census 2011 40 Some of the observed lower productivity of female-dominated firms is explained by sector selection (i.e., the different sectoral allocations of men and women); but there is a negative correlation between firms’ female employment share and productivity level even when controlling for sector effects. Firms’ value added per worker varies from one sector to the next, but does not vary much within sectors when comparing the productivity distributions of majority female and majority male firms (Figure 32). We also tested whether women were any more prevalent in firms at either the bottom or the top of the productivity spectrum, given the wide disparity in productivity between the least and most productive firms in Paraguay30; this test did not yield large differences. But regression analysis indicates that firms with a higher share of female employment are less productive when controlling for sector, firm size, age, and region (Annex C, Table C2). Firms with at least 50 percent female employees are 9 percent less productive than majority male firms, other things being equal. And as the female employment share rises, firm productivity falls. Figure 32. Kernel Distribution of Value-Added per Worker, Comparing Majority Female and Majority Male Firms by Sector Source: Authors’ calculations based on Economic Census 2011 30 Ruppert Bulmer and Scutaru (2018) 41 4.4. DYNAMIC IMPLICATIONS FOR CLOSING GENDER GAPS IN JOB QUALITY Small firm size is correlated with slower job creation within firms. Smaller firms struggle to grow, possibly due to a lack of scale economies, making it difficult to compete with larger, more productive firms. Firms with 50 or more workers had significantly higher employment growth between 2010 and 2014 compared to smaller firms – on the order of 30 percent higher (Annex C, Table C4). Other impediments to job creation may stem from the challenging regulatory environment facing firms; Paraguay ranks below average on a number of Doing Business indicators31. Furthermore, many sectors in Paraguay are highly concentrated; in some subsectors, the top four firms (measured by sales) account for more than half of all sales in their sub- sector, and in some cases over 90 percent32. This may impede new firm entry and dynamism among small firms. Even in the retail sector, where firm entry is relatively easy, sector concentration is unexpectedly high. The limited growth of micro-sized firms impacts women more than men, and will intensify the gender divide in job quality due to dynamic effects. The higher allocation of women in micro-sized firms, together with micro firms’ lower productivity and wages translate into significantly lower female earnings, consistent with the labor supply analysis. Because women are predominantly working in smaller, less productive, slower-growing firms, they are unlikely to be able to catch up to men. The inferior labor outcomes of women compared to men are compounded by – and help to reinforce – the observed sectoral segregation of men and women. By selecting into more accessible but less productive sectors such as retail, hotels and restaurants, and other services, women’s output and earnings are in effect constrained. For those women who enter manufacturing, they are more likely to find work in the apparel or food and beverage sectors, both characterized by below-average productivity. Note that the preceding productivity analysis does not consider profitability or other aspects of firm performance, due to data limitations. There is evidence from Germany that less productive firms with a high share of female employees are not necessarily less profitable, suggesting that lower female productivity is compensated by lower female wages33. Sectors that have higher shares of female employees pay significantly lower wages, according to regression analysis. After controlling for firm size, age and region, we find that commerce sector firms pay 8-20 percent less on average compared to manufacturing firms (across a range of regression specifications), and services sectors pay 5-42 percent less (see regressions 2-9 in Annex C Table C3). Moreover, we find that average wages are lower in firms employing more women, and even lower in majority female firms, controlling for firm size and sector (as discussed above). The findings from the firm-level analysis of Paraguay’s private sector suggest that women most likely face impediments to accessing more productive, better paid work in larger firms and in non-traditional sectors. These impediments may be cultural and self-perpetuating as a result of women’s selection into commerce and services, or may be due to explicit bias (e.g., unskilled construction work largely excludes women but pays well), or may be a function of the business environment or other factors that constrain women more than men (e.g., access to credit, access to physical or human capital, lack of land assets or housing, transportation constraints, access to technology). 31 www.doingbusiness.org 32 Ruppert Bulmer and Scutaru (2018) 33 Pfeifer and Wagner (2012) 42 The large differences in job quality between men and women are likely the result of a combination of factors. The following chapter explores policy areas for consideration to address these likely obstacles to better female labor outcomes. 43 5. CHALLENGES FOR THE FUTURE AND POLICY OPTIONS This analysis documents the existence of gender gaps in almost every labor market indicator: participation, unemployment, work status, sector of work, and earnings. The various drivers of these gender gaps have a compounding effect along the life cycle. Girls may focus their education or skills acquisition in areas that are not in demand by the labor market. The poor wage prospects for young women keep many from entering the labor market. Once women enter, they face more barriers and receive lower returns to effort than men. Cultural norms around homecare responsibilities result in shorter spells of labor force participation or rotating entry and exit, reinforcing self-selection into informal work with shorter or flexible work hours. Women’s higher labor mobility costs, such as those stemming from exposure to security risks, also reduce the marginal gains from working. Exclusion from networks where business deals are made or recruitment takes place further limits women’s access to productive work opportunities. Sectoral segregation distributes female and male workers unequally across sectors and firm-types and reduces women’s likelihood of being in formal waged employment or being employers owning larger businesses. The existence of gender segregation across sectors results in worse labor outcomes and lower earnings for women in Paraguay. Firms with a high proportion of female labor tend to be small and grow more slowly. Women are more prevalent in the less productive and less well-paid sectors of commerce, hotels and restaurants, apparel manufacturing and other services. The strong economic gains observed in Paraguay over the past decade – reflected in robust job growth, reduced informality, higher wages, diminishing poverty and greater economic participation by women – belie persistent gender gaps and a truncated structural transformation. The marked shift from agricultural employment toward services has fallen short of achieving large productivity gains, due to the continued prevalence of low productivity sectors like retail, hotels and restaurants and other services. The limited growth of micro and small firms has not generated significant wage employment, and this in turn has affected and is affected by limited competition, diversification and integration with regional and external markets and the associated productivity gains. Whereas female workers have benefited on average from recent job growth, women are relatively more implicated by the various distortions prevalent in Paraguay’s economy. Progress toward improved female labor outcomes has stalled, and the large remaining gaps vis-à-vis men are costly for Paraguay’s economy. The fact that women are not participating optimally in the economy – whether due to lower labor force attachment or through low productivity and earnings – translates into significant forgone production 34 . Lower female earnings have dynamic implications for development outcomes by affecting consumption and investment decisions within the household. Inadequate household investment in human capital reduces labor productivity, both in present and future generations, whereas improving women’s access to productive employment generates positive development externalities such as increased human capital, better quality of life for women, improved child health and education outcomes, lower poverty, and enhanced social cohesion within society35. Economic production that is insufficiently inclusive of half of the population reduces the potential for sustained economic growth in the future. Although the available data do not allow us to identify which causal factors generate the largest effect on gender gaps in Paraguay, it is likely that multiple sources of differential gender treatment and 34 Teignier and Cuberes (2015) estimate per capita income losses from gender inequality at about 25 percent for Turkey and 26 percent for India. 35 World Bank (2015a) 44 discriminatory attitudes are at play. Policy options for reducing segregation and increasing women’s access to and ability to compete for good jobs can be designed to address supply-side constraints or demand-side constraints, given that employment is the result of a match between labor supply and labor demand. Moreover, causal factors may affect both supply and demand, for example relating to regulatory barriers, social norms, and implicit bias in the workplace. Increasing the economic contribution of women can be considered through four main policy channels: (i) enhancing the capacity of women to obtain better jobs; (ii) reducing the barriers facing women in accessing quality employment; (iii) increasing the productivity and earnings of informally employed women; and (iv) stimulating the creation of new service-sector jobs that will be attractive to women in the future. There are many policy approaches and entry points for achieving more inclusive growth through enhanced female labor outcomes. The discussion below presents various policy options for consideration, organized around these four main channels, and drawing on lessons from international experience where possible. 5.1. ENHANCING WOMEN’S CAPACITY TO OBTAIN BETTER JOBS Addressing the various ways in which women fail to qualify or compete for better jobs in Paraguay could focus on increasing job market readiness through better education and skills, or through fostering better school-to-work transitions and more efficient matching of job seekers with available jobs. a) Education and Skills Paraguay can modernize its educational system to improve the curricula and provide programs that facilitate school-to-work transitions. Paraguay’s education system is structured around curricula that emphasize traditional subjects and learning methods, and the development of certain abilities not necessarily in demand by private employers. International trends in education reform have introduced a greater emphasis on reasoning skills rather than rote memorization, learning assessments and metrics, and skills that are needed in the labor market. Chile provides a good example of effective reform that used learning metrics and PISA’s reading framework to guide national curriculum reform 36 . Vietnam has successfully modernized its school system by adapting the Colombian model “Escuela Nueva” that focused on improving teamwork and developing critical thinking37. To improve education quality that prepares boys and girls for the labor market, Paraguay can prioritize investments in skills development programs to introduce or update technical and non-technical (e.g., reasoning, team work) skills38. Skills development programs could focus on enabling female workers in male-dominated fields, particularly for individuals with lower education levels. In Paraguay, women have higher rates of tertiary education, but are concentrated in female-dominated fields associated with lower returns. It is important to analyze potential gender differences in educational choice and fields of specialization that may be 36 In Chile after the reform, the proportion of 15-year-olds who achieved reading scores at or above a Program for International Student Assessment (PISA) level of proficiency increased from 52 percent to 69 percent between 2000 and 2015 (WDR, 2018) 37 Vietnam’s recent educational results have surpassed the OECD countries in many levels (World Bank World Development Report, 2016). 38 The Agenda Educativa 2013-2018 was already aligned with international practices and can provide a framework for future educational reform (World Bank, Paraguay Policy Note, 2018). 45 causing lower shares of female workers to enter high-productivity male-dominated sectors such as utilities, financial services, and construction. Interventions to address these gaps range, for example, from encouraging girls to enter technical school subjects such as science, IT and math, to enhancing tertiary curricula in areas of support services to industry (e.g., marketing, logistics, design, communications, IT), to targeting new vocational trainings to women. Cost-effective instruments that provide information on earnings tied to educational choice can also be useful. Investments in education can be associated with beliefs around the relative job prospects and earnings of men and women. When there are higher returns to working for men, as in Paraguay, families tend to invest more in the education of their sons39. This can reinforce lower educational attainment and skill acquisition of women. Providing information on the returns to education is a low-cost way of increasing educational attainment. Under a program in the Dominican Republic, individuals received data on the relative earnings of primary, secondary and university graduates. An impact study found that students significantly underestimated the returns to secondary education, and that providing them with information led them to complete 0.2 additional years of schooling (Jensen, 2010). A program implemented in Kenya in 2008 evaluated the impact of knowledge on real returns to vocational training and presented persuasive messaging (e.g., a video of female auto-mechanics) to encourage female participation in traditionally male- dominated, more remunerative fields. According to baseline data, males overwhelmingly preferred traditionally male-dominated courses like motor-vehicle mechanics and women almost exclusively chose traditionally female-dominated courses such as hairdressing. The program increased female enrollment in training for male-dominated occupations, and females had the greatest financial returns amongst voucher recipients (Hicks et al. 2015). However, women were not more likely to get a job in these male-dominated fields, suggesting the presence of additional barriers to entry. Vocational and technical skills training programs – whether targeting initial skills acquisition or life-long learning – have proved to be effective in improving labor outcomes for women. In Latin America, these programs have been largely successful at increasing labor force participation and earnings among female participants (Ibarraran and Shady 2009; Attanasio et al. 2017). The government can focus on labor training programs that promote equal gender participation – since these have been found to have disproportionately positive effects on outcomes for women trainees, especially in labor markets with substantial gender differences (see Box 1). 39 Shrestha and Palaniswamy (2016) conducted a study of siblings in Nepal and found evidence for this type of behavior in poorer households which led to negative spillovers for girls’ education. 46 Box 1. Hard and Soft Skills Training Program in LAC • Colombia: The program Jóvenes en Acción provided three months of classroom training to low- income unemployed young adults, followed by three months of apprenticeship in a private company. The average wage and the probability of being a paid employee increased by 7 percent for women. More recent evaluations of the program show sustained long-term effects. The program had a positive and significant effect on the probability of working in the formal sector for women eight years after randomization. Beneficiaries of the program were more likely to work for a large firm and had higher earnings. A cost-benefit analysis found that the gains were stable over time, with an internal rate of return of 20% for women (Attanasio et al., 2015; Attanasio et al, 2017). • Peru: The training and internship program ProJoven encouraged female participation in training for traditionally male-dominated occupations and provided subsidies so that women with children could participate. Eighteen months after participation in the program, employment rates for women improved by about 15% (while employment for men fell by 11%), gender occupational segregation diminished by 30% and women’s labor income improved by 93% (while men’s earnings increased by 11%) (Nopo et al., 2007). • Panamá: The PROCAJOVEN program provides short-term training for low-income unemployed youth 18-29 years old. Classroom training has two parts: job readiness skills and technical training (120 and 150 hours, respectively), followed by 172 hours of internship in a firm. A second modality focuses on the transition for first-time job seekers with complete secondary education, providing job readiness and a longer internship (344 hours). The program had a significant effect on employment rates (44% were employed in the treatment group compared to 32% in the control groups), particularly for those living in Panama City (47% for treatment, 32% for control) (Ibarrarán and Rosas, 2009). • Dominican Republic: Two soft-skills interventions were tested. For one group, the program provided a combination of soft skills and technical training tailored to the needs of employers and a two-month internship upon graduation. For the second group, the program provided only soft skills and an internship. Women who participated in the two groups were found to have higher employment rates 12 months after the program was completed. This indicates that soft skills play a key role in increasing labor force participation and reducing segregation. In the long run, participating women also showed higher self-esteem and lower fertility rates (Acevedo et al, 2017). The importance of non-technical skills (or soft skills) for improving jobs outcomes for women is being increasingly recognized in the literature and by policymakers. The development of soft-skills such as self- esteem, communication and decision-making can be very effective in reducing gender gaps. Many programs combine vocational and hard skills sector-specific training with soft skills. Even in advanced economies with high levels of educational attainment and high female labor force participation, there is evidence that cultural norms reduce women’s ambitions and slow their career advancement. Mentorship and soft skills development programs can change women’s beliefs about their skills and enhance their confidence and self-advocacy abilities. Dominican Republic provides an example of a program implemented through private institutions targeting improved employment opportunities for low-income youth (see Box 1). In Nigeria, a program prepares university graduates for work in the emerging sector of information and communication technology (ICT)-enabled services. The randomized control trial had no gender component and provided ICT training with hard and soft skills in three competency areas: oral and written communication, basic computing, and cognitive skills. The results showed an increase in women’s probability of working in ICT and reduced gender-based employment segregation, particularly among women who revealed self-defeating implicit gender biases in the baseline (Croke et al. 2017). 47 b) Labor Insertion and Intermediation There are many examples of programs that address the challenges of school-to-work transitions through employer-specific training and work experience. The German dual vocational education and training system is widely regarded as one of the best models. The program, which trains apprentices in a specific company while enrolled in a public vocational school, is credited as a key factor in the country’s relatively low youth unemployment rate40. There are some critics however who argue that the model excludes low- achieving youth, who are not able to enter fully-qualifying VET programs (Dunbar 2015). Paraguay itself has two ongoing apprenticeship pilots: MOPADUAL and SAPE’A. The Modelo Paraguayo de Formación Dual (MOPADUAL) program, launched in 2016 in cooperation with the BIBB (Bundesinstitut fur Berufsbildung, Germany’s Federal Institute for Vocational Education and Training), provides youth with a dual track education that combines classroom-based theory courses with practical work experience through apprenticeship. The Ministry of Labor has also been working with Plan International on the SAPE’A program, launched in 2015, which includes an internship program combining technical and soft-skills training with mentorship within a private company during a defined period of internship. Both programs are in the early stages and operate at a small scale. Neither explicitly targets young women. Initial results from SAPE’A are positive, but careful impact analysis, program design adjustments, and addressing cost- effectiveness and scalability are essential next steps, and should be taken using a gender lens to identify gaps or opportunities for improving female labor outcomes. Government can also assist by connecting employees and employers. Coordinating the different parties to job matching – job seekers, potential employers, and training and intermediation service providers – can be fostered through information-sharing to improve job seekers’ awareness and understanding of private sector opportunities. There are examples of electronic platforms for labor services matching, but these have been most effective when private-sector led. In Peru, for example, integrating mobile phones into traditional public intermediation services increased employment among job seekers by 8 percentage points in the short term by increasing the speed with which workers were matched to jobs (Dammert et al. 2015). The Government of Paraguay offers a range of labor intermediation services through the Dirección General de Empleo, targeting youth and adults for labor insertion and job search support. The effectiveness of these programs is not well understood due to lack of tracking and evaluation processes. Given that the Ministry of Labor’s recent survey of employers concluded that most recruitment takes place through personal connections41, it would be an opportune moment to revisit the current service offerings, and introduce monitoring and evaluation systems. In addition to government-provided job matching services – especially important for hard-to-place job seekers – the Government could contract with third- party providers to train and place job seekers, where remuneration of the service-provider is tied to successful job matches (i.e., performance contracts). Without knowledge of job opportunities, female job seekers are less likely to seek better work options when job search is costly. Subsidies for job search can be effective in increasing search intensity. Rural residents are less likely to search in urban centers, for instance. In settings with high labor mobility costs, a transportation stipend for job search may be effective in bridging hurdles to job search, especially for low-income workers. A study conducted in Ethiopia evaluated the impact of search costs on labor market outcomes for unemployed youth in spatially dislocated areas of urban Addis Ababa (Franklin 2015). Some job seekers were randomly assigned to receive a transport subsidy twice a week, covering the cost of travel 40 Dunbar (2015). 41 Observatorio Laboral (2017). 48 from the outskirts to the city center where information about vacancies is available. Receiving the transport subsidy increased the likelihood of finding permanent employment by 7 percentage points after 4 months. Enhancing social networks for women can also have positive impacts on wage employment. Because recruitment is to a large degree through personal connections, the fact that women have more tenuous connections to business contacts – due to their typically informal work status – may be another constraining factor. Women could therefore benefit from establishing or strengthening their own networks or seeking out male mentors; these types of efforts could be led by industry bodies (industry associations or chambers of commerce) or other civil society organizations. 5.2. REDUCING BARRIERS FACING WOMEN IN ACCESSING QUALITY EMPLOYMENT Women face both explicit and implicit barriers to good quality jobs. Some of these obstacles may be physical constraints to accessing school or work locations. Some obstacles are embedded in the regulatory framework, and therefore can be addressed through direct Government action. Other impediments may be less tangible and thus harder to grapple with, but may be even more binding; as such, they represent a priority area for policy intervention. a) Safety and Mobility Improving women’s access to employment requires physical access to schooling and places of employment. This can prove challenging in many settings, due to lack of physical infrastructure, or due to high crime rates or unsafe transportation options that can disproportionately affect women, reducing female educational attainment and labor supply or female entry into industries in locations far from home or in jobs requiring night travel42. Evidence from many countries shows that female self-employed are much more likely to work from home and that female-owned firms have their customers within a smaller radius than male-owned firms 43 . The implied limited market may be restricting their success. The Government should measure the extent to which labor mobility is constrained by improving data collection on safety and violence issues in rural and urban areas that may be impacting women’s schooling or labor decisions. Where gaps in safe transport options exist, Governments should allocate resources and introduce mechanisms to address risk of violence against women and girls in public transportation systems. Physical infrastructure and economic planning at the local and regional levels needs to take account of gender-specific challenges; future planning should therefore incorporate an explicit gender analysis. Programs such as UN Habitat’s Safer Cities Program are currently improving data collection on safety. Some other programs in LAC are addressing the risk of violence against women and girls in public transportation systems. In Rio de Janeiro, a new program called “Via Lilas” installed electronic kiosks, placed at stations along many transport routes. In Colombia, the Supervia suburban rail lines provide information about how women can seek support for gender-based violence. 42 Borker (2017), as cited in Das et al. (2018), provides evidence that schooling decisions by Delhi women are affected by poor security through reliance on safer but very costly transport options and selection into lower quality universities that are safer to access. 43 In Sri Lanka, for example, the customer-base for female-owned businesses is significantly smaller geographically than that for men: 48 percent of female owned firms had all of their customers within 1 kilometer, but this was true for only 30 percent of men (De Mel et al. 2008). 49 b) Recruitment and Hiring The Government has an important regulatory role to play to ensure that gender discrimination in recruitment practices does not occur. Clarifying to employers the regulations prohibiting job postings that define positions as male-preferred or male-only or require a specific marital status or physical traits for women – or introducing new regulations if needed – should be accompanied by monitoring and enforcement mechanisms. Companies seeking a better gender balance in recruits can reduce “male” wording in job postings44, implement gender-blind hiring processes and reduce their reliance on informal networks, inter alia45. c) Legal Barriers and Labor Regulations The labor code spells out worker protections to ensure safe working conditions and other worker rights, in compliance with international standards codified in ILO conventions, but enforcement is not universal. A review of national workplace regulations should be carried out to ensure no limits on female work, and adequate accommodation of female employees, including basic facilities such as separate toilet facilities, changing rooms, and breastfeeding rooms. Law 496/1995 already prohibits discrimination – including wage discrimination – on the basis of gender, guarantees that women and men enjoy the same rights and duties in the workplace, and establishes sexual harassment as grounds for dismissal. Application of the law is imperfect, however. The Ministry of Labor has the mandate to enforce implementation of these measures through awareness campaigns and labor inspections. There is scope for scaling up government oversight activities in light of this report’s findings, and civil society partners could play a complementary role. Most countries have legislation in place that addresses domestic violence and harassment at work. In Paraguay, Law 1600/2000 affords special and urgent protection to the victims of domestic violence, and the recently approved Act 5777 starting to criminalize femicide, as well as other forms of violence, will provide shelter and legal assistance to victims. Because of declining marriage rates, domestic violence is increasingly likely to occur between unmarried intimate partners; legislation and public safety campaigns may need to be extended to this category. Other types of gender-based violence, while less extreme or less explicit, are harmful to women: to their physical health, their human capital and their economic well- being. It is therefore important to extend laws to protect against economic coercion and violence. Additionally, although there is legislation on sexual harassment and criminal penalties in employment, there are no civil remedies established in Paraguayan legislation46. Maternity leave policies have the potential to increase female labor supply, but can also discourage demand for female employees; the Government should review and consider amending female-only and non-flexible leave policies. Mandating employer-financed maternity leave places an undue burden on employers, discouraging them from hiring women of child-bearing age. Shifting the financial burden to the state would remove this bias. This could be part of a more comprehensive reform of social insurance programs that expands their coverage beyond the currently small pool of formal workers and reduces explicit and implicit taxes on labor that disincentivize formal work. The Government could consider introducing more flexible parental leave options, such as allowing leave-sharing between mother and father following the birth of a child. Whereas traditional maternity leave requirements are intended to protect new mothers and babies, they have the unintended effect of increasing employment segregation by reinforcing women as primary caregivers, ultimately decreasing their labor force attachment. 44 See Flory et al. (2015), Gaucher et al. (2011) 45 See Das et al. (2018) for a detailed discussion of implicit gender bias in hiring. 46 Women Business and the Law, World Bank (2018b) 50 d) Social Norms and Attitudes in Gender Roles Social norms are an important constraint affecting labor supply decisions, sector choice, and many other aspects of women’s employment outcomes. The persistent view of the household division of responsibilities in which men are providers and protectors, and women are household managers and caretakers often leads women to opt for working from home or in jobs that permit reduced and/or flexible hours to accommodate these caretaker responsibilities. Cultural norms – both at the household and community levels – reinforce employment segregation, further restricting women’s choices. Acceptable professional jobs for women, especially in some rural areas, are often just extensions of their caretaker roles at home: teacher, nurse, and cook (Muñoz Boudet et al. 2013). Evidence from opinion surveys indicates some preference for men as breadwinners, especially in times of job scarcity. Almost two-fifths of Paraguayans believe that, in times of job scarcity, men should have more right to paid work than women47. A 2008 survey48 found that a third of respondents believe that wives must obey their husbands regardless of whether they agree with him or not, and nearly one quarter of women perceive that wife-beating was sometimes justified. These views, although not strongly held by a majority of women, nevertheless translate into biases in the distribution of jobs and economic opportunities toward men (World Bank 2015b). The observed social norms around gender roles and the perceptions of women’s rights negatively affect women’s agency, limiting their ability to participate fully in the economy . In addition to constraining women’s labor market behavior, work status and occupational choice, social norms may favor the passivity of women within relationships, communities, and society, which contradicts the exercise of agency and economic empowerment. The relatively low rates of political participation by women, high rates of domestic violence and prevalence of teenage pregnancy in Paraguay all reflect a lack of women’s agency (World Bank 2015b). Whereas social norms and perceptions are not set by or the responsibility of the state, governments can use their position to influence perceptions or promote change. A positive attitude from political leaders towards women business owners was found to be critical to the development of a strong female entrepreneurial community (Burr and Strickland 1992). Government can influence social norms regarding gender roles in the private sphere – e.g., domestic responsibilities, child care – and the public sphere – e.g., participation in government, civil society, or types of work suitable for or attractive to women. Improving data on cultural perceptions and promoting information campaigns are essential to better understanding and relaxing restrictive social norms. Surveys can be implemented to assess the main perceptions regarding the role of women in society and in the labor market. Information campaigns can be designed and targeted to foster discussion of women’s various roles and rights, or targeted to girls and their parents to make it socially acceptable for women to have freedom of movement outside of home, interact with males, delay marriage and child-bearing, register assets in their name, and pursue traditionally male-dominated occupations. Information campaigns can also be targeted to female job seekers to promote non-traditional careers and change negative perceptions of work in economic zones (e.g., garment factories, ITC sector). 47 Based on 2012 data from the Latin American Public Opinion Project (LAPOP). 48 Survey was carried out by the Sociedad de Estudios Rurales y Cultura Popular (cited in World Bank 2015b). 51 e) Social Protection and Child Care Expanding social protection for women can increase their willingness to enter the labor market, or search for better-quality wage employment, or take entrepreneurial risks. Women are less likely to be covered by social insurance, leaving them at higher risk of income shocks in the event of job-loss due to, for example, injury, illness or childbirth. Increasing social protection coverage of working families can free women to supply more labor to market-based activities. Access to preventive health services and health insurance coverage can reduce the incidence and severity of illness, both for workers and for dependents requiring care by women in the household. Paraguay’s fragmented health care system is characterized by low quality of services for most of the population, and particularly in rural areas. Recent progress was observed with regards to expanding basic service delivery and reducing chronic malnutrition and teenage pregnancy, but maternal mortality rates remain among the highest in South America (World Bank 2018a). The care of elderly family members also typically falls to women, and we already observe worse labor outcomes for caretakers (namely a lower likelihood of formal employment). As Paraguay’s population ages, the needs of the elderly will create a growing burden on household resources, with potentially far-reaching implications for the labor market and for the economy at large. In this context, effective public policies to support the incomes and health outcomes of the elderly can mitigate the potentially negative effects. Because social insurance is linked to a formal employment contract, informally employed workers are largely excluded from coverage. Voluntary access by the registered self-employed was introduced in recent years, but has seen low take-up rates. Several countries have introduced ambitious reforms that de- link social insurance access from the labor contract. Chile reformed its pension system in 2008 due to its limited coverage of the labor force (Winkler et al. 2017; Attanasio et al. 2011). To incentivize workers to join the formal sector and expand social protection, Paraguay could consider establishing an integrated system of means-tested subsidies to supplement workers’ and firms’ (reduced) social insurance contributions, with subsidies financed through general tax revenue. The lack of social insurance for most self-employed workers and employers of small firms makes them more risk averse with respect to their work activities. This, in turn, can reduce earnings in both the short and long runs. Women may be even more affected than men, due to the additional income risks associated with their care responsibilities. In West Africa, an experiment offering a choice of agricultural insurance or savings instruments found that female farmers disproportionately preferred savings instruments, even though both male and female farmers were exposed to the same production risks. The difference in preferences, which may be driven by the additional life-cycle risks facing women (e.g., related to fertility and childcare), led to lower spending on inputs and lower yields for female farmers compared to their male counterparts (Delavallade et al. 2015). Improving access to social protection and needs-based social assistance for the self-employed and small entrepreneurs would increase the incentive to work outside the home and to risk personal assets to invest in their own business gains, knowing that they would be covered by a safety net in case of job loss. The lack of affordable, high-quality childcare presents a major obstacle to parents, especially working mothers of young children. This challenge is not unique to Paraguay. There are measures the Government can take to foster the development of a network of childcare providers, such as creating incentives to establish childcare businesses (e.g., through tax credits), ensuring quality and safety through an accreditation system with standards and monitoring, and providing training (technical, occupational health and safety, and/or business skills) for childcare entrepreneurs. 52 f) Workplace Culture and Environment When workers lack agency and voice, they cannot get redress for work-related grievances or unfair treatment that is prohibited by law, including gender-based harassment. Women are more likely to be subject to sexual or psychological harassment that may arise within traditional cultural practices and attitudes, but are outdated, diminish productivity and dignity, and ultimately hold back female job performance49. Each of these factors widens existing gender gaps. The state has a responsibility to ensure a fair workplace environment for women. The Government should (i) establish training and advisory services to promote behavioral change, gender sensitization, and grievance mechanisms and penalties for sexual and psychological harassment, (ii) conduct these trainings within Government agencies, leading by example, and (iii) incentivize private firms to carry out similar training. Employers’ and workers’ associations can be effective promoters of gender-sensitive workplace culture in terms of fostering equal-treatment values and attitudes towards women and men, and ensuring human resource policies that are at a minimum non-discriminatory, but may encompass efforts to increase female entry, retention, and promotion. Government can play a role in developing and sharing guidelines and introducing their own gender-sensitive HR policies within the civil service. Sensitivity to discriminatory workplace attitudes and equal-treatment values could also be integrated into the Government’s various soft-skills training efforts (for example, as part of SNPP-SMART-ARANDU). g) Promotion and Retention High labor turnover rates are a constraint to firm productivity, because of the need to recruit and train replacement staff. They also limit workers’ skills acquisition on the job. High turnover can stem from tenuous contract terms and low pay, which could be addressed by offering open-ended contracts or tenure-linked incentives. But high turnover also results from workers’ perceived lack of career path. Employers – including the Government – need to improve the dissemination of career opportunities in their firms. Intermediation service providers can assist with reducing information asymmetries, which can impact women more negatively than men. In most countries, women are under-represented in leadership positions, whether in the public or private sector. Formal analysis of this effect has not been conducted in Paraguay, in part due to data gaps. Analyzing administrative data and conducting new surveys – including firm-level surveys assuming that information is collected on occupational roles (currently missing from existing firm-level data) – would help measure the extent to which women are present or absent at the management level. Understanding the nature and sources of gender gaps in leadership would enable policymakers to develop a strategy for addressing them. One option that has been implemented in other countries is a system of government incentives to promote gender parity at different organizational levels by recognizing companies with work cultures that are supportive of women. Gender quotas in leadership in the public and private sectors is shown to be effective in countering taste- based discrimination 50 and statistical discrimination against women, by transforming perceptions of 49 ILO (2017) 50 Taste-based discrimination is a discrimination based in preferences and not in statistical evidence, where employers hold a ‘taste for discrimination,’ meaning that there is a dis-amenity value to employing minority workers (Becker 1957). Female and minority workers may have to ‘compensate’ employers by being more productive at a given wage or, equivalently, by accepting a lower wage for identical productivity. 53 women’s ability to lead. Quotas can be effective when gender disparities in education and other selection problems are not binding, which seems to be the case of Paraguay. Quotas may have positive externalities for women down the hierarchy of companies or for girls’ aspirations, due to role model effects and networking potential. Causal evidence on political quotas in India shows several positive effects: greater participation by women, changes in policy investments, greater aspirations and educational attainment among girls, and less corruption (Pande and Ford 2011). However, little evidence is available on long-term impacts of corporate quotas. In Norway, the boards of directors’ quota reform of 2003 was evaluated for its impact on firm performance. Applying difference-in-difference approaches to accounting data for a short period, Matsa and Miller (2011) and Ahern and Dittmar (2010) both find short-term losses that can be fully explained by increased hiring or the change in the characteristics of board members other than their gender. Even if the impact on firms is low, quotas can have far-reaching impacts on social norms, especially in the case of women in very visible positions at the highest levels of political office or as managers in large companies (e.g., SOEs or PPPs) or business associations. 5.3. INCREASING THE PRODUCTIVITY OF INFORMALLY EMPLOYED WOMEN Given the large number of women who work informally – many in home-based agricultural production or non-farm activities or in their own micro-enterprises – improving the labor outcomes of this group will require interventions to boost their productivity through increased access to productive assets. a) Increasing Small Farm Productivity Farming activities occupy a large economic space for women in Paraguay; increasing their productivity through better inputs would go a long way towards increasing farm incomes. One fifth of employed women engage in farming as either their primary or secondary occupation. The equal rights of women to access land, hold land titles and access credit are enshrined in law, but evidence suggests that gender stereotypes persist in how these assets are distributed. For example, during the period 2000 to 2009, in Eastern Paraguay only a third of property titles were held by women, and only one-fifth in the Western regions (INDERT 2008, cited in USAID 2011). Female heads of rural households are more likely to live in extreme poverty and/or on small farms (less than 5 hectares), and small farms are less likely to generate a sustainable livelihood or receive agricultural technical assistance to overcome the inherent challenges of small farmer agricultural development (Deere et al. 2010, cited in USAID 2011). Government could play a role in fostering linkages between small agricultural producers and local and external markets to boost production and earnings. Rural road networks are particularly problematic in Paraguay, limiting the extent to which small holders can bring goods to market or connect with aggregators. Production cooperatives are one route to overcoming the limited scale economies of small producers. Connecting to value chains also has the potential to raise farmers’ productivity and earnings, although current patterns of farm production – dominated by very large producers of soy and cattle – translate into very short domestic value chains with limited transformation or value addition. Innovation and identifying or creating niche markets are important pathways to diversifying production and raising value added content. Incipient efforts in these areas would benefit from increased policy attention to small farmers where results have been positive. 54 b) Supporting Female Micro-Entrepreneurs and the Self-Employed There is a growing body of evidence on effective ways to increase the productivity and earnings of entrepreneurs, including programs that target women. Mechanisms include technical support, business training, monetary support, and networking development, with many combinations in between (McKenzie and Woodruff 2008). The Government of Paraguay already has multiple programs to support entrepreneurship. The programs are provided through the Servicio Nacional de Promoción Profesional (SNPP), the SINAFOCAL Emprende, which already targets women, the Entrepreneurship Directorate under the Ministry of Labor’s Employment Office (Dirección General de Empleo), and initiatives at the Commerce and Industry Ministry. The various entrepreneurship programs are not sufficiently coordinated, nor is their effectiveness measured by tracking impacts on beneficiaries following the intervention. Two key steps would be important for addressing these shortcomings: (i) a comprehensive review of these potentially overlapping initiatives, with a view to reducing duplication and providing access to the widest audience possible; and (ii) introduce monitoring and evaluation systems to assess impact and increase effectiveness. And to better address the needs of women, an important first step would be to assess course content for gender sensitivity. Firms in Paraguay report difficulty in accessing financing. Certain aspects of the business climate in Paraguay are burdensome and thus likely to dissuade firm entry or investment. The World Bank Doing Business surveys indicate that the time and cost to start a business in Paraguay are high compared to the LAC average. Entrepreneurs face even bigger obstacles for accessing capital, due to lack of collateral or a performance track record, or a high-risk profile. And female entrepreneurs are likely to be especially challenged51. Gender gaps in accessing capital can stem from differences in property and inheritance rights, lower female earnings and decision-making power over earnings, lower levels of financial inclusion, or discrimination in capital markets. Cross-country data shows that female-managed firms are less likely to obtain a bank loan and are charged higher interest rates when loan applications are approved. Female borrowers are also more likely to pay higher interest rates and have higher collateral requirements (Coleman 2000, Riding and Swift 1990). Limited access to capital may constrain women’s risk-taking ability and acquisition of the human and physical capital required for income generating activities. Difficulty accessing capital may be driving both the small firm size and the sectors in which women operate businesses. The labor demand analysis above indicates that female self-employed are more concentrated in specific subsectors, especially in services, with lower productivity and remuneration52. Bruhn (2009) finds that female-owned businesses are generally smaller because they lack start-up capital as a consequence of lower asset endowments, and because small firms in the service sector often lack collateral, which constrains access to credit to grow their businesses, creating a vicious circle. Providing female entrepreneurs with capital can improve business outcomes under the right conditions. Microfinance and matching grant programs to women have been introduced in many countries, with mixed results (Beck 2015). In urban Ghana, an experiment randomly provided cash or in-kind grants to male- and female-owned microenterprises. For women running subsistence enterprises there was no gain from either treatment. For women with larger businesses, only in-kind grants increased profits, and cash grants did not have an effect (Fafchamps et al. 2014). Although microfinance may not be very effective in the entry of firms, it can be successful in promoting business growth. Banerjee et al.’s (2015) long-term study finds that eighteen months after gaining access to credit, borrowers for start-ups are no more likely to be 51 There is ample evidence that women are less likely to access formal financial services (Demirguc-Kunt, Klapper, and Singer 2013; Aterido, Beck, and Iacovone 2013). 52 This is consistent with the findings of Halford and Leonard (2006) and Bowden and Mummery (2009). 55 entrepreneurs, although the existing entrepreneurs invested more in their businesses. Policymakers should be aware that short repayment periods and high interest rates may undermine business growth (Blattman and Ralston 2015). Results are often dependent on targeting and training (Todd 2012), and some interventions are less successful for women than for men, given intrahousehold and other constraints on women. Bernhardt et al. (2018) find that cash grants tend to be allocated to the most profitable business venture within a household, which is often the male-run business, but when women are the sole household enterprise operator, positive capital shocks lead to large increases in profits. These results illustrate the importance of designing interventions to account for specific constraints faced by women. Recent innovations such as business plan competitions have generated positive returns. In Nigeria, a large-scale national business plan competition pilot randomly assigned grants to applicants in one group, while in another group, only competition winners were provided grants of approximately US$50,000. Surveys tracked beneficiaries for three years and showed that competition winners had greater firm entry rates, higher survival rates, greater profits and sales, and higher employment, including increases of over 20 percentage points in the likelihood of a firm having 10 or more workers (McKenzie 2015). Business training programs can also have positive impacts on profits and business growth. Most studies find that existing firm owners implement some of the practices taught in training, but the magnitudes of improvement are often modest. In Chile, a study evaluated the impact of business training versus asset transfer on self-employment and income. Around 95 percent of the participants of the program were female, although the results are not disaggregated by gender. The randomized evaluation found positive impacts on total employment, income, and positive business practices (Martinez et al. 2017). Self- employment increased by 18 and 28 percentage points in the two groups of the program, and short-term and long-term income increased by 17 and 52 percent, respectively. In Kenya, business training of female- run businesses was also found to contribute to increased profits, sales and owner well-being (McKenzie and Puerto 2017). In a study evaluating the impact of a business development program in Lima, Peru, using an experimental design, one group received only general training, while another group received technical assistance in addition. Valdivia (2015) finds that both groups showed increased sales revenues, self- reported adoption of recommended business practices and growth above 15 percent two years after the end of the treatment. But the group that received only general training took a longer period to show positive results. Networks can be effective in linking entrepreneurs with potential buyers and investors, mentors, or facilitators in public or private agencies such as government procurement officers, chambers of commerce and trade associations. Women tend to have difficulty accessing these types of traditionally male networks; Government could play a role here by facilitating female entrepreneurs’ networking opportunities. This may be especially effective for female entrants into male-dominated sectors, who can feel excluded from the network of their new occupation and thus struggle to find partners and suppliers (Campos et al. 2015). An innovative study in India evaluated the impact of peers for entrepreneurial success, and finds that women who train with a friend are more likely to have business loans, are less likely to be housewives, and report increased business activity and higher household income (Field et al. 2015). This suggests that when female entrepreneurs have more connections with other women entrepreneurs, it can be beneficial to their success. Feigenberg et al. (2014) also find that more frequent meetings among microfinance groups in India help build valuable social capital among women. Policies aimed at strengthening women’s networks can be focused on creating networking opportunities or connecting them with successful business owners. 56 5.4. STIMULATING JOB CREATION ATTRACTIVE TO WOMEN Many of the observed gender gaps in labor outcomes are related to the fact that there is insufficient labor demand in the formal private sector, which in turn constrains opportunities for wage work, especially formal wage work. Various factors limit the dynamism and job creation of Paraguay’s private sector, factors ranging from global economic trends and competitive pressures in the region, to the macroeconomic environment, to high levels of market concentration, to the structure of economic production. A full analysis of these factors is beyond the scope of this report, but the discussion below identifies several areas where policy interventions could yield better job opportunities for women. a) Core Economic and Regulatory Policies Macroeconomic policies related to investment or trade may benefit particular industries and therefore workers in those industries; applying a gender lens when assessing the employment impact of new policies is essential to avoid further disenfranchising women, and to identify opportunities to increase women’s welfare. For instance, the terms of an investment deal may foster job creation in construction or other industries where women have a low presence. Reviewing trade policies or fiscal incentives targeting FDI with a view to understanding the gender implications could help to narrow gender gaps by ensuring that women also benefit from the projected higher returns associated with a particular investment or trade reform. Impediments to firm entry are likely culprits in stymying job creation for both men and women. Young firms account for a low share of firms and a low share of employment in Paraguay compared to other developing economies. The World Bank Doing Business surveys indicate that the time and cost to start a business in Paraguay are high compared to the LAC average. Paraguay also ranks in the bottom half of the global sample for access to electricity and access to credit, protection of minority investors, and insolvency resolution. Firms need to perceive the benefits of formalizing, or they will remain in the shadows. The Government is already addressing some of these constraints to facilitate entry. Ongoing efforts include simplifying firm registration, reforming the insolvency regime, creating a moveable assets collateral registry to foster credit access, and introducing a risk-based customs control system. The wide productivity gaps between very large firms and the rest of the private sector may arise from differences in tax or regulatory treatment. The data show a positive correlation between firm productivity and firm size, especially among the most productive firms. Many sectors are highly concentrated, dominated by a limited number of firms; smaller firms struggle to compete in this setting. Government action to review its competition policy framework and level the playing field through neutral policies to foster the entry and growth of other firms would be helpful. It is not enough to attract large firms within the enclave model of export processing zones and tax-free imports, because this approach does not encourage the use of domestic inputs, nor does it create adequate opportunities for local innovation to meet related opportunities. Both of these effects limit potential job creation. Paraguay’s low entry and firm growth rates may stem from broader business climate challenges that are structural in nature. Structural challenges might include unfavorable economic conditions, small market size, poor connectivity to external markets, and a weak institutional environment for protecting property rights, enforcing contracts, levying fair and transparent taxes, and guaranteeing regulatory compliance. These types of challenges may be harder to address, but if tackled they may play a larger role in encouraging firm entry and diversification. Paraguay’s small domestic market and land-locked setting in particular highlight the need for better trade links between local producers and foreign investors in export manufacturing and external consumers, and for lower export costs. Transport costs are high, and the time 57 and cost of exporting exceed the LAC averages, rendering Paraguayan products less competitive in the international market. Trade facilitation measures to improve trade logistics will encourage diversification of the highly concentrated export basket by reducing the cost of exporting. Government can play a role in increasing market access through trade policies as well as physical infrastructure investment to link local, regional, and external markets. These types of transversal measures promote firm entry as well as higher productivity of both large and small firms, young and old, formal and informal. b) Promoting High-Skilled Services Sectors The future prospects for increased economic inclusion of women through better labor outcomes are integrally linked to the performance of the private sector and its capacity to create medium-to-high skill jobs. Low firm productivity is correlated with low wages and therefore low job quality; firms that can increase their output and sales can generate the space to offer higher wages and attract more productive workers. Two routes to enhancing firms’ labor productivity are through (i) building the technical and market knowledge and business skills of employers, and (ii) enhancing the quality of employees’ work. Education and skills are fundamental components of labor productivity, yet 29 percent of surveyed employers in Paraguay cite the inadequately educated workforce as a major constraint to firm operations, and 13 percent cite it as their largest obstacle 53 . The Ministry of Labor conducted a small survey of employers to identify key skill gaps and types of training that would be useful for their workforce. Respondents expressed demand for customer service and marketing skills, administrative functions (for example, sales, HR, accounting, and finance) and ICT. The cross-cutting skills in greatest demand were basic computer operation and foreign language proficiency (mostly English)54. To enhance the marketable skills of the future labor force to meet firms’ labor demand, broad changes in the traditional education curriculum are required. As Paraguay’s economy continues its transition toward services-based production, reaping productivity gains from structural transformation will depend on the private sector’s capacity to expand higher-level services activities. A key component of making firms more competitive and increasing domestic value- added is to include more domestic inputs of high quality, especially services. Whereas Paraguay’s large firms are relatively more productive, account for a large share of formal employment, and add more formal jobs to the economy, the findings above also show that certain manufacturing subsectors have low productivity. These include apparel and food and beverage manufacturing that involve low-tech transformation of inputs into final products with low levels of sophistication and minimal value addition. Local input could be extended to include pre- and post-production skilled services relating to design, product R&D, marketing, or logistics, among others. Tertiary curricula would need to be revised to produce graduates possessing these competencies. Another component essential for attracting investments with the potential to create high-quality skilled jobs is an ecosystem that can support technology-dependent investments (for example, large manufacturing plants) or innovative start-ups in IT or manufacturing support services. Investors are likely to look for local capacity to maintain and upgrade operating equipment, as well as an education system that can produce a cohort of internationally competitive service providers. These types of jobs would attract women in particular, given the dearth of alternatives in the private sector. 53 www.enterprisesurveys.org 54 These pilot survey findings will be useful for adjusting course content targeted to employers (through SINAFOCAL), and the Ministry of Labor is currently considering how to use the initial findings and scale-up the survey. For details, see Observatorio Laboral (2017). And for a description of active labor market policies in Paraguay, see Corvo and Ruppert Bulmer (2017). 58 REFERENCES Acevedo, P., G. Cruces, P. Gertler and S. Martínez. 2017. “Living Up to Expectations: How Job Training Made Women Better Off and Men Worse Off”, Mimeo Ahern, K. and A. Dittmar. 2010. The Changing of the Boards: The Value Effect of a Massive Exogenous Shock. Mimeo, University of Michigan. Aterido, Reyes, Beck, Thorsten and Iacovone, Leonardo, (2013). Access to Finance in Sub-Saharan Africa: Is There a Gender Gap? World Development, 47, issue C, p. 102-120, https://EconPapers.repec.org/RePEc:eee:wdevel:v:47:y:2013:i:c:p:102-120. Attanasio, O., A. Guarn, C. Medina, and C. Meghir. 2015. "Long Term Impacts of Vouchers for Vocational Training: Experimental Evidence for Colombia." NBER Working Paper No. 21390, National Bureau of Economic Research. Attanasio, O., A. Kugler, and C. Meghir. 2017. "Vocational Training for Disadvantaged Youth in Colombia: A Long-Term Follow-Up." American Economic Journal: Applied Economics 2017, 9(2): 131–143 Attanasio, O., Kugler, A., & Meghir, C. 2011. “Subsidizing Vocational Training for Disadvantaged Youth in Colombia: Evidence from a randomized trial.” American Economic Journal, Applied Economics 3 (3): 188-220. Beck, T. 2015. “Microfinance: A Critical Literature Survey.” IEG working paper No.4;. Washington, DC. Bernhardt, A, E. Field, R. Pande, and N. Rigol. 2017. “Household Matters: Revisiting the Returns to Capital among Gemale Micro-entrepreneurs”, NBER Working Paper 23358, National Bureau of Economic Research. Blattman, C. and L. Ralston. 2015. “Generating Employment in Poor and Fragile States: Evidence from Labor Market and Entrepreneurship Programs”. Available at SSRN: https://ssrn.com/abstract=2622220 or http://dx.doi.org/10.2139/ssrn.2622220 Bloom, D. & Williamson, J., 1998. Demographic Transition and Economic Miracles in Emerging Asia. World Bank Economic Review, 12(3): 419-456. Bloom, D., Canning, D. and Malaney, P., 2000. Population Dynamics and Economic Growth in Asia. Population and Development Review, 26(Suppl.): 257-290. Borker, G. 2017. “Safety First: Perceived Risk of Street Harassment and Educational Choices of Women”, Department of Economics, Brown University. Bowden, P. and J. Mummery. 2009. Understanding Feminism, New York: Acumen Press. Bruhn, M., and D. McKenzie. 2009. "In Pursuit of Balance: Randomization in Practice in Development Field Experiments." American Economic Journal: Applied Economics, 1 (4): 200-232. Brummund, P., C. Mann and C. Rodriguez-Castelan. 2016. “Job Quality and Poverty in Latin America”, Policy Research Working Paper No. 7927, Washington, DC, World Bank. Burr, S. G., M. Strickland. 1992. “Creating a Positive Business Climate for Women: An Approach to Small Business Development”. Economic Development Review, 10 (1), 63-66. 59 Campos, F., M. P. Goldstein, L. Mcgorman, A. M. Munoz Boudet, & O. Pimhidzai. 2015. “Breaking the Metal Ceiling: Female Entrepreneurs who Succeed in Male-dominated Sectors”. (English). Africa Region Gender Practice Policy Brief; no. 9. Washington, DC; World Bank Group. Coleman, S. 2000. “Access to Capital and Terms of Credit: A Comparison of Men- and Women-owned Small Businesses.” Journal of Small Business Management, 38(3): 37–52. Croke, K., M. Goldstein and A. Holla. 2017. “Can Job Training Decrease Women’s Self-Defeating Biases? Experimental Evidence from Nigeria”, World Bank Policy Research Working Paper No. 8141. Cuberes, D.; M. Teignier. 2015. “How Costly are Labor Gender Gaps? Estimates for the Balkans and Turkey”. Policy Research Working Paper No. 7319. World Bank, Washington, DC. © World Bank. Dammert, A. C., J. Galdo, and V. Galdo. 2015. “Integrating mobile phone technologies into labor-market intermediation: a multi-treatment experimental design.” IZA Journal of Labor and Development, 4(11). Das, S., E. Carranza and A. Kotikula. 2018. “Addressing Gender-Based Employment Segregation”, Jobs Group, World Bank. Deere, C. D., G. Alvarado, and J. Twyman. 2010. “Poverty, Headship, and Gender Inequality in Asset Ownership in Latin America.”, Center for Gender in Global Context, Michigan State University. Delavallande, C., F. Dizon, R. V. Hill and J. P. Petraud. 2015. "Managing Risk with Insurance and Savings: Experimental Evidence for Male and Female Farm Managers in West Africa:" IFPRI Discussion Papers 1426, International Food Policy Research Institute (IFPRI). De Mel, S., D. Pablo McKenzie, and C. Woodruff (2008). “Returns to Capital: Results from a Randomized Experiment.” Quarterly Journal of Economics 123(4): 1329–72 Demirguc-Kunt, Asli; Klapper, Leora; Singer, Dorothe. 2013. Financial Inclusion and Legal Discrimination Against Women: Evidence from Developing Countries. Policy Research Working Paper No. 6416, World Bank, Washington, DC. Dougherty, C. 2005. “Why are the Returns to Schooling Higher for Women than for Men?”, The Journal of Human Resources 40 (4). Dunbar, M. 2015. Secondary Vocational Education International Experience: Final Report, April 2015 (English). Washington, D.C.: World Bank Group. Eastwood, R. & Lipton, M., 2011. “Demographic Transition in Sub-Saharan Africa: How Big will the Economic Dividend Be? “Population Studies, 65(1): 9-35. Fafchamps, M., D. McKenzie, S. Quinn, C. Woodruff. 2014. "Microenterprise Growth and the Flypaper Effect: Evidence from a Randomized Experiment in Ghana," Journal of Development Economics, Elsevier, 106(C): 211-226. Feigenberg, B., E. Field, R. Pande, N. Rigol, and S. Sarkar. 2014. “Do Group Dynamics Influence Social Capital Gains among Microfinance Clients? Evidence from a Randomized Experiment in Urban India”. Forthcoming, Journal of Policy Analysis and Management. Field, E., S. Jayachandran, R. Pande, and N. Rigol. 2015. “Friendship at Work: Can Peer Effects Catalyze Female Entrepreneurship?” NBER Working Paper 21093. The National Bureau of Economic Research, Cambridge, MA. 60 Flory, J. A., Leibbrandt, A., & List, J. A. 2014. “Do Competitive Workplaces Deter Female Workers? A Large- scale Natural Field Experiment on Job Entry Decisions.” The Review of Economic Studies, 82(1): 122- 155. Franklin, S. 2015. “Location, Search Costs and Youth Unemployment: A Randomized Trial of Transport Subsidies in Ethiopia”. CSAE Working Paper WPS/2015-11. Gaucher, D., Friesen, J., & Kay, A. C. (2011). “Evidence that Dendered Wording in Job Advertisements Exists and Sustains Gender Inequality.” Journal of Personality and Social Psychology, 101(1): 109-128. Halford, S. and P. Leonard .2006. Negotiating Gendered Identities at Work: Place, Space and Time, Basingstoke. Palgrave, Macmillan. Hicks, J. H., Kremer, M., Mbiti, I., & Miguel, E. 2015. Vocational Education in Kenya: Evidence from a Randomized Evaluation. International Initiative for Impact Evaluation (3IE). Hubbard, W. 2011. “The Phantom Gender Difference in the College Wage Premium”, The Journal of Human Resources 46 (3). Ibarrarán, P. & Shady, D.R. 2009. “Evaluating the Impact of Job Training Programmes in Latin America: Evidence from IDB funded operations.” Journal of Development Effectiveness 1.2: 195-216. ILO. 2017. “Ending Violence and Harassment against Women and Men in the World of Work”. Conference paper, 12 May 2017, Geneva, Switzerland. INDERT. 2008. Paraguay Instituto Nacional de Desarrollo Rural de la Tierra (INDERT), 2008 statistics. http://www.indert.gov.py/index.php?a=historia Lagarde, C., Solberg, E. 2018. “Why 2018 must be the Year for Women to Thrive”. World Economic Forum, 2018. Martinez A., C., E. Puentes, and J. Ruiz-Tagle. 2017. "The Effects of Micro Entrepreneurship Programs on Labor Market Performance: Experimental Evidence from Chile" Working Paper, March 2017. Matsa, D. and A. Miller. 2011. Chipping Away at the Glass Ceiling: Gender Spillovers in Corporate Leadership. American Economic Review: Papers & Proceedings (May 2011). McKenzie, D. and C. Woodruff. 2008. “Experimental Evidence on Returns to Capital and Access to Finance in Mexico”, Mimeo. World Bank. McKenzie, D. J. and S. Puerto. 2017. “Growing Markets Through Business Training for Female Entrepreneurs: A Market-Level Randomized Experiment in Kenya” (March 3, 2017). World Bank Policy Research Working Paper No. 7993. McKenzie, D. J. 2015. “Identifying and Spurring High-growth Entrepreneurship: Experimental Evidence from a Business Plan Competition” (English). Policy Research working paper; no. WPS 7391; Impact Evaluation series. Washington, D.C.: World Bank Group. Montenegro, C. and H. Patrinos. 2014. “Comparable Estimates of Returns to Schooling around the World”, Policy Research Working Paper 7020, World Bank, Washington, DC. Muñoz Boudet, A. M.; P. Petesch; C. Turk; A. Thumala. 2013. “On Norms and Agency: Conversations about Gender Equality with Women and Men in 20 Countries”, Directions in Development: Human Development. Washington, DC: World Bank. 61 Ñopo, H.; Robles, M.; Saavedra, J. 2007. “Occupational Training to Reduce Gender Segregation: The impacts of ProJoven”, Research Department Working Paper No. 623 (Washington, DC, InterAmerican Development Bank). Observatorio Laboral. 2017. Encuesta sobre las Necesidades de Capacitación Laboral. Resultados Finales (Mayo de 2017). SINAFOCAL. Pande, R, and D. Ford. 2012. “Gender Quotas and Female Leadership”. Washington, DC: World Bank. Pfeifer, C., J. Wagner. 2012. "Age and Gender Composition of the Workforce, Productivity and Profits: Evidence from a New Type of Data for German Enterprises," IZA Discussion Papers 6381, Institute for the Study of Labor (IZA). Riding, A. & Swift, C. 1990. “Women Business Owners and Terms of Credit: Some Empirical Findings of the Canadian Experience.” Journal of Business Venturing, 5(5): 327–340. Ruppert Bulmer, E. and A. Scutaru. 2018. Firm Productivity and Employment in Paraguay 2010-2014. Jobs Working Paper Issue no. 19. Washington, D.C.: World Bank Group. Ruppert Bulmer, E.; S. Watson; D. de Padua and A. Garlati. 2017. Jobs Diagnostic Paraguay: The Dynamic Transformation of Employment in Paraguay (English). Jobs Series no. 10. Washington, D.C.: World Bank Group. Shrestha, Slesh A. & Palaniswamy, Nethra. 2017. "Sibling Rivalry and Gender Gap: Intrahousehold Substitution of Male and Female Educational Investments from Male Migration Prospects," Journal of Population Economics, 30(4): 1355-1380. Todd, P. 2012. “Effectiveness of Interventions Aimed at Improving Women's Employability and Quality of Work: A Critical Review.” World Bank Policy Research Working Paper no 6189, World Bank. USAID. 2011. “USAID/Paraguay Gender Assessment”, prepared by DevTech Systems, Inc. Valdivia, M. 2015. "Business Training Plus for Female Entrepreneurship? Short and Medium-term Experimental Evidence from Peru," Journal of Development Economics, Elsevier, 113(C): 33-51. Winkler, H. J., E. N. Ruppert Bulmer, H. K. Mote. 2017. “Expanding Social Insurance Coverage to Informal Workers (English).” Jobs Working Paper no. 6. Washington, D.C.: World Bank Group. World Bank. 2015a. World Bank Group Gender Strategy (FY16-23): Gender Equality, Poverty Reduction and Inclusive Growth. Washington, D.C.: World Bank Group. World Bank. 2015b. Volatility and Inequality as Constraints to Shared Prosperity: Paraguay Equity Assessment. Washington, 2015. World Bank. 2016. World development report 2016: digital dividends overview (English). World development report. Washington, D.C.: World Bank Group. World Bank. 2018a. Paraguay Systematic Country Diagnostic. Washington, D.C.: World Bank Group. World Bank. 2018b. Paraguay - Policy notes 2018: Paraguay - Notas de política 2018 (Spanish). Washington, D.C.: World Bank Group. 62 ANNEX A – DATA SOURCES AND VARIABLE DEFINITIONS Labor supply analysis using household-level data 1. Encuesta Permanente de Hogares The labor supply analysis relies primarily on microdata from the annual household income survey, Encuesta Permanente de Hogares (EPH), for 2001 through 2016, carried out by the Dirección General de Estadísticas, Encuestas y Censos. The total sample of households includes the 17 departments of the country plus Asunción, although the departments of Boquerón and Alto Paraguay were included only beginning in 201555. The sample size in 2016 was 37,800 individuals in 10,200 households. Reliable information on formality status is available only from 2008 onwards. A worker’s formality status is defined as follows. Informal work comprises (i) farmers, herders, and fishers (self-employed or employers of firms with no Registro Único de Contribuyente, or RUC); (ii) unpaid family workers; (iii) non-farmer self- employed, employee, or employer of firm with no RUC; and (iv) wage employees not contributing to Social Security. Formal employment comprises (i) wage employees contributing to Social Security; (ii) employers of a registered firm (with a RUC); and (iii) self-employed workers with a registered firm (with a RUC). Real monthly wages are calculated as reported monthly earnings from the main occupation transformed into 2005 USD PPP using SEDLAC CPI and PPP conversion factors. Wages include monetary income and all other income related with the job, such as bonuses and implicit rent/food/uniform value, received regularly. Hourly wage is computed as the monthly wage divided by weekly hours worked multiplied by 4.33. Annual wage is computed as the monthly wage multiplied by twelve. The variable on potential experience used in the Mincer regressions is calculated as age minus years of education plus six. Tenure is defined as the number of years in the current main occupation. Education categories are defined as follows: incomplete primary or less if less than 6 years of education; primary complete if with 6 years of education and not enrolled; secondary incomplete if with 6 years of education and enrolled or has (6-12) years of education; secondary complete if with 12 years of education and not enrolled; tertiary incomplete if with 12 years of education and enrolled or has (12-15) years of education; tertiary complete if with more than 15 years of education. 2. Encuesta Continua de Empleo The second main data source for household-level worker and employment characteristics is the labor force survey, Encuesta Continua de Empleo (ECE). We use the ECE panel component for 2010 to 2014 to analyze between-period labor transitions. The total ECE panel dataset includes 64,056 observations covering 20 quarters from 2010q1 to 2014q4. The geographic coverage includes Asunción as well as the urban areas in the Central department, representing about 40 percent of the national workforce and a little more than 60 percent of the urban workforce (2011 figures). The panel is unbalanced, i.e., data is not available for all quarters for all individuals. The panel rotation scheme is based on quarterly periodicity. A sample-selected household is interviewed in each of five consecutive quarters, after which it is dropped from the panel. A 55 Population of these departments represent 2% of the total country. 63 partial sample replacement is done every sixth quarter and the sample is completely changed at the end of 5 years (20 quarters). Information on the RUC is unavailable for years 2010-2011, but data on pension contributions are available for all years; we therefore revise the formality definition as follows. Workers defined to be formally employed include (i) wage employees contributing to a pension fund; (ii) employers contributing to a pension fund; and (iii) self-employed and domestic workers contributing to a pension fund. Workers defined as informally employed include (i) farmers/herders/fisherman (self-employed or employer not contributing to pension fund); (ii) unpaid family workers; (iii) self-employed, employee or employer not contributing to a pension fund; (iv) wage employees not contributing to a pension fund; and (v) domestic workers not contributing to a pension fund. Labor demand analysis using firm-level data There are four main data sources used to analyze private sector firm characteristics in Paraguay. Survey data was collected by the Dirección General de Estadística, Encuestas y Censos, and provides information at the establishment level, including employment level, annual sales, wage bill, value of capital assets, and value added. Employment is differentiated by gender and by remuneration method, i.e., paid, unpaid, and commission-based pay, but wages are reported at the establishment level (inclusive of all employees). Additional information is collected on firm characteristics such as formality status, sector, firm age, and firm location (at the district level). Only a small subset of firms has information on capital assets, and there is no information collected on workers’ education, skill level or occupation. The datasets are not perfectly comparable; the analysis therefore uses each source separately and provides caveats when making comparisons. 1. Censo Económico The 2011 Firm Census – Censo Económico – is a national census that includes data on firm characteristics such as employment, sales and value-added during the reference year 2010. The census covers all regions of Paraguay, and identifies 211,042 establishments employing 799,153 workers, including paid, unpaid and commissioned56 employees. The census includes both formal and informal firms, as well as single-person firms (i.e., self-employed). The census includes all sectors except agriculture. The number of firms reporting at least one paid employee is 51,457, and these employ a total of 537,025 workers (recall Table 2). The firm size thresholds defined by DGEEC are the following: 1-10 employees or annual sales up to PRY G 5,000,000; 11-30 employees or annual sales between PRY G 5,000,000 and PRY G 2,500,000,000; 31-50 employees or annual sales between PRY G 2,500,000,000 and PRY G 6,000,000,000; and 51 or more employees or annual sales over PRY G 6,000,000,000. For the purposes of this analysis, we define new firm- size categories to facilitate international comparisons (i.e., 1-9 paid employees; 10-19; 20-99, 100 and over). Data quality is largely good, with high response rates, although two key variables have large numbers of missing responses: firm age, and level of capital assets. • Only 57 percent of firms report their age. This may be driven by firms that operated informally prior to registering. Under-reporting of firm age may be motivated by evasion, or may simply be a result of uncertainty on the part of the respondent. The sectoral and geographical distribution of firms not reporting age do not appear to be biased. 56 Tercerizado or comisionista. 64 • Related to the age variable, no explicit data is collected on firm entry or exit. New entrants are therefore identified as those with age equal to 1 year. It is not possible to deduce firm exits. • There is a very low response rate for the level of capital assets (only two percent of firms report this information). This is likely due to the difficulty and time required to value firm assets. And for micro- sized firms in particular, respondents may be unclear about what constitutes a capital asset. Although we omit the variable from the regression analysis, due to the sharply reduced sample size, we test the variable for significance as a robustness check to see how its inclusion alters the regression estimates (for more detail, refer to Ruppert Bulmer and Scutaru 2018). 2. Encuesta de Empresas A Firm Survey – Encuesta de Empresas – was conducted in two waves in 2015 and 2016, on a national sample of formal (i.e., registered) firms drawn from the Firm Census. The firm survey excludes agriculture, mining, construction and the financial services sector, and is representative at the 2-digit level for manufacturing, commerce and non-financial services. The first survey wave comprised micro and small firms; the second wave comprised medium and large firms. The reference year is 2014. The survey of 6,523 firms covered the weighted equivalent of 52,727 firms with 439,612 employees. The survey response rate for micro and small firms was relatively low, and large firms – which account for the highest share of employment and output – are over-represented in the sample, but these sampling biases are corrected through survey weights. Similar to the firm census, there is relatively low reporting of capital assets. 3. Constructed Panel of Firms 2010-2014 To exploit the time variation in firms’ performance and growth between 2010 and 2014, a panel of 502 firms was created comprising formal firms with at least 1 paid employee and that appear in both the Firm Census and the Firm Survey (using the unique firm identification number to match firms). This panel dataset covers only manufacturing, commerce and non-financial services sectors. A comparison of summary statistics and the distribution of key variables – namely firm size, productivity and wages – shows that the panel is highly representative of both the Firm Census and the Firm Survey, although the panel firms are slightly larger, marginally more productive and pay modestly higher wages. The panel regression analysis estimates the correlates of the following dependent variables: the change in firm employment, the change in firm productivity, and the change in labor cost. 4. Pre-Censo A Pre-Census Survey – Pre-Censo – was conducted annually in reference to 2013, 2014 and 2015 to update the firm registry for a very large sample of firms drawn from the 2011 Firm Census. The pre-census comprised 89,000 firms in 2013, 88,000 firms in 2014, and 104,000 firms in 2015, accounting for about three-quarters of all registered firms. The sample is representative with respect to firm size, sector and region distributions. Firm entry and exit are only partly observed, because the dataset reflects a large sample rather than a complete census. 65 ANNEX B – LABOR SUPPLY ANALYSIS DETAILS Table B1. Logit estimates and Average Marginal Effects (AME), Age 15+, 2015 Inactive (0) Unemployed (0) Informal (0) Unemployed (0) to Active (1) to Employed (1) to Formal (1) to inactive (1) Estimate AME Estimate AME Estimate AME Estimate AME (1) (2) (3) (4) (5) (6) (7) (8) Age 0.326*** 0.013*** 0.262*** 0.006*** 0.256*** 0.009*** -0.139** -0.001 (0.024) (0.001) (0.052) (0.001) (0.049) (0.001) (0.059) (0.001) Age^2 -0.004*** -0.003*** -0.003*** 0.002*** (0.000) (0.001) (0.001) (0.001) Male 1.992*** 0.275*** 0.486*** 0.026*** 0.374*** 0.042*** -1.671*** -0.164*** (0.063) (0.008) (0.120) (0.006) (0.100) (0.011) (0.143) (0.013) Attending any education level -1.174*** -0.162*** 0.582*** 0.031*** -0.295* -0.033* 2.036*** 0.199*** (0.102) (0.014) (0.191) (0.010) (0.162) (0.018) (0.227) (0.021) Language most spoken at home (relative to Spanish) Guaraní 0.270** 0.038** 0.511*** 0.026*** -0.541*** -0.061*** 0.319 0.031 (0.106) (0.015) (0.186) (0.010) (0.148) (0.017) (0.203) (0.020) Guaraní and Spanish 0.167* 0.024* 0.105 0.006 -0.112 -0.013 0.058 0.006 (0.089) (0.013) (0.172) (0.011) (0.118) (0.014) (0.181) (0.019) Other -0.068 -0.010 1.170*** 0.047*** 0.599** 0.074** 0.920** 0.075*** (0.158) (0.023) (0.372) (0.012) (0.250) (0.032) (0.393) (0.027) Highest education level (relative to incomplete primary or less) Primary complete 0.343*** 0.055*** -0.350 -0.020 0.619*** 0.051*** -0.540** -0.043** (0.104) (0.017) (0.227) (0.013) (0.232) (0.018) (0.232) (0.018) Secondary incomplete 0.357*** 0.057*** -0.213 -0.011 1.388*** 0.134*** -0.506** -0.040** (0.111) (0.018) (0.200) (0.010) (0.228) (0.019) (0.231) (0.017) Secondary complete 0.824*** 0.125*** -0.025 -0.001 1.892*** 0.199*** -0.635** -0.052*** (0.121) (0.018) (0.220) (0.011) (0.234) (0.021) (0.250) (0.020) Tertiary incomplete 1.312*** 0.186*** -0.237 -0.013 2.687*** 0.316*** -1.331*** -0.133*** (0.144) (0.019) (0.241) (0.013) (0.253) (0.027) (0.278) (0.028) Tertiary complete 1.985*** 0.253*** 0.382 0.016 2.781*** 0.330*** -1.268*** -0.125*** (0.172) (0.019) (0.331) (0.013) (0.255) (0.028) (0.364) (0.042) 66 Logit estimates and Average Marginal Effects (AME), Age 15+, 2015 (continued) Inactive (0) Unemployed (0) Informal (0) Unemployed (0) to Active (1) to Employed (1) to Formal (1) to inactive (1) Estimate AME Estimate AME Estimate AME Estimate AME (1) (2) (3) (4) (5) (6) (7) (8) Household size -0.027 -0.004 0.001 0.000 -0.025 -0.003 0.038 0.004 (0.029) (0.004) (0.050) (0.003) (0.043) (0.005) (0.057) (0.006) Number children younger than 13 in hh -0.015 -0.002 0.020 0.001 0.045 0.005 0.003 0.000 (0.044) (0.006) (0.079) (0.004) (0.067) (0.007) (0.087) (0.009) Number youth in hh 0.149*** 0.021*** -0.012 -0.001 -0.003 -0.000 -0.175** -0.017** (0.042) (0.006) (0.072) (0.004) (0.066) (0.007) (0.076) (0.007) Number elder age>=60 in hh -0.037 -0.005 -0.247** -0.013** -0.032 -0.004 -0.082 -0.008 (0.059) (0.008) (0.110) (0.006) (0.093) (0.010) (0.109) (0.011) Urban -0.050 -0.007 -0.498*** -0.027*** 0.251** 0.028** -0.388*** -0.038*** (0.073) (0.010) (0.139) (0.007) (0.127) (0.014) (0.146) (0.014) Region (relative to Asunción) San Pedro 0.153 0.021 -0.283 -0.012 -0.925*** -0.104*** -0.526* -0.046 (0.149) (0.021) (0.282) (0.013) (0.209) (0.023) (0.317) (0.029) Caaguazú 0.006 0.001 0.480 0.015 -0.942*** -0.106*** 0.605* 0.037* (0.136) (0.019) (0.311) (0.009) (0.186) (0.021) (0.330) (0.020) Itapúa 0.308* 0.042* -0.189 -0.008 -0.294 -0.035 -0.561* -0.050 (0.161) (0.022) (0.322) (0.014) (0.216) (0.026) (0.338) (0.032) Alto Paraná 0.130 0.018 -1.179*** -0.075*** -0.473*** -0.056*** -1.164*** -0.123*** (0.122) (0.017) (0.221) (0.014) (0.167) (0.020) (0.246) (0.024) Central 0.111 0.015 -0.341 -0.015 -0.330** -0.039** -0.479* -0.042** (0.120) (0.017) (0.219) (0.009) (0.154) (0.019) (0.246) (0.021) Rest -0.042 -0.006 -0.382* -0.017** -0.549*** -0.064*** -0.340 -0.028 (0.110) (0.016) (0.197) (0.008) (0.138) (0.017) (0.226) (0.018) Secondary sector (relative to Retail) Agriculture, cattle and fishing -1.286*** -0.147*** (0.260) (0.026) Manufacture and mining -0.050 -0.007 (0.139) (0.019) 67 Logit estimates and Average Marginal Effects (AME), Age 15+, 2015 (continued) Inactive (0) Unemployed (0) Informal (0) Unemployed (0) to Active (1) to Employed (1) to Formal (1) to inactive (1) Estimate AME Estimate AME Estimate AME Estimate AME (1) (2) (3) (4) (5) (6) (7) (8) Electricity, gas and water 0.413 0.057 (0.418) (0.060) Construction -1.015*** -0.121*** (0.218) (0.024) Transport and communication 0.042 0.006 (0.211) (0.028) Finance and real state 0.221 0.030 (0.190) (0.026) Govt/public administration 1.063*** 0.155*** (0.188) (0.029) Other services -0.801*** -0.098*** (0.154) (0.018) Firm size (relative to [2-5]) Alone -0.470*** -0.052*** (0.138) (0.015) [6-10] 0.661*** 0.085*** (0.141) (0.019) [11-20] 0.991*** 0.132*** (0.164) (0.023) [21-50] 1.404*** 0.194*** (0.158) (0.023) [50+] 1.731*** 0.243*** (0.175) (0.026) Constant -5.827*** -1.810** -8.121*** 4.919*** (0.427) (0.861) (0.875) (0.969) Weighted observations 15893 15893 11479 11479 10248 10248 5164 5164 Pseudo R^2 0.267 0.106 0.411 0.189 68 Table B2. Logit regressions from Panel Data, Age 15+, Estimates and Average Marginal Effects (AME), quarterly transitions 2010-2014 Employed if Active if started as started as Formal if started as Inactive if started Inactive Unemployed or Informal as Unemployed Inactive Estimate AME Estimate AME Estimate AME Estimate AME Male 0.671*** 0.119*** 0.926*** 0.147*** -0.452*** -0.026*** -0.527*** -0.115*** (0.087) (0.015) (0.077) (0.012) (0.116) (0.007) (0.186) (0.040) Age group (relative to [15-19]) [20-24] 0.242* 0.045* 0.377*** 0.057*** 0.500** 0.024** -0.180 -0.039 (0.125) (0.024) (0.111) (0.017) (0.231) (0.010) (0.208) (0.045) [25-29] 0.416** 0.080** 0.760*** 0.126*** 0.755*** 0.039*** 0.058 0.013 (0.169) (0.034) (0.150) (0.026) (0.257) (0.012) (0.327) (0.072) [30-34] 0.043 0.008 0.714*** 0.117*** 0.820*** 0.043*** 0.037 0.008 (0.191) (0.034) (0.168) (0.030) (0.279) (0.014) (0.372) (0.082) [35-39] -0.187 -0.031 0.255 0.037 0.457 0.021 -0.075 -0.016 (0.193) (0.032) (0.177) (0.027) (0.289) (0.013) (0.467) (0.102) [40-44] -0.018 -0.003 0.708*** 0.116*** 0.779*** 0.040*** 0.482 0.107 (0.197) (0.034) (0.179) (0.032) (0.288) (0.015) (0.565) (0.124) [45-49] -0.298 -0.049* 0.282 0.041 0.912*** 0.049*** -0.203 -0.044 (0.185) (0.029) (0.175) (0.027) (0.290) (0.015) (0.569) (0.123) [50-54] -0.433 -0.069 0.060 0.008 0.993* 0.054 0.605 0.133 (0.295) (0.042) (0.265) (0.037) (0.514) (0.033) (0.711) (0.154) [55-59] 60 or more Attending any education level -1.032*** -0.184*** -0.844*** -0.134*** 0.227 0.013 0.806*** 0.176*** (0.126) (0.022) (0.113) (0.018) (0.172) (0.010) (0.255) (0.054) Highest education level (relative to incomplete primary or less) Primary complete 0.021 0.004 -0.056 -0.009 0.138 0.007 -0.060 -0.013 (0.187) (0.033) (0.175) (0.028) (0.394) (0.020) (0.586) (0.130) Secondary incomplete -0.179 -0.031 -0.184 -0.029 0.089 0.005 -0.022 -0.005 (0.172) (0.030) (0.161) (0.026) (0.384) (0.020) (0.521) (0.115) Secondary complete -0.025 -0.004 -0.101 -0.016 0.340 0.019 -0.536 -0.120 (0.175) (0.031) (0.163) (0.026) (0.380) (0.020) (0.527) (0.117) Tertiary incomplete 0.481** 0.094*** 0.226 0.038 0.332 0.018 -1.169** -0.255** (0.192) (0.036) (0.178) (0.030) (0.397) (0.021) (0.556) (0.120) Tertiary complete 0.462** 0.090** 0.322 0.056 0.311 0.017 -0.992 -0.219* (0.229) (0.045) (0.204) (0.035) (0.402) (0.021) (0.604) (0.131) 69 Logit regressions from Panel Data, Age 15+, Estimates and Average Marginal Effects (AME), quarterly transitions 2010-2014 Employed if Active if started as started as Formal if started as Inactive if started Inactive Unemployed or Informal as Unemployed Inactive Estimate AME Estimate AME Estimate AME Estimate AME Household size 0.006 0.001 -0.032 -0.005 0.111* 0.006* -0.066 -0.014 (0.037) (0.007) (0.033) (0.005) (0.060) (0.004) (0.076) (0.017) hh Number children younger than 14 in-0.023 -0.004 0.077* 0.012* -0.180** -0.010** 0.082 0.018 (0.050) (0.009) (0.047) (0.007) (0.083) (0.005) (0.128) (0.028) Number youth in hh 0.020 0.004 0.087* 0.014* -0.048 -0.003 0.039 0.008 (0.055) (0.010) (0.049) (0.008) (0.090) (0.005) (0.113) (0.025) Number elder age>=60 in hh 0.011 0.002 -0.040 -0.006 -0.185* -0.011* 0.188 0.041 (0.065) (0.012) (0.063) (0.010) (0.103) (0.006) (0.154) (0.033) Job Tenure 0.005 0.000 (0.010) (0.001) Secondary sector (relative to Retail) Agriculture, cattle and fishing -0.428 -0.026 (0.539) (0.029) Manufacture and mining 0.002 0.000 (0.171) (0.011) Electricity, gas and water 0.316 0.022 (0.683) (0.051) Construction -1.274*** -0.062*** (0.341) (0.013) Transport and communication -0.125 -0.008 (0.232) (0.015) Finance and real state 0.093 0.006 (0.199) (0.013) Govt/public administration -1.563*** -0.070*** (0.213) (0.009) Other services 0.198 0.014 (0.156) (0.011) Firm size at time t (relative to [2-5]) Alone -1.784*** -0.017*** (0.416) (0.003) [6-10] 1.308*** 0.050*** (0.212) (0.010) [11-20] 1.858*** 0.094*** (0.224) (0.016) [21-50] 2.123*** 0.123*** (0.224) (0.019) [50+] 3.575*** 0.370*** (0.183) (0.020) 70 Logit regressions from Panel Data, Age 15+, Estimates and Average Marginal Effects (AME), quarterly transitions 2010-2014 Employed if Active if started as started as Formal if started as Inactive if started Inactive Unemployed or Informal as Unemployed Inactive Estimate AME Estimate AME Estimate AME Estimate AME Year-quarter (relative to 2010q2) 2010q3 -0.249 -0.046 -0.324* -0.057* -0.404 -0.025 -0.226 -0.050 (0.184) (0.034) (0.169) (0.030) (0.268) (0.017) (0.471) (0.104) 2010q4 -0.019 -0.004 -0.098 -0.018 -0.735** -0.042** -0.362 -0.081 (0.186) (0.036) (0.167) (0.031) (0.294) (0.017) (0.456) (0.102) 2011q1 0.020 0.004 -0.473*** -0.081*** -0.095 -0.006 -0.083 -0.018 (0.184) (0.036) (0.177) (0.030) (0.261) (0.017) (0.450) (0.098) 2011q2 -0.132 -0.025 -0.137 -0.025 -0.293 -0.019 -0.489 -0.110 (0.326) (0.061) (0.283) (0.051) (0.416) (0.026) (0.613) (0.139) 2011q3 -0.335* -0.061* -0.579*** -0.096*** -0.689** -0.040** -0.973** -0.222** (0.202) (0.037) (0.185) (0.031) (0.310) (0.018) (0.454) (0.100) 2011q4 -0.162 -0.031 -0.344* -0.060* -0.761*** -0.043** -0.365 -0.082 (0.191) (0.036) (0.180) (0.031) (0.294) (0.017) (0.461) (0.103) 2012q1 -0.246 -0.046 -0.679*** -0.110*** -0.155 -0.010 -2.090*** -0.438*** (0.198) (0.037) (0.194) (0.031) (0.285) (0.019) (0.530) (0.095) 2012q2 -0.165 -0.031 -0.477*** -0.081*** -0.451 -0.028 -0.826* -0.188** (0.192) (0.036) (0.177) (0.030) (0.310) (0.019) (0.428) (0.095) 2012q3 -0.412 -0.074 -0.396* -0.069* 0.156 0.011 -1.306** -0.295** (0.265) (0.046) (0.234) (0.039) (0.393) (0.028) (0.652) (0.138) 2012q4 -0.197 -0.037 -0.427** -0.073** -0.113 -0.007 -0.912** -0.208** (0.198) (0.037) (0.183) (0.031) (0.311) (0.021) (0.453) (0.100) 2013q1 -0.187 -0.035 -0.499*** -0.084*** -0.309 -0.020 -1.191*** -0.270*** (0.201) (0.038) (0.188) (0.032) (0.308) (0.019) (0.451) (0.097) 2013q2 -0.194 -0.036 -0.189 -0.034 0.003 0.000 -0.826* -0.188* (0.197) (0.037) (0.177) (0.032) (0.278) (0.019) (0.451) (0.100) 2013q3 -0.216 -0.040 -0.368** -0.064** -0.545* -0.033* -0.878** -0.200** (0.195) (0.036) (0.179) (0.031) (0.279) (0.017) (0.445) (0.099) 2013q4 -0.409* -0.073* -0.587*** -0.097*** -0.318 -0.020 -1.560*** -0.346*** (0.229) (0.040) (0.202) (0.033) (0.298) (0.019) (0.543) (0.110) 2014q1 -0.158 -0.030 -0.466** -0.080** -0.405 -0.025 -1.515*** -0.337*** (0.202) (0.038) (0.184) (0.031) (0.305) (0.019) (0.495) (0.102) 2014q2 -0.713*** -0.119*** -0.578*** -0.096*** -0.014 -0.001 -1.352*** -0.304*** (0.219) (0.036) (0.193) (0.032) (0.274) (0.019) (0.464) (0.098) 2014q3 -0.242 -0.045 -0.438** -0.075** 0.040 0.003 -0.865** -0.197** (0.207) (0.038) (0.191) (0.032) (0.291) (0.020) (0.423) (0.093) 2014q4 -0.213 -0.040 -0.248 -0.044 -0.739** -0.042** -0.645 -0.146 (0.205) (0.038) (0.180) (0.032) (0.322) (0.018) (0.457) (0.102) Year-quarter (relative to 2015q2) 2015q3 2015q4 Constant -0.693*** -1.208*** -4.421*** 1.220* (0.250) (0.234) (0.510) (0.636) Weighted Observations 5481 5481 6989 6989 7476 7476 893 893 Pseudo R sq 0.058 0.073 0.300 0.089 71 Logit regressions from Panel Data, Age 15+, Estimates and Average Marginal Effects (AME), quarterly transitions Notes Wei ghted Logi t regres s i ons from Pa nel Da ta . Sta nda rd errors i n pa renthes es .* p<0.1 ** p<0.05 *** p<0.01 Incompl ete pri ma ry or l es s i f l es s tha n 6 yea rs of educa ti on; Pri ma ry compl ete i f wi th 6 yea rs of educa ti on a nd not enrol l ed; Seconda ry i ncompl ete i f wi th 6 yea rs of educa ti on a nd enrol l ed or ha s (6-12) yea rs of educa ti on; Seconda ry compl ete i f wi th 12 yea rs of educa ti on a nd not enrol l ed; Terti a ry i ncompl ete i f wi th 12 yea rs of educa ti on a nd enrol l ed or ha s (12-16) yea rs of educa ti on; Terti a ry compl ete i f wi th more tha n 15 yea rs of educa ti on. Yea rs of educa ti on ca l cul a ted by Pa ra gua y's Sta ti s ti ca l Depa rtment (DGEEC). As RUC da ta una va i l a bl e for yea rs 2010-2011 but pens i on contri buti ons a va i l a bl e for a l l yea rs then forma l i ty defi ni ti on i s cha nged for ECE pa nel da ta s et. Forma l i f (i ) wa ge empl oyees contri buti ng to a pens i on fund, (i i ) empl oyers contri buti ng to a pens i on fund, (i i i ) s el f-empl oyed a nd domes ti c workers contri buti ng to a pens i on fund; Informa l i f (i ) fa rmers /herders /fi s herma n (s el f-empl oyed or empl oyer not contri buti ng to pens i on fund), (i i ) unpa i d fa mi l y worker, (i i i ) s el f- empl oyed, empl oyee or empl oyer not contri buti ng to a pens i on fund, (i v) wa ge empl oyees not contri buti ng to a pens i on fund a nd (v) domes ti c workers not contri buti ng to a pens i on fund. La ngua ge s poken dropped for cl a ri ty of expos i ti on. Regres s i on s a mpl e too s ma l l . La ngua ge groups genera te bi a s ed es ti ma tes . Da ta des cri pti on. Tota l obs erva ti ons 62957, i ndi vi dua l s 19150, Qua rters 20, Yea rs 2010-2014. Geogra phi ca l l y i t covers As unci ón a nd Urba n a rea s i n the Centra l depa rtment, whi ch repres ents a bout 40% of the Na ti ona l workforce a nd a l i ttl e more tha n 60% of urba n workforce. Unba l a nced pa nel : da ta not a va i l a bl e for a l l qua rters for a l l i ndi vi dua l s . 72 Table B3. Logit estimates and Average Marginal Effects (AME), Age 15+, 2016 Inactive (0) Informal (0) to Employed (1) to Formal (1) Female Male Female Male Estimate AME Estimate AME Estimate AME Estimate AME (1) (2) (3) (4) (5) (6) (7) (8) Age 0.231*** 0.012*** 0.592*** 0.017*** 0.253*** 0.010*** 0.254*** 0.009*** (0.026) (0.001) (0.045) (0.001) (0.054) (0.001) (0.042) (0.001) Age^2 -0.003*** -0.008*** -0.002*** -0.003*** (0.000) (0.001) (0.001) (0.001) Attending any education level -0.503*** -0.102*** -1.607*** -0.116*** -0.002 -0.000 -0.200 -0.024 (0.104) (0.021) (0.137) (0.010) (0.185) (0.020) (0.159) (0.019) Language most spoken at home (relative to Spanish) Guaraní -0.098 -0.020 0.440*** 0.032*** -0.290* -0.031* -0.481*** -0.059*** (0.099) (0.020) (0.165) (0.012) (0.164) (0.018) (0.142) (0.018) Guaraní and Spanish -0.018 -0.004 0.156 0.012 -0.003 -0.000 -0.099 -0.013 (0.086) (0.017) (0.129) (0.010) (0.132) (0.014) (0.111) (0.014) Other -0.598*** -0.124*** -0.383 -0.032 -0.784** -0.080** 0.506 0.068 (0.167) (0.035) (0.401) (0.035) (0.393) (0.038) (0.308) (0.043) Highest education level (relative to incomplete primary or less) Primary complete 0.046 0.010 0.650*** 0.052*** 0.975*** 0.072*** 0.446** 0.045** (0.096) (0.021) (0.234) (0.019) (0.305) (0.022) (0.201) (0.020) Secondary incomplete 0.143 0.032 0.650*** 0.052*** 1.507*** 0.127*** 0.885*** 0.096*** (0.099) (0.022) (0.198) (0.016) (0.284) (0.021) (0.192) (0.019) Secondary complete 0.497*** 0.109*** 0.627*** 0.050*** 2.148*** 0.205*** 1.460*** 0.174*** (0.114) (0.025) (0.227) (0.019) (0.282) (0.022) (0.193) (0.020) Tertiary incomplete 0.882*** 0.187*** 0.897*** 0.069*** 2.811*** 0.298*** 2.039*** 0.262*** (0.126) (0.026) (0.244) (0.019) (0.289) (0.025) (0.212) (0.025) Tertiary complete 1.608*** 0.308*** 0.757** 0.059** 2.974*** 0.322*** 2.057*** 0.265*** (0.154) (0.026) (0.302) (0.023) (0.291) (0.026) (0.219) (0.027) Household size -0.037 -0.008 -0.052 -0.004 0.022 0.002 -0.037 -0.005 (0.030) (0.006) (0.046) (0.003) (0.057) (0.006) (0.048) (0.006) Number children younger than 13 in hh -0.073* -0.015* 0.292*** 0.021*** -0.177** -0.019** 0.023 0.003 (0.043) (0.009) (0.073) (0.005) (0.086) (0.009) (0.069) (0.008) Number youth in hh 0.084** 0.017** 0.050 0.004 -0.049 -0.005 0.053 0.006 (0.042) (0.008) (0.065) (0.005) (0.079) (0.008) (0.069) (0.008) Number elder age>=60 in hh 0.074 0.015 -0.189** -0.014** -0.275** -0.029** -0.051 -0.006 (0.061) (0.012) (0.086) (0.006) (0.113) (0.012) (0.093) (0.011) 73 Logit estimates and Average Marginal Effects (AME), Age 15+, 2016 Inactive (0) Informal (0) to Employed (1) to Formal (1) Female Male Female Male Estimate AME Estimate AME Estimate AME Estimate AME (1) (2) (3) (4) (5) (6) (7) (8) Urban 0.195*** 0.039*** -0.387*** -0.028*** 0.265* 0.028* 0.115 0.014 (0.072) (0.015) (0.123) (0.009) (0.144) (0.015) (0.117) (0.014) Region (relative to Asunción) San Pedro -0.471*** -0.095*** 0.092 0.007 0.134 0.015 -0.293 -0.035 (0.155) (0.031) (0.240) (0.018) (0.279) (0.031) (0.249) (0.029) Caaguazú -0.347** -0.070** -0.284 -0.023 -0.497** -0.054** -0.424* -0.049* (0.146) (0.029) (0.220) (0.018) (0.249) (0.027) (0.236) (0.027) Itapúa 0.059 0.011 0.490** 0.034** 0.015 0.002 0.406** 0.052** (0.139) (0.027) (0.201) (0.014) (0.228) (0.025) (0.190) (0.024) Alto Paraná -0.320*** -0.064*** 0.162 0.012 -0.227 -0.025 -0.013 -0.002 (0.123) (0.024) (0.179) (0.013) (0.195) (0.021) (0.172) (0.021) Central -0.220* -0.044* 0.182 0.013 -0.446** -0.048** 0.306* 0.039* (0.125) (0.025) (0.171) (0.013) (0.188) (0.020) (0.164) (0.020) Rest -0.320*** -0.064*** 0.116 0.009 -0.382** -0.041** -0.190 -0.023 (0.115) (0.023) (0.160) (0.012) (0.175) (0.019) (0.158) (0.019) Secondary sector (relative to Retail) Agriculture, cattle and fishing -1.904*** -0.188*** -1.431*** -0.175*** (0.309) (0.024) (0.174) (0.019) Manufacture and mining 0.102 0.013 0.093 0.014 (0.201) (0.026) (0.128) (0.019) Electricity, gas and water 0.305 0.040 -0.184 -0.027 (0.881) (0.117) (0.456) (0.065) Construction 0.757 0.102 -1.197*** -0.152*** (0.660) (0.094) (0.171) (0.020) Transport and communication 0.526 0.070 -0.177 -0.026 (0.458) (0.063) (0.178) (0.026) Finance and real state 0.685*** 0.092*** 0.426** 0.066** (0.213) (0.030) (0.180) (0.029) Govt/public administration 0.421** 0.055** 0.765*** 0.122*** (0.196) (0.027) (0.180) (0.030) Other services -0.868*** -0.099*** -0.821*** -0.110*** (0.165) (0.018) (0.227) (0.027) 74 Logit estimates and Average Marginal Effects (AME), Age 15+, 2016 Inactive (0) Informal (0) to Employed (1) to Formal (1) Female Male Female Male Estimate AME Estimate AME Estimate AME Estimate AME (1) (2) (3) (4) (5) (6) (7) (8) Firm size (relative to [2-5]) Alone -0.537*** -0.059*** -0.272** -0.031** (0.167) (0.018) (0.134) (0.015) [6-10] 0.832*** 0.112*** 0.663*** 0.089*** (0.194) (0.027) (0.132) (0.018) [11-20] 1.257*** 0.175*** 0.907*** 0.125*** (0.197) (0.029) (0.151) (0.022) [21-50] 1.768*** 0.254*** 1.441*** 0.208*** (0.210) (0.032) (0.157) (0.024) [50+] 1.869*** 0.270*** 2.004*** 0.299*** (0.214) (0.033) (0.163) (0.026) Constant -3.621*** -7.322*** -8.479*** -7.412*** (0.440) (0.708) (0.984) (0.773) Weighted observations 9053 9053 9154 9154 5141 5141 7687 7687 Pseudo R^2 0.129 0.375 0.428 0.361 Notes Logit regressions with weighted observations Incomplete primary or less if less than 6 years of education; Primary complete if with 6 years of education and not enrolled; Secondary incomplete if with 6 years of education and enrolled or has (6-12) years of education; Secondary complete if with 12 years of education and not enrolled; Tertiary incomplete if with 12 years of education and enrolled or has (12-16) years of education; Tertiary complete if with more than 15 years of education. Formal if (i) wage employees contributing to Social Security, (ii) employers of a registered firm (RUC), (iii) self-employed workers with a registered firm (RUC); Informal if (i) farmers/herders/fisherman (self-employed or employer of firm with no RUC), (ii) unpaid family worker, (iii) self-employed, employee or employer of firm with no RUC, (iv) wage employees not contributing to Social Security. Standard errors in parentheses.* p<0.1 ** p<0.05 *** p<0.01 Govt/public administration and Firm size [21-50] not estimable as perfectly predict formality. Alone category in Firm size includes self-employed and domestic employees. (+) Not estimable as perfectly predicts zero. Source: Own calculations based on SEDLAC data. 75 Table B4. Average Wages by Labor State, Sector and Education Level (2005 USD PPP), Age 15+, 2016, Full-time workers Monthly wages Hourly wages Labor state Sector Education level Female Male Female Male Farmer 250.2 413.6 1.35 2.00 Employer 1119.1 1605.7 4.68 7.12 Self-employed (non-farmer) 493.9 608.7 2.12 2.65 Informal wage 443.8 583.6 2.10 2.66 Formal wage private 720.0 781.4 3.58 3.54 Formal wage public 995.0 1023.6 5.35 5.05 Agriculture, cattle and fishing 309.9 578.8 1.58 2.66 Manufacture and mining 540.8 690.3 2.45 3.09 Electricity, gas and water 1663.3 1529.2 9.48 7.96 Construction 625.3 591.8 3.35 2.81 Retail, restaurant and hotels 536.5 670.8 2.30 2.87 Transport and communication 691.9 786.9 3.42 3.39 Finance and real state 856.8 858.5 4.37 4.14 Govt/public administration 931.3 983.3 5.03 4.87 Other services 457.8 1014.2 2.18 4.64 of which domestic workers 304.1 424.7 2.00 2.29 Primary incomplete or less 322.0 481.1 1.41 2.15 Primary complete 401.4 579.0 1.73 2.57 Secondary incomplete 419.1 618.0 1.84 2.74 Secondary complete 516.3 627.9 2.30 2.77 Tertiary incomplete 602.1 947.5 2.87 4.37 Tertiary complete 1043.7 1425.9 5.49 7.01 Source: EPH data 76 Table B5 Mincer Regressions, OLS, Dep. variable Log Monthly real wage (2005 USD PPP), Paid workers age 15+, 2015 (1) (2) (3) (4) (5) Experience 0.042*** 0.053*** 0.047*** 0.047*** 0.038*** (0.003) (0.003) (0.003) (0.003) (0.003) Experience^2 -0.000*** -0.001*** -0.001*** -0.001*** -0.000*** (0.000) (0.000) (0.000) (0.000) (0.000) Male 0.426*** 0.469*** 0.423*** 0.320*** 0.397*** (0.021) (0.021) (0.024) (0.022) (0.023) Formal 0.536*** 0.417*** 0.477*** (0.023) (0.027) (0.026) Public 0.228*** (0.027) Job tenure (years) -0.002 -0.001 0.006*** -0.002* 0.004*** (0.001) (0.001) (0.002) (0.001) (0.001) Language most spoken at home (relative to Spanish) Guaraní -0.280*** -0.357*** -0.266*** -0.270*** -0.212*** (0.039) (0.040) (0.040) (0.039) (0.039) Guaraní and Spanish -0.035 -0.071*** -0.079*** -0.037 -0.050* (0.026) (0.026) (0.026) (0.026) (0.025) Other 0.511*** 0.505*** 0.606*** 0.499*** 0.604*** (0.072) (0.076) (0.074) (0.071) (0.071) Does not speak -0.051 0.154*** 0.201*** -0.204** 0.012 (0.109) (0.044) (0.043) (0.088) (0.098) Highest education level (relative to incomplete primary or less) Primary complete 0.134*** 0.157*** 0.101** 0.113*** 0.088** (0.043) (0.043) (0.041) (0.041) (0.041) Secondary incomplete 0.280*** 0.363*** 0.257*** 0.269*** 0.196*** (0.048) (0.049) (0.048) (0.047) (0.047) Secondary complete 0.457*** 0.610*** 0.476*** 0.427*** 0.360*** (0.050) (0.051) (0.055) (0.051) (0.052) Tertiary incomplete 0.631*** 0.851*** 0.694*** 0.543*** 0.511*** (0.053) (0.054) (0.057) (0.054) (0.055) Tertiary complete 0.928*** 1.198*** 1.024*** 0.854*** 0.799*** (0.055) (0.056) (0.059) (0.056) (0.058) Region (relative to Asunción) San Pedro -0.435*** -0.522*** -0.366*** -0.443*** -0.301*** (0.050) (0.052) (0.051) (0.050) (0.050) Caaguazú -0.371*** -0.454*** -0.366*** -0.393*** -0.300*** (0.044) (0.047) (0.045) (0.045) (0.043) Itapúa -0.203*** -0.259*** -0.214*** -0.216*** -0.172*** (0.052) (0.054) (0.053) (0.051) (0.052) Alto Paraná -0.064* -0.107*** -0.120*** -0.099*** -0.078** (0.036) (0.037) (0.036) (0.037) (0.035) Central -0.090*** -0.108*** -0.109*** -0.127*** -0.093*** (0.033) (0.034) (0.034) (0.034) (0.032) Rest -0.323*** -0.388*** -0.313*** -0.349*** -0.263*** (0.031) (0.033) (0.032) (0.032) (0.030) 77 Mincer Regressions, OLS, Dep. variable Log Monthly real wage (2005 USD PPP), Paid workers age 15+, 2015 (1) (2) (3) (4) (5) Sector (relative to Retail) Agriculture, cattle and fishing -0.619*** -0.552*** (0.051) (0.050) Manufacture and mining 0.014 -0.007 (0.036) (0.035) Electricity, gas and water 0.544*** 0.429*** (0.091) (0.088) Construction 0.122*** 0.187*** (0.036) (0.036) Transport and communication 0.206*** 0.176*** (0.043) (0.044) Finance and real state 0.107** 0.067 (0.043) (0.042) Govt/public administration 0.175*** 0.028 (0.031) (0.031) Other services -0.196*** -0.124*** (0.033) (0.033) Firm size (relative to [2-5]) Alone -0.381*** (0.030) [6-10] 0.165*** (0.029) [11-20] 0.161*** (0.035) [21-50] 0.087*** (0.032) [50+] 0.124*** (0.035) Constant 4.741*** 4.614*** 4.845*** 4.854*** 4.923*** (0.088) (0.092) (0.097) (0.094) (0.092) Weighted observations 11838 11845 11845 11319 11838 R sq 0.384 0.345 0.387 0.419 0.417 Notes OLS regres s i ons wi th robus t s ta nda rd errors a nd i ncome wei ghts . Sta nda rd errors i n pa renthes es .* p<0.1 ** p<0.05 *** p<0.01 Wa ges ca l cul a ted a s a vera ge of monthl y ea rni ngs from ma i n occupa ti on tra ns formed i n 2005 USD PPP us i ng SEDLAC CPI a nd PPP convers i on fa ctors . Incl ude moneta ry i ncome a nd a l l other i ncome rel a ted wi th the job, l i ke bonus es a nd i mpl i ci t rent/food/uni form va l ue, recei ved regul a rl y. Tra ns formed i n 2005 USD PPP us i ng SEDLAC CPI a nd PPP convers i on fa ctors . Experi ence = Age - Yea rs of educa ti on + 6 Forma l i f (i ) wa ge empl oyees contri buti ng to Soci a l Securi ty, (i i ) empl oyers of a regi s tered fi rm (RUC), (i i i ) s el f-empl oyed workers wi th a regi s tered fi rm (RUC); Informa l i f (i ) fa rmers /herders /fi s herma n (s el f-empl oyed or empl oyer of fi rm wi th no RUC), (i i ) unpa i d fa mi l y worker, (i i i ) s el f-empl oyed, empl oyee or empl oyer of fi rm wi th no RUC, (i v) wa ge empl oyees not contri Tenure buti defi nga ned to Soci a l Securi s number of yeaty. rs i n ma i n occupa ti on. Incompl ete pri ma ry or l es s i f l es s tha n 6 yea rs of educa ti on; Pri ma ry compl ete i f wi th 6 yea rs of educa ti on a nd not enrol l ed; Seconda ry i ncompl ete i f wi th 6 yea rs of educa ti on a nd enrol l ed or ha s (6-12) yea rs of educa ti on; Seconda ry compl ete i f wi th 12 yea rs of educa ti on a nd not enrol l ed; Terti a ry i ncompl ete i f wi th 12 yea rs of educa ti on a nd enrol l ed or ha s (12-15) yea rs of educa ti on; Terti a ry compl ete i f wi th more tha n 15 yea rs of educa ti on. To expl ore for s el ecti on bi a s we es ti ma ted Heckma n model s . Res ul ts a re i denti ca l to OLS a nd ca n be s uppl i ed by reques t. Source: Sta ff ca l cul a ti ons ba s ed on EPH da ta . 78 Table B6 Mincer Regressions, OLS, Dep. variable Log hourly real wage (2005 USD PPP), full-time wage workers age 15+, 2015 (1) (2) (3) (4) (5) Experience 0.033*** 0.038*** 0.036*** 0.033*** 0.030*** (0.004) (0.004) (0.004) (0.004) (0.004) Experience^2 -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** (0.000) (0.000) (0.000) (0.000) (0.000) Male 0.193*** 0.211*** 0.138*** 0.144*** 0.138*** (0.020) (0.020) (0.023) (0.024) (0.022) Formal 0.274*** 0.194*** 0.235*** (0.022) (0.026) (0.023) Public 0.239*** (0.030) Job tenure (years) 0.010*** 0.010*** 0.010*** 0.010*** 0.009*** (0.002) (0.002) (0.002) (0.002) (0.002) Language most spoken at home (relative to Spanish) Guaraní -0.145*** -0.179*** -0.175*** -0.116*** -0.151*** (0.032) (0.033) (0.032) (0.034) (0.032) Guaraní and Spanish -0.112*** -0.134*** -0.132*** -0.092*** -0.118*** (0.024) (0.025) (0.024) (0.026) (0.024) Other 0.278*** 0.258*** 0.283*** 0.297*** 0.302*** (0.074) (0.081) (0.085) (0.073) (0.078) Does not speak -0.234*** -0.061 -0.056 -0.347*** -0.179*** (0.048) (0.050) (0.044) (0.048) (0.037) Highest education level (relative to incomplete primary or less) Primary complete 0.052 0.076* 0.074* 0.055 0.060 (0.040) (0.040) (0.040) (0.041) (0.040) Secondary incomplete 0.116*** 0.164*** 0.158*** 0.117*** 0.119*** (0.040) (0.041) (0.042) (0.042) (0.040) Secondary complete 0.287*** 0.367*** 0.360*** 0.285*** 0.289*** (0.043) (0.044) (0.044) (0.045) (0.043) Tertiary incomplete 0.506*** 0.584*** 0.569*** 0.490*** 0.468*** (0.046) (0.047) (0.047) (0.047) (0.047) Tertiary complete 0.865*** 0.952*** 0.913*** 0.838*** 0.804*** (0.051) (0.053) (0.054) (0.054) (0.053) Region (relative to Asunción) San Pedro -0.251*** -0.317*** -0.283*** -0.271*** -0.248*** (0.047) (0.048) (0.047) (0.048) (0.046) Caaguazú -0.255*** -0.309*** -0.292*** -0.268*** -0.256*** (0.044) (0.046) (0.045) (0.046) (0.043) Itapúa -0.145*** -0.172*** -0.173*** -0.153*** -0.152*** (0.048) (0.047) (0.046) (0.049) (0.046) Alto Paraná -0.066* -0.083** -0.073** -0.092** -0.051 (0.034) (0.035) (0.034) (0.036) (0.033) Central -0.072** -0.078** -0.071** -0.111*** -0.063** (0.032) (0.033) (0.033) (0.034) (0.032) Rest -0.148*** -0.185*** -0.166*** -0.178*** -0.143*** (0.031) (0.032) (0.031) (0.033) (0.030) 79 Mincer Regressions, OLS, Dep. variable Log hourly real wage (2005 USD PPP), full-time wage workers age 15+, 2015 (1) (2) (3) (4) (5) Sector (relative to Retail) Agriculture, cattle and fishing -0.026 -0.007 (0.038) (0.037) Manufacture and mining 0.098*** 0.081*** (0.030) (0.029) Electricity, gas and water 0.567*** 0.518*** (0.102) (0.100) Construction 0.160*** 0.189*** (0.035) (0.036) Transport and communication 0.143*** 0.128*** (0.048) (0.048) Finance and real state 0.100** 0.079* (0.043) (0.041) Govt/public administration 0.280*** 0.223*** (0.032) (0.032) Other services -0.129*** -0.090** (0.035) (0.035) Firm size (relative to [2-5]) Alone -0.127*** (0.040) [6-10] 0.108*** (0.028) [11-20] 0.174*** (0.035) [21-50] 0.137*** (0.030) [50+] 0.241*** (0.035) Constant -0.043 -0.082 -0.048 -0.055 0.014 (0.077) (0.081) (0.083) (0.085) (0.081) Weighted observations 5219 5219 5219 4812 5219 R sq 0.433 0.413 0.434 0.453 0.456 Notes OLS regres s i ons wi th robus t s ta nda rd errors a nd i ncome wei ghts . Sta nda rd errors i n pa renthes es .* p<0.1 ** p<0.05 *** p<0.01 Wa ges ca l cul a ted a s a vera ge of monthl y ea rni ngs from ma i n occupa ti on tra ns formed i n 2005 USD PPP us i ng SEDLAC CPI a nd PPP convers i on fa ctors . Incl ude moneta ry i ncome a nd a l l other i ncome rel a ted wi th the job, l i ke bonus es a nd i mpl i ci t rent/food/uni form va l ue, recei ved regul a rl y. Tra ns formed i n 2005 USD PPP us i ng SEDLAC CPI a nd PPP convers i on fa ctors . Experi ence = Age - Yea rs of educa ti on + 6 Forma l i f (i ) wa ge empl oyees contri buti ng to Soci a l Securi ty, (i i ) empl oyers of a regi s tered fi rm (RUC), (i i i ) s el f-empl oyed workers wi th a regi s tered fi rm (RUC); Informa l i f (i ) fa rmers /herders /fi s herma n (s el f-empl oyed or empl oyer of fi rm wi th no RUC), (i i ) unpa i d fa mi l y worker, (i i i ) s el f-empl oyed, empl oyee or empl oyer of fi rm wi th no RUC, (i v) wa ge empl oyees not contri buti ng to Soci a l Securi Tenure ty. a s number of yea rs i n ma i n occupa ti on. defi ned Incompl ete pri ma ry or l es s i f l es s tha n 6 yea rs of educa ti on; Pri ma ry compl ete i f wi th 6 yea rs of educa ti on a nd not enrol l ed; Seconda ry i ncompl ete i f wi th 6 yea rs of educa ti on a nd enrol l ed or ha s (6-12) yea rs of educa ti on; Seconda ry compl ete i f wi th 12 yea rs of educa ti on a nd not enrol l ed; Terti a ry i ncompl ete i f wi th 12 yea rs of educa ti on a nd enrol l ed or ha s (12-15) yea rs of educa ti on; Terti a ry compl ete i f wi th more tha n 15 yea rs of educa ti on. To expl ore for s el ecti on bi a s we es ti ma ted Heckma n model s . Res ul ts a re i denti ca l to OLS a nd ca n be s uppl i ed by reques t. Source: Sta ff ca l cul a ti ons ba s ed on EPH da ta . 80 Table B7 Mincer Regressions, OLS, Dependent variable Log monthly real wage (2005 USD PPP), Paid workers age 15+ 2015 Male Female Male Female Male Female Male Female Male Female (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Experience 0.049*** 0.034*** 0.060*** 0.043*** 0.054*** 0.038*** 0.053*** 0.038*** 0.044*** 0.029*** (0.004) (0.005) (0.004) (0.005) (0.004) (0.005) (0.004) (0.005) (0.004) (0.005) Experience^2 -0.001*** -0.000*** -0.001*** -0.000*** -0.001*** -0.000*** -0.001*** -0.000*** -0.000*** -0.000*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Formal 0.474*** 0.633*** 0.387*** 0.484*** 0.439*** 0.543*** (0.029) (0.040) (0.034) (0.045) (0.032) (0.042) Public -0.002 -0.004* -0.002 -0.002 0.004* 0.010*** -0.002 -0.004** 0.002 0.007*** (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) Job tenure 0.128*** 0.325*** (years) (0.035) (0.043) Language most spoken at home (relative to Spanish) Guaraní -0.289*** -0.276*** -0.368*** -0.340*** -0.283*** -0.244*** -0.262*** -0.271*** -0.222*** -0.204*** (0.051) (0.057) (0.053) (0.059) (0.053) (0.056) (0.051) (0.058) (0.051) (0.055) Guaraní and Spanish -0.022 -0.047 -0.054* -0.094** -0.066** -0.098** -0.015 -0.053 -0.038 -0.066 (0.032) (0.042) (0.032) (0.043) (0.031) (0.042) (0.033) (0.042) (0.031) (0.041) Other 0.563*** 0.337** 0.539*** 0.374** 0.649*** 0.438*** 0.568*** 0.324** 0.662*** 0.404*** (0.084) (0.137) (0.087) (0.153) (0.089) (0.134) (0.082) (0.138) (0.086) (0.126) Does not speak -0.034 0.211*** 0.180*** -0.201 -0.034 (0.165) (0.076) (0.070) (0.143) (0.137) Highest education level (relative to incomplete primary or less) Primary complete 0.099* 0.211*** 0.125** 0.226*** 0.080* 0.118 0.073 0.188** 0.062 0.117 (0.051) (0.076) (0.052) (0.078) (0.048) (0.077) (0.050) (0.074) (0.048) (0.075) Secondary incomplete 0.232*** 0.391*** 0.313*** 0.465*** 0.229*** 0.293*** 0.230*** 0.356*** 0.168*** 0.237*** (0.060) (0.080) (0.061) (0.081) (0.060) (0.080) (0.059) (0.080) (0.059) (0.078) Secondary complete 0.444*** 0.499*** 0.586*** 0.661*** 0.483*** 0.460*** 0.443*** 0.416*** 0.374*** 0.335*** (0.062) (0.084) (0.064) (0.085) (0.069) (0.085) (0.063) (0.084) (0.065) (0.084) Tertiary incomplete 0.547*** 0.737*** 0.758*** 0.973*** 0.651*** 0.702*** 0.506*** 0.588*** 0.485*** 0.498*** (0.068) (0.085) (0.069) (0.086) (0.073) (0.087) (0.068) (0.085) (0.072) (0.086) Tertiary complete 0.876*** 0.987*** 1.135*** 1.271*** 1.010*** 0.989*** 0.864*** 0.851*** 0.801*** 0.751*** (0.069) (0.089) (0.068) (0.091) (0.072) (0.091) (0.070) (0.089) (0.071) (0.091) 81 Mincer Regressions, OLS, Dependent variable Log monthly real wage (2005 USD PPP), Paid workers age 15+ 2015 Male Female Male Female Male Female Male Female Male Female (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Region (relative to Asunción) San Pedro -0.432*** -0.449*** -0.513*** -0.532*** -0.368*** -0.362*** -0.437*** -0.456*** -0.300*** -0.318*** (0.063) (0.085) (0.064) (0.089) (0.065) (0.083) (0.063) (0.085) (0.063) (0.081) Caaguazú -0.338*** -0.422*** -0.408*** -0.528*** -0.331*** -0.429*** -0.381*** -0.405*** -0.271*** -0.359*** (0.056) (0.073) (0.059) (0.077) (0.058) (0.071) (0.058) (0.073) (0.056) (0.069) Itapúa -0.171** -0.256*** -0.215*** -0.333*** -0.184** -0.295*** -0.189*** -0.251*** -0.149** -0.248*** (0.074) (0.070) (0.076) (0.074) (0.076) (0.071) (0.073) (0.067) (0.074) (0.066) Alto Paraná -0.017 -0.133** -0.057 -0.183*** -0.077* -0.180*** -0.065 -0.154** -0.038 -0.141** (0.044) (0.060) (0.046) (0.063) (0.045) (0.060) (0.048) (0.060) (0.044) (0.058) Central -0.084** -0.099* -0.088** -0.142** -0.105** -0.133** -0.157*** -0.090* -0.100** -0.106** (0.042) (0.051) (0.043) (0.055) (0.043) (0.054) (0.044) (0.052) (0.042) (0.051) Rest -0.302*** -0.357*** -0.353*** -0.446*** -0.286*** -0.377*** -0.343*** -0.354*** -0.244*** -0.321*** (0.041) (0.046) (0.043) (0.050) (0.043) (0.048) (0.043) (0.047) (0.041) (0.046) Secondary sector (relative to Retail) Agriculture, cattle and fishing -0.472*** -1.100*** -0.412*** -1.015*** (0.059) (0.096) (0.058) (0.095) Manufacture and mining 0.102** -0.175** 0.086** -0.200*** (0.040) (0.081) (0.039) (0.075) Electricity, gas and water 0.558*** 0.531*** 0.471*** 0.353** (0.110) (0.156) (0.105) (0.158) Construction 0.160*** 0.254*** 0.226*** 0.272 (0.041) (0.081) (0.041) (0.179) Transport and communication 0.240*** 0.116 0.230*** -0.009 (0.049) (0.107) (0.049) (0.108) Finance and real state 0.129** 0.068 0.101** 0.007 (0.051) (0.078) (0.050) (0.077) Govt/public administration 0.135*** 0.181*** 0.014 -0.002 (0.042) (0.049) (0.041) (0.049) Other services -0.222*** -0.248*** -0.160*** -0.168*** (0.062) (0.041) (0.061) (0.041) 82 Mincer Regressions, OLS, Dependent variable Log monthly real wage (2005 USD PPP), Paid workers age 15+ 2015 Male Female Male Female Male Female Male Female Male Female (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Firm size (relative to [2-5]) Alone -0.404*** -0.343*** (0.042) (0.045) [6-10] 0.195*** 0.101* (0.034) (0.055) [11-20] 0.209*** 0.104* (0.043) (0.061) [21-50] 0.091** 0.101* (0.038) (0.056) [50+] 0.105** 0.190*** (0.044) (0.057) Constant 5.068*** 4.850*** 4.962*** 4.774*** 5.091*** 5.143*** 5.052*** 4.999*** 5.152*** 5.204*** (0.114) (0.126) (0.120) (0.132) (0.122) (0.136) (0.118) (0.132) (0.116) (0.132) Weighted observations 7195 4643 7201 4644 7201 4644 6876 4443 7195 4643 R sq 0.356 0.409 0.322 0.362 0.365 0.420 0.395 0.438 0.393 0.456 Notes OLS regres s i ons wi th robus t s tanda rd errors a nd i ncome wei ghts . Standa rd errors i n pa renthes es .* p<0.1 ** p<0.05 *** p<0.01 Wa ges ca l cul a ted a s a vera ge of monthl y ea rni ngs from ma i n occupa tion tra ns formed i n 2005 USD PPP us i ng SEDLAC CPI a nd PPP convers i on fa ctors . Incl ude monetary i ncome a nd a l l other i ncome rel a ted wi th the job, l i ke bonus es a nd i mpl i ci t rent/food/uni form va l ue, recei ved regul a rl y. Tra ns formed i n 2005 USD PPP us i ng SEDLAC CPI a nd PPP convers i on fa ctors . Experi ence = Age - Yea rs of educa tion + 6 Forma l i f (i ) wa ge empl oyees contri buting to Soci a l Securi ty, (i i ) empl oyers of a regi s tered fi rm (RUC), (i i i ) s el f-empl oyed workers wi th a regi s tered fi rm (RUC); Informa l i f (i ) fa rmers /herders /fi s herma n (s el f-empl oyed or empl oyer of fi rm wi th no RUC), (i i ) unpa i d fa mi l y worker, (i i i ) s el f- empl oyed, empl oyee or empl oyer of fi rm wi th no RUC, (i v) wa ge empl oyees not contri buting to Soci a l Securi ty. Tenure defi ned a s number of yea rs i n ma i n occupa tion. Incompl ete pri ma ry or l es s i f l es s tha n 6 yea rs of educa tion; Pri ma ry compl ete i f wi th 6 yea rs of educa tion a nd not enrol l ed; Seconda ry i ncompl ete i f wi th 6 yea rs of educa tion a nd enrol l ed or ha s (6-12) yea rs of educa tion; Seconda ry compl ete i f wi th 12 yea rs of educa tion a nd not enrol l ed; Tertia ry i ncompl ete i f wi th 12 yea rs of educa tion a nd enrol l ed or ha s (12-15) yea rs of educa tion; Tertia ry compl ete i f wi th more tha n 15 yea rs of educa To expl tion. ore for s el ection bi a s we a l s o es tima ted Heckma n model s . Res ul ts a re i dentica l to OLS a nd ca n be s uppl i ed by reques t. Source: Own ca l cul a tions ba s ed on EPH da ta. 83 Table B8 Mincer Regressions, OLS, Dependent variable log monthly real wage (2005 USD PPP), full-time wage workers age 15+, 2015 Male Female Male Female Male Female Male Female Male Female (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Experience 0.036*** 0.028*** 0.042*** 0.034*** 0.040*** 0.033*** 0.037*** 0.026*** 0.034*** 0.026*** (0.004) (0.006) (0.004) (0.006) (0.004) (0.006) (0.004) (0.006) (0.004) (0.006) Experience^2 -0.000*** -0.000*** -0.001*** -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** -0.000*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Formal 0.243*** 0.298*** 0.174*** 0.212*** 0.228*** 0.239*** (0.023) (0.034) (0.026) (0.043) (0.024) (0.038) Public 0.134*** 0.186*** (0.033) (0.045) Job tenure 0.011*** 0.003 0.011*** 0.004* 0.011*** 0.005** 0.010*** 0.004 0.010*** 0.003 (years) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) Language most spoken at home (relative to Spanish) Guaraní -0.132*** -0.159*** -0.166*** -0.184*** -0.156*** -0.162*** -0.114*** -0.133*** -0.130*** -0.146*** (0.037) (0.046) (0.037) (0.050) (0.037) (0.048) (0.040) (0.047) (0.037) (0.046) Guaraní and Spanish -0.082*** -0.104*** -0.097*** -0.133*** -0.095*** -0.122*** -0.070** -0.089** -0.084*** -0.102*** (0.029) (0.034) (0.029) (0.036) (0.028) (0.036) (0.030) (0.037) (0.028) (0.034) Other 0.322*** 0.033 0.300*** 0.015 0.311*** 0.024 0.349*** 0.021 0.330*** 0.045 (0.077) (0.073) (0.084) (0.081) (0.088) (0.078) (0.074) (0.078) (0.081) (0.074) Does not speak -0.269*** -0.104** -0.124*** -0.356*** -0.240*** (0.044) (0.044) (0.046) (0.057) (0.050) Highest education level (relative to incomplete primary or less) Primary complete 0.039 0.212*** 0.059 0.231*** 0.059 0.196*** 0.034 0.199*** 0.045 0.195*** (0.043) (0.067) (0.044) (0.069) (0.043) (0.068) (0.045) (0.068) (0.043) (0.067) Secondary incomplete 0.079* 0.312*** 0.127*** 0.350*** 0.121*** 0.296*** 0.083* 0.266*** 0.081* 0.273*** (0.042) (0.064) (0.042) (0.068) (0.043) (0.066) (0.044) (0.064) (0.043) (0.063) Secondary complete 0.256*** 0.444*** 0.335*** 0.527*** 0.323*** 0.447*** 0.261*** 0.372*** 0.254*** 0.389*** (0.044) (0.068) (0.045) (0.070) (0.046) (0.069) (0.046) (0.071) (0.045) (0.068) Tertiary incomplete 0.406*** 0.665*** 0.497*** 0.767*** 0.481*** 0.665*** 0.397*** 0.566*** 0.387*** 0.573*** (0.049) (0.068) (0.050) (0.071) (0.052) (0.072) (0.052) (0.073) (0.053) (0.070) Tertiary complete 0.747*** 0.947*** 0.856*** 1.050*** 0.821*** 0.939*** 0.729*** 0.851*** 0.714*** 0.846*** (0.055) (0.074) (0.056) (0.077) (0.059) (0.077) (0.059) (0.081) (0.058) (0.076) 84 Mincer Regressions, OLS, Dependent variable log monthly real wage (2005 USD PPP), full-time wage workers age 15+, 2015 Male Female Male Female Male Female Male Female Male Female (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Region (relative to Asunción) San Pedro -0.260*** -0.219*** -0.309*** -0.281*** -0.296*** -0.241*** -0.279*** -0.231*** -0.254*** -0.225*** (0.053) (0.074) (0.054) (0.078) (0.055) (0.077) (0.055) (0.074) (0.053) (0.074) Caaguazú -0.224*** -0.341*** -0.261*** -0.411*** -0.259*** -0.372*** -0.234*** -0.344*** -0.223*** -0.343*** (0.052) (0.071) (0.054) (0.076) (0.054) (0.073) (0.054) (0.074) (0.053) (0.069) Itapúa -0.097* -0.205*** -0.117** -0.244*** -0.117** -0.242*** -0.111** -0.210*** -0.099* -0.219*** (0.054) (0.070) (0.055) (0.072) (0.054) (0.068) (0.055) (0.067) (0.053) (0.066) Alto Paraná -0.029 -0.008 -0.049 -0.027 -0.048 -0.026 -0.072* -0.018 -0.023 -0.010 (0.038) (0.050) (0.040) (0.051) (0.040) (0.051) (0.040) (0.052) (0.038) (0.050) Central -0.047 -0.051 -0.053 -0.063 -0.057 -0.052 -0.095** -0.070 -0.048 -0.048 (0.036) (0.049) (0.037) (0.051) (0.037) (0.050) (0.038) (0.052) (0.036) (0.049) Rest -0.097*** -0.221*** -0.120*** -0.276*** -0.116*** -0.245*** -0.142*** -0.233*** -0.094*** -0.221*** (0.034) (0.048) (0.036) (0.050) (0.036) (0.049) (0.036) (0.050) (0.035) (0.048) Secondary sector (relative to Retail) Agriculture, cattle and fishing -0.022 -0.138 -0.006 -0.118 (0.040) (0.141) (0.039) (0.137) Manufacture and mining 0.062* 0.010 0.046 -0.008 (0.032) (0.053) (0.031) (0.050) Electricity, gas and water 0.457*** 0.380** 0.415*** 0.316* (0.124) (0.174) (0.120) (0.174) Construction 0.057 0.143 0.084** 0.154 (0.035) (0.094) (0.035) (0.138) Transport and communication 0.191*** -0.018 0.178*** -0.053 (0.045) (0.081) (0.045) (0.089) Finance and real state 0.098** 0.040 0.073* 0.028 (0.042) (0.070) (0.040) (0.069) Govt/public administration 0.192*** 0.140*** 0.133*** 0.088* (0.037) (0.051) (0.035) (0.051) Other services -0.106 -0.199*** -0.097 -0.145*** (0.066) (0.042) (0.064) (0.044) 85 Mincer Regressions, OLS, Dependent variable log monthly real wage (2005 USD PPP), full-time wage workers age 15+, 2015 Male Female Male Female Male Female Male Female Male Female (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Firm size (relative to [2-5]) Alone -0.088 -0.131*** (0.137) (0.045) [6-10] 0.143*** 0.063 (0.029) (0.047) [11-20] 0.209*** 0.130** (0.033) (0.063) [21-50] 0.162*** 0.130** (0.031) (0.053) [50+] 0.248*** 0.203*** (0.037) (0.059) Constant 5.511*** 5.234*** 5.447*** 5.209*** 5.441*** 5.332*** 5.434*** 5.329*** 5.508*** 5.374*** (0.089) (0.103) (0.093) (0.110) (0.096) (0.112) (0.095) (0.115) (0.094) (0.108) Weighted observations 3357 1862 3357 1862 3357 1862 3091 1721 3357 1862 R sq 0.407 0.484 0.379 0.454 0.394 0.476 0.432 0.507 0.421 0.499 Notes OLS regres s i ons wi th robus t s tanda rd errors a nd i ncome wei ghts . Standa rd errors i n pa renthes es .* p<0.1 ** p<0.05 *** p<0.01 Wa ges ca l cul a ted a s a vera ge of monthl y ea rni ngs from ma i n occupa tion tra ns formed i n 2005 USD PPP us i ng SEDLAC CPI a nd PPP convers i on fa ctors . Incl ude monetary i ncome a nd a l l other i ncome rel a ted wi th the job, l i ke bonus es a nd i mpl i ci t rent/food/uni form va l ue, recei ved regul a rl y. Tra ns formed i n 2005 USD PPP us i ng SEDLAC CPI a nd PPP convers i on fa ctors . Experi ence = Age - Yea rs of educa tion + 6 Forma l i f (i ) wa ge empl oyees contri buting to Soci a l Securi ty, (i i ) empl oyers of a regi s tered fi rm (RUC), (i i i ) s el f-empl oyed workers wi th a regi s tered fi rm (RUC); Informa l i f (i ) fa rmers /herders /fi s herma n (s el f-empl oyed or empl oyer of fi rm wi th no RUC), (i i ) unpa i d fa mi l y worker, (i i i ) s el f- empl oyed, empl oyee or empl oyer of fi rm wi th no RUC, (i v) wa ge empl oyees not contri buting to Soci a l Securi ty. Tenure defi ned a s number of yea rs i n ma i n occupa tion. Incompl ete pri ma ry or l es s i f l es s tha n 6 yea rs of educa tion; Pri ma ry compl ete i f wi th 6 yea rs of educa tion a nd not enrol l ed; Seconda ry i ncompl ete i f wi th 6 yea rs of educa tion a nd enrol l ed or ha s (6-12) yea rs of educa tion; Seconda ry compl ete i f wi th 12 yea rs of educa tion a nd not enrol l ed; Tertia ry i ncompl ete i f wi th 12 yea rs of educa tion a nd enrol l ed or ha s (12-15) yea rs of educa tion; Tertia ry compl ete i f wi th more tha n 15 yea rs of educa To expl tion. ore for s el ection bi a s we a l s o es tima ted Heckma n model s . Res ul ts a re i dentica l to OLS a nd ca n be s uppl i ed by reques t. Source: Own ca l cul a tions ba s ed on EPH da ta. 86 Table B9 Mincer Regressions, OLS, Dependent variable log monthly real wage (2005 USD PPP), full-time public wage workers age 15+, 2015 Male Female Male Female Male Female Male Female Male Female (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Experience 0.029** 0.054*** 0.027** 0.067*** 0.027** 0.067*** 0.035*** 0.063*** 0.029** 0.054*** (0.011) (0.015) (0.013) (0.016) (0.013) (0.016) (0.011) (0.016) (0.011) (0.015) Experience^2 -0.000** -0.001*** -0.000* -0.001*** -0.000* -0.001*** -0.000*** -0.001*** -0.000** -0.001*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Formal 0.322*** 0.189* 0.296*** 0.213* 0.322*** 0.189* (0.070) (0.101) (0.071) (0.111) (0.070) (0.101) Job tenure 0.010** -0.004 0.015*** -0.003 0.015*** -0.003 0.008* -0.007 0.010** -0.004 (years) (0.004) (0.005) (0.004) (0.005) (0.004) (0.005) (0.004) (0.005) (0.004) (0.005) Language most spoken at home (relative to Spanish) Guaraní -0.152* -0.112 -0.180** -0.093 -0.180** -0.093 -0.122 -0.145* -0.152* -0.112 (0.081) (0.080) (0.090) (0.082) (0.090) (0.082) (0.083) (0.084) (0.081) (0.080) Guaraní and Spanish -0.151** -0.112* -0.157** -0.107 -0.157** -0.107 -0.100* -0.111 -0.151** -0.112* (0.060) (0.068) (0.061) (0.069) (0.061) (0.069) (0.060) (0.081) (0.060) (0.068) Other -0.131* -0.071 -0.071 -0.194** -0.131* (0.067) (0.072) (0.072) (0.096) (0.067) Highest education level (relative to incomplete primary or less) Primary complete -0.046 -0.289 0.045 -0.350 0.045 -0.350 0.006 -0.661 -0.046 -0.289 (0.144) (0.340) (0.162) (0.328) (0.162) (0.328) (0.156) (0.430) (0.144) (0.340) Secondary incomplete -0.065 -0.028 -0.005 -0.039 -0.005 -0.039 0.038 -0.457 -0.065 -0.028 (0.139) (0.337) (0.177) (0.326) (0.177) (0.326) (0.145) (0.431) (0.139) (0.337) Secondary complete 0.296** 0.001 0.401** 0.015 0.401** 0.015 0.350** -0.403 0.296** 0.001 (0.118) (0.332) (0.167) (0.325) (0.167) (0.325) (0.137) (0.441) (0.118) (0.332) Tertiary incomplete 0.428*** 0.428 0.639*** 0.459 0.639*** 0.459 0.517*** 0.015 0.428*** 0.428 (0.115) (0.332) (0.158) (0.323) (0.158) (0.323) (0.141) (0.438) (0.115) (0.332) Tertiary complete 0.705*** 0.535 0.877*** 0.564* 0.877*** 0.564* 0.805*** 0.134 0.705*** 0.535 (0.118) (0.332) (0.162) (0.324) (0.162) (0.324) (0.147) (0.437) (0.118) (0.332) 87 Mincer Regressions, OLS, Dependent variable log monthly real wage (2005 USD PPP), full-time public wage workers age 15+, 2015 Male Female Male Female Male Female Male Female Male Female (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Region (relative to Asunción) San Pedro -0.216** -0.317** -0.259** -0.359*** -0.259** -0.359*** -0.261** -0.309** -0.216** -0.317** (0.105) (0.126) (0.111) (0.127) (0.111) (0.127) (0.117) (0.139) (0.105) (0.126) Caaguazú -0.189 -0.356** -0.237* -0.399** -0.237* -0.399** -0.251* -0.368** -0.189 -0.356** (0.128) (0.153) (0.124) (0.165) (0.124) (0.165) (0.148) (0.163) (0.128) (0.153) Itapúa -0.216** -0.410*** -0.307*** -0.441*** -0.307*** -0.441*** -0.305*** -0.421*** -0.216** -0.410*** (0.098) (0.142) (0.088) (0.147) (0.088) (0.147) (0.098) (0.153) (0.098) (0.142) Alto Paraná -0.128 -0.197 -0.203 -0.222* -0.203 -0.222* -0.153 -0.198 -0.128 -0.197 (0.174) (0.128) (0.172) (0.128) (0.172) (0.128) (0.182) (0.133) (0.174) (0.128) Central -0.000 -0.213* -0.037 -0.238** -0.037 -0.238** -0.080 -0.243** -0.000 -0.213* (0.086) (0.118) (0.090) (0.118) (0.090) (0.118) (0.092) (0.121) (0.086) (0.118) Rest -0.102 -0.267** -0.124 -0.291** -0.124 -0.291** -0.178* -0.285** -0.102 -0.267** (0.084) (0.116) (0.087) (0.117) (0.087) (0.117) (0.098) (0.125) (0.084) (0.116) Firm size (relative to [2-5]) Alone -0.135 -0.177* (0.136) (0.104) [6-10] -0.085 -0.064 (0.104) (0.080) [11-20] -0.024 0.039 (0.085) (0.097) [21-50] -0.002 -0.030 (0.089) (0.076) [50+] 0.109 0.109 (0.088) (0.108) Constant 5.681*** 5.531*** 5.758*** 5.460*** 5.758*** 5.460*** 5.555*** 5.832*** 5.681*** 5.531*** (0.234) (0.394) (0.262) (0.397) (0.262) (0.397) (0.259) (0.485) (0.234) (0.394) Weighted observations 363 417 363 417 363 417 327 377 363 417 R sq 0.516 0.308 0.473 0.292 0.473 0.292 0.548 0.353 0.516 0.308 Note s OLS re gre s s i ons wi th robus t s ta nda rd e rrors a nd i ncome we i ghts . Sta nda rd e rrors i n pa re nthe s e s .* p<0.1 ** p<0.05 *** p<0.01 Wa ge s ca l cul a te d a s a ve ra ge of monthl y e a rni ngs from ma i n occupa ti on tra ns forme d i n 2005 USD PPP us i ng SEDLAC CPI a nd PPP conve rs i on fa ctors . Incl ude mone ta ry i ncome a nd a l l othe r i ncome re l a te d wi th the job, l i ke bonus e s a nd i mpl i ci t re nt/food/uni form va l ue , re ce i ve d re gul a rl y. Tra ns forme d i n 2005 USD PPP us i ng SEDLAC CPI a nd PPP conve rs i on fa ctors . Expe ri e nce = Age - Ye a rs of e duca ti on + 6 Forma l i f (i ) wa ge e mpl oye e s contri buti ng to Soci a l Se curi ty, (i i ) e mpl oye rs of a re gi s te re d fi rm (RUC), (i i i ) s e l f-e mpl oye d worke rs wi th a re gi s te re d fi rm (RUC); Informa l i f (i ) fa rme rs /he rde rs /fi s he rma n (s e l f-e mpl oye d or e mpl oye r of fi rm wi th no RUC), (i i ) unpa i d fa mi l y worke r, (i i i ) s e l f- e mpl oye d, e mpl oye e or e mpl oye r of fi rm wi th no RUC, (i v) wa ge e mpl oye e s not contri buti ng to Soci a l Se curi ty. Te nure de fi ne d a s numbe r of ye a rs i n ma i n occupa ti on. Incompl e te pri ma ry or l e s s i f l e s s tha n 6 ye a rs of e duca ti on; Pri ma ry compl e te i f wi th 6 ye a rs of e duca ti on a nd not e nrol l e d; Se conda ry i ncompl e te i f wi th 6 ye a rs of e duca ti on a nd e nrol l e d or ha s (6-12) ye a rs of e duca ti on; Se conda ry compl e te i f wi th 12 ye a rs of e duca ti on a nd not e nrol l e d; Te rti a ry i ncompl e te i f wi th 12 ye a rs of e duca ti on a nd e nrol l e d or ha s (12-15) ye a rs of e duca ti on; Te rti a ry compl e te i f wi th more tha n 15 Toae ye rs xplof e duca ore for s tieon. l e cti on bi a s we a l s o e s ti ma te d He ckma n mode l s . Re s ul ts a re i de nti ca l to OLS a nd ca n be s uppl i e d by re que s t. Source : Own ca l cul a ti ons ba s e d on EPH da ta . 88 Table B10 Mincer Regressions, OLS, Dependent variable log hourly real wage (2005 USD PPP), full-time public wage workers age 15+, 2015 Male Female Male Female Male Female Male Female Male Female (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Experience 0.036*** 0.054*** 0.035** 0.069*** 0.035** 0.069*** 0.039*** 0.063*** 0.036*** 0.054*** (0.013) (0.016) (0.014) (0.017) (0.014) (0.017) (0.013) (0.018) (0.013) (0.016) Experience^2 -0.000** -0.001*** -0.000** -0.001*** -0.000** -0.001*** -0.000** -0.001*** -0.000** -0.001*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Formal 0.224** 0.209** 0.211** 0.238** 0.224** 0.209** (0.090) (0.102) (0.098) (0.111) (0.090) (0.102) Job tenure 0.005 -0.004 0.009* -0.003 0.009* -0.003 0.002 -0.006 0.005 -0.004 (years) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005) Language most spoken at home (relative to Spanish) Guaraní -0.045 -0.091 -0.065 -0.071 -0.065 -0.071 -0.003 -0.130 -0.045 -0.091 (0.102) (0.082) (0.107) (0.083) (0.107) (0.083) (0.104) (0.086) (0.102) (0.082) Guaraní and Spanish -0.132* -0.124* -0.136* -0.118* -0.136* -0.118* -0.070 -0.110 -0.132* -0.124* (0.079) (0.070) (0.079) (0.070) (0.079) (0.070) (0.078) (0.083) (0.079) (0.070) Other -0.021 0.046 0.046 -0.077 -0.021 (0.103) (0.125) (0.125) (0.086) (0.103) Does not speak Highest education level (relative to incomplete primary or less) Primary complete -0.002 -0.231 0.061 -0.299 0.061 -0.299 0.085 -0.729* -0.002 -0.231 (0.136) (0.344) (0.155) (0.324) (0.155) (0.324) (0.156) (0.422) (0.136) (0.344) Secondary incomplete -0.064 -0.027 -0.022 -0.039 -0.022 -0.039 0.052 -0.498 -0.064 -0.027 (0.128) (0.333) (0.159) (0.321) (0.159) (0.321) (0.139) (0.400) (0.128) (0.333) Secondary complete 0.346** 0.067 0.419** 0.082 0.419** 0.082 0.404*** -0.392 0.346** 0.067 (0.137) (0.327) (0.172) (0.319) (0.172) (0.319) (0.148) (0.411) (0.137) (0.327) Tertiary incomplete 0.462*** 0.466 0.609*** 0.500 0.609*** 0.500 0.593*** 0.018 0.462*** 0.466 (0.121) (0.330) (0.151) (0.319) (0.151) (0.319) (0.152) (0.408) (0.121) (0.330) Tertiary complete 0.870*** 0.547* 0.990*** 0.578* 0.990*** 0.578* 0.990*** 0.102 0.870*** 0.547* (0.134) (0.329) (0.162) (0.319) (0.162) (0.319) (0.171) (0.408) (0.134) (0.329) 89 Mincer Regressions, OLS, Dependent variable log hourly real wage (2005 USD PPP), full-time public wage workers age 15+, 2015 Male Female Male Female Male Female Male Female Male Female (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Region (relative to Asunción) San Pedro -0.204* -0.325** -0.234** -0.371*** -0.234** -0.371*** -0.225* -0.316** -0.204* -0.325** (0.111) (0.128) (0.117) (0.130) (0.117) (0.130) (0.130) (0.143) (0.111) (0.128) Caaguazú -0.220* -0.334** -0.254* -0.381** -0.254* -0.381** -0.263* -0.349** -0.220* -0.334** (0.133) (0.150) (0.132) (0.162) (0.132) (0.162) (0.156) (0.162) (0.133) (0.150) Itapúa -0.182 -0.471*** -0.245** -0.505*** -0.245** -0.505*** -0.243* -0.485*** -0.182 -0.471*** (0.131) (0.173) (0.115) (0.183) (0.115) (0.183) (0.143) (0.178) (0.131) (0.173) Alto Paraná -0.552*** -0.252* -0.605*** -0.280** -0.605*** -0.280** -0.551*** -0.245* -0.552*** -0.252* (0.097) (0.130) (0.103) (0.131) (0.103) (0.131) (0.119) (0.134) (0.097) (0.130) Central -0.119 -0.208* -0.144 -0.235** -0.144 -0.235** -0.144 -0.221* -0.119 -0.208* (0.101) (0.120) (0.105) (0.119) (0.105) (0.119) (0.114) (0.123) (0.101) (0.120) Rest -0.195** -0.268** -0.210** -0.294** -0.210** -0.294** -0.256** -0.285** -0.195** -0.268** (0.086) (0.117) (0.090) (0.118) (0.090) (0.118) (0.106) (0.127) (0.086) (0.117) Firm size (relative to [2-5]) Alone -0.111 -0.333*** (0.143) (0.117) [6-10] -0.057 -0.092 (0.109) (0.086) [11-20] -0.022 0.018 (0.104) (0.102) [21-50] 0.101 -0.060 (0.107) (0.080) [50+] 0.153 0.085 (0.110) (0.112) Constant 0.268 0.308 0.322 0.229 0.322 0.229 0.093 0.657 0.268 0.308 (0.262) (0.393) (0.281) (0.393) (0.281) (0.393) (0.327) (0.461) (0.262) (0.393) Weighted observations 363 417 363 417 363 417 327 377 363 417 R sq 0.471 0.287 0.453 0.269 0.453 0.269 0.517 0.338 0.471 0.287 Notes OLS regres s i ons wi th robus t s ta nda rd errors a nd i ncome wei ghts . Sta nda rd errors i n pa renthes es .* p<0.1 ** p<0.05 *** p<0.01 Wa ges ca l cul a ted a s a vera ge of monthl y ea rni ngs from ma i n occupa ti on tra ns formed i n 2005 USD PPP us i ng SEDLAC CPI a nd PPP convers i on fa ctors . Incl ude moneta ry i ncome a nd a l l other i ncome rel a ted wi th the job, l i ke bonus es a nd i mpl i ci t rent/food/uni form va l ue, recei ved regul a rl y. Experi ence = Age - Yea rs of educa ti on + 6 Forma l i f (i ) wa ge empl oyees contri buti ng to Soci a l Securi ty, (i i ) empl oyers of a regi s tered fi rm (RUC), (i i i ) s el f-empl oyed workers wi th a regi s tered fi rm (RUC); Informa l i f (i ) fa rmers /herders /fi s herma n (s el f-empl oyed or empl oyer of fi rm wi th no RUC), (i i ) unpa i d fa mi l y worker, (i i i ) s el f- Tenure defi ned a s number of yea rs i n ma i n occupa ti on. Incompl ete pri ma ry or l es s i f l es s tha n 6 yea rs of educa ti on; Pri ma ry compl ete i f wi th 6 yea rs of educa ti on a nd not enrol l ed; Seconda ry i ncompl ete i f wi th 6 yea rs of educa ti on a nd enrol l ed or ha s (6-12) yea rs of educa ti on; Seconda ry compl ete i f wi th 12 yea rs of educa ti on a nd not enrol l ed; To expl ore for s el ecti on bi a s we a l s o es ti ma ted Heckma n model s . Res ul ts a re i denti ca l to OLS a nd ca n be s uppl i ed by reques t. Source: Own ca l cul a ti ons ba s ed on EPH da ta . 90 Table B11 Blinder-Oaxaca decomposition, Dependent variable Log Monthly real wage (2005 USD PPP), Employed age 15+, 2016 Specification (1) (2) (3) (4) (5) (6) Male 6.053*** 6.055*** 6.055*** 6.041*** 6.053*** 6.051*** (0.012) (0.012) (0.012) (0.012) (0.012) (0.011) Female 5.746*** 5.746*** 5.746*** 5.731*** 5.746*** 5.737*** (0.017) (0.017) (0.017) (0.017) (0.017) (0.016) Difference 0.307*** 0.308*** 0.308*** 0.310*** 0.307*** 0.314*** (0.021) (0.021) (0.020) (0.021) (0.020) (0.019) Endowments -0.120*** -0.159*** -0.167*** -0.020 -0.179*** -0.078*** (0.015) (0.014) (0.035) (0.018) (0.037) (0.027) Coefficients 0.413*** 0.446*** 0.409*** 0.291*** 0.386*** 0.396*** (0.018) (0.018) (0.022) (0.020) (0.022) (0.023) Interaction 0.013 0.021** 0.067* 0.038** 0.100*** -0.004 (0.009) (0.009) (0.035) (0.015) (0.037) (0.028) Explanatory variables Experience x x x x x Formal x x Public sector x Job tenure (years) x x x x x Education x x x x x Region x x x x x Sector x x x Firm size x Notes: Endowments represents the mean change in women’s wages if they had the same characteristics as men. Coefficients represents the change in women’s wages when applying the men’s coefficients to the women’s characteristics. Interaction measures the simultaneous effect of differences in endowments and coefficients. 91 ANNEX C – LABOR DEMAND REGRESSIONS Table C1. OLS Regression: Log Employment (paid employees per firm), Firm Census 2011 data on formal firms with at least one paid employee (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Age 0.0132*** 0.00378*** 0.00346***0.00368***0.00370***0.00834***0.0118*** 0.0112*** 0.0122*** 0.0116*** (0.00285) (0.000805) (0.000781)(0.000789)(0.000797)(0.00246) (0.00242) (0.00236) (0.00254) (0.00248) NorteOccidental -0.157 -0.0785 -0.0820 -0.0729 -0.0768 -0.0767 -0.102 -0.152 -0.142 -0.158 -0.148 (0.134) (0.0532) (0.0541) (0.0532) (0.0521) (0.0526) (0.0754) (0.124) (0.124) (0.128) (0.128) CentroSur -0.211 -0.0268 -0.0297 -0.0221 -0.0262 -0.0265 -0.00555 -0.182 -0.170 -0.198 -0.180 (0.132) (0.0544) (0.0552) (0.0534) (0.0538) (0.0540) (0.0892) (0.120) (0.121) (0.125) (0.125) Este -0.133 -0.112** -0.118** -0.111** -0.109** -0.110** -0.231*** -0.151 -0.136 -0.148 -0.136 (0.137) (0.0534) (0.0549) (0.0528) (0.0528) (0.0532) (0.0630) (0.126) (0.128) (0.129) (0.130) Text&Garm&Leath -0.337** (0.131) Chemicals&Rubber&Plastic 0.775*** (0.236) Metals&Machinery -0.406*** (0.0957) Other manuf. -0.307*** (0.0764) Utilities 0.122 (0.322) Construction 0.0508 (0.283) Trade -0.519*** (0.167) Hotels&Restaur. -0.552*** (0.160) Transport&Commun. -0.208 (0.277) Finance&Bus. -0.333 (0.234) Other services -0.249 (0.236) 92 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) MinUtilConstr 0.0217 0.0258 -0.184 0.0223 0.0229 -0.138 0.241 0.230 0.259 0.243 (0.131) (0.133) (0.145) (0.131) (0.131) (0.193) (0.244) (0.224) (0.243) (0.224) Commerce -0.450*** -0.456*** -0.450*** -0.429*** -0.429*** -0.751*** -0.383*** -0.323*** -0.357*** -0.310*** (0.0452) (0.0465) (0.0450) (0.0439) (0.0436) (0.0785) (0.109) (0.111) (0.113) (0.114) Services -0.135* -0.139* -0.252*** -0.0863 -0.0870 -0.0387 -0.165 -0.161 -0.158 -0.161 (0.0786) (0.0791) (0.0569) (0.0810) (0.0788) (0.177) (0.155) (0.157) (0.156) (0.158) Real Sales (log) 0.440*** 0.441*** 0.439*** 0.437*** 0.439*** 0.527*** (0.0455) (0.0459) (0.0453) (0.0451) (0.0453) (0.0379) Young -0.0440*** (0.0101) Tradable sectors -0.208*** (0.0609) Firm 50+% female employees -0.111*** (0.0184) Female Employment Share -0.122*** (0.0233) Capital -real (log) 0.0664*** (0.0160) Sales per worker -log 0.169*** (0.0446) Value Added per worker -real LCU (log) 0.177*** (0.0450) Output Worker Q2 -0.00376 (0.0302) Output Worker Q3 0.0379 (0.0412) Output Worker Q4 0.309*** (0.0912) Value Added Worker Q2 0.107*** (0.0197) Value Added Worker Q3 0.155*** (0.0261) Value Added Worker Q4 0.343*** (0.0827) Constant 1.227*** -4.311*** -4.264*** -4.097*** -4.265*** -4.280*** -6.576*** -0.899* -0.831* 0.933*** 0.860*** (0.165) (0.563) (0.565) (0.564) (0.558) (0.560) (0.650) (0.487) (0.456) (0.115) (0.117) 93 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Observations 25,111 25,617 25,617 25,617 25,617 25,617 455 24,889 24,823 24,889 24,823 R-squared 0.061 0.554 0.553 0.556 0.556 0.556 0.617 0.077 0.072 0.066 0.064 Food&Bev - omitted YES Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 94 Table C2 OLS Regression: Log Value Added per Worker (firm value added per paid employee), Firm Census 2011 data on formal firms with at least one paid employee (1) (2) (3) (4) (5) (6) Employment (log) 0.150*** (0.0190) age 0.00569*** 0.00562***0.00592***0.00594***0.00327 (0.000831) (0.000992)(0.00113) (0.00112) (0.00331) NorteOccidental -0.0609 -0.0686 -0.0618 -0.0665 -0.0664 -0.325*** (0.0406) (0.0414) (0.0373) (0.0416) (0.0413) (0.0944) CentroSur -0.174*** -0.199*** -0.193*** -0.198*** -0.198*** -0.211*** (0.0353) (0.0357) (0.0333) (0.0360) (0.0361) (0.0747) Este 0.0106 0.00392 0.00711 0.00825 0.00757 -0.00839 (0.0392) (0.0373) (0.0360) (0.0375) (0.0375) (0.0723) Text&Garm&Leath -0.260*** (0.0711) Chemicals&Rubber&Plastic 0.123 (0.117) Metals&Machinery 0.00875 (0.0260) Other manuf. -0.0216 (0.0426) Utilities -0.251*** (0.0907) Construction 0.275*** (0.0328) Trade 0.262*** (0.0263) Hotels&Restaur. -0.142*** (0.0205) Transport&Commun. 0.197*** (0.0579) Finance&Bus. 0.233*** (0.0307) Other services -0.0139 (0.0731) 95 (1) (2) (3) (4) (5) (6) Size 10-19 0.387*** 0.381*** 0.377*** 0.384*** -0.137 (0.0422) (0.0453) (0.0442) (0.0439) (0.106) Size 20-99 0.751*** 0.728*** 0.732*** 0.741*** -0.137 (0.0786) (0.0882) (0.0833) (0.0833) (0.229) Size 100+ 1.092*** 1.035*** 1.044*** 1.060*** 0.467 (0.0897) (0.110) (0.101) (0.1000) (0.606) Young -0.146*** (0.0126) MinUtilConstr 0.187*** -0.0756 0.185*** 0.186*** 0.206 (0.0616) (0.0843) (0.0636) (0.0638) (0.220) Commerce 0.317*** 0.317*** 0.335*** 0.334*** 0.410*** (0.0266) (0.0264) (0.0300) (0.0303) (0.142) Services 0.0755** -0.0725*** 0.116*** 0.111*** 0.145 (0.0340) (0.0234) (0.0399) (0.0411) (0.210) Tradable sectors -0.264*** (0.0551) Firm 50+% female employees -0.0924*** (0.0234) Female Employment Share -0.0911*** (0.0317) Capital -real (log) 0.0475*** (0.0156) Constant 10.16*** 10.31*** 10.46*** 10.20*** 10.20*** 9.934*** (0.0396) (0.0335) (0.0691) (0.0400) (0.0408) (0.371) Observations 24,336 24,823 24,823 24,823 24,823 439 R-squared 0.072 0.074 0.077 0.074 0.074 0.063 Food&Bev - omitted YES Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 96 Table C3 OLS Regression: Log Wage (firm wage bill per paid employee), Firm Census 2011 data on formal firms with at least one paid employee (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Employment (log) 0.259*** (0.0156) Age 0.00652*** 0.00506*** 0.00437*** 0.00489*** 0.00489*** 7.65e-05 0.00597*** 0.00489*** 0.00627*** 0.00511*** (0.000674) (0.000756) (0.000432) (0.000690) (0.000670) (0.00189) (0.000873) (0.000601) (0.000890) (0.000669) NorteOccidental -0.220*** -0.235*** -0.236*** -0.223*** -0.233*** -0.231*** -0.260*** -0.247*** -0.221*** -0.238*** -0.207*** (0.0309) (0.0209) (0.0215) (0.0121) (0.0213) (0.0209) (0.0470) (0.0204) (0.0213) (0.0232) (0.0235) CentroSur -0.275*** -0.235*** -0.236*** -0.224*** -0.233*** -0.234*** -0.238*** -0.261*** -0.223*** -0.263*** -0.220*** (0.0221) (0.0186) (0.0192) (0.0121) (0.0183) (0.0182) (0.0497) (0.0192) (0.0164) (0.0210) (0.0199) Este -0.0783***-0.114*** -0.116*** -0.111*** -0.107*** -0.108*** -0.132*** -0.131*** -0.112*** -0.116*** -0.0980*** (0.0228) (0.0105) (0.0109) (0.00898) (0.0130) (0.0135) (0.0357) (0.0115) (0.0123) (0.0147) (0.0155) Text&Garm&Leath -0.0572 (0.0417) Chemicals&Rubber&Plastic 0.00736 (0.0774) Metals&Machinery 0.212*** (0.0341) Other manuf. 0.115*** (0.0411) Utilities -0.0624 (0.0803) Construction 0.174*** (0.0642) Trade 0.0996*** (0.0237) Hotels&Restaur. -0.286*** (0.0241) Transport&Commun. 0.170*** (0.0568) Finance&Bus. 0.176*** (0.0239) Other services 0.0130 (0.0637) 97 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Size 10-19 -0.0837* -0.0812* -0.0989* -0.100** -0.0883* -0.0444 0.380*** 0.377*** 0.426*** 0.403*** (0.0471) (0.0472) (0.0507) (0.0463) (0.0477) (0.0508) (0.0290) (0.0349) (0.0338) (0.0373) Size 20-99 -0.0406 -0.0320 -0.0773 -0.0571 -0.0456 -0.000537 0.494*** 0.471*** 0.583*** 0.571*** (0.0719) (0.0722) (0.0697) (0.0697) (0.0703) (0.109) (0.0457) (0.0546) (0.0414) (0.0508) Size 100+ 0.0987 0.120 0.0525 0.0914 0.103 -0.801** 0.549*** 0.528*** 0.705*** 0.686*** (0.155) (0.157) (0.152) (0.153) (0.155) (0.307) (0.0835) (0.0958) (0.0843) (0.0819) MinUtilConstr -0.0469 -0.0446 -0.525*** -0.0449 -0.0438 -0.197 -0.0150 -0.0157 0.00931 0.0111 (0.0634) (0.0637) (0.0633) (0.0628) (0.0625) (0.171) (0.0528) (0.0537) (0.0525) (0.0495) Commerce -0.145*** -0.147*** -0.149*** -0.104*** -0.0988***-0.125** -0.201*** -0.128*** -0.152*** -0.0829*** (0.0136) (0.0131) (0.0118) (0.0162) (0.0166) (0.0553) (0.0171) (0.0140) (0.0192) (0.0154) Services -0.151*** -0.151*** -0.422*** -0.0503* -0.0393 0.0656 -0.193*** -0.189*** -0.191*** -0.179*** (0.0217) (0.0219) (0.0144) (0.0261) (0.0268) (0.0646) (0.0229) (0.0189) (0.0248) (0.0220) Real Sales (log) 1.023*** 1.013*** 1.044*** 1.045*** 1.031*** 0.0684 (0.0913) (0.0936) (0.0912) (0.0897) (0.0922) (0.303) Real Sales (log) squared -0.0290***-0.0287***-0.0297***-0.0300***-0.0294***0.00376 (0.00371) (0.00380) (0.00370) (0.00365) (0.00375) (0.0103) Young -0.116*** (0.0174) Tradable sectors -0.482*** (0.0147) Firm has 50+% female employees -0.229*** (0.0243) Female Employment Share -0.283*** (0.0324) Capital -real (log) 0.00526 (0.0134) Sales per worker -log 0.261*** (0.00970) Value Added per worker -real LCU (log) 0.348*** (0.0185) 98 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Output Worker Q2 0.352*** (0.0261) Output Worker Q3 0.430*** (0.0249) Output Worker Q4 0.613*** (0.0287) Value Added Worker Q2 0.527*** (0.0193) Value Added Worker Q3 0.611*** (0.0195) Value Added Worker Q4 0.717*** (0.0239) Constant 9.194*** 1.468** 1.644*** 1.812*** 1.382** 1.477** 8.318*** 6.627*** 5.959*** 9.228*** 9.105*** (0.0409) (0.563) (0.582) (0.559) (0.553) (0.570) (2.235) (0.106) (0.187) (0.0278) (0.0239) Observations 25,111 25,617 25,617 25,617 25,617 25,617 455 24,889 24,823 24,889 24,823 R-squared 0.187 0.361 0.362 0.383 0.379 0.382 0.212 0.234 0.291 0.206 0.273 Food&Bev - omitted YES Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 99 Table C4 OLS Regression: Employment Growth (change in paid employment per firm, 2010-2014), panel dataset (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) sza_10to19 -0.00173 -0.0183 -0.0102 0.00771 0.00695 0.0117 -0.0183 -0.0344 -0.0241 -0.0187 -0.0363 -0.0295 -0.00652 -0.0342 (0.0767) (0.0874) (0.0918) (0.0912) (0.0907) (0.0932) (0.0874) (0.0882) (0.0911) (0.0816) (0.0907) (0.0821) (0.0791) (0.0845) sza_20to49 0.0977 0.00568 0.0353 0.0930 0.0852 0.0933 0.00568 -0.00921 0.00195 0.0384 0.0116 0.0358 0.0425 -0.0252 (0.136) (0.157) (0.153) (0.152) (0.147) (0.153) (0.157) (0.155) (0.157) (0.152) (0.156) (0.146) (0.160) (0.156) sza_50to99 0.286*** 0.309*** 0.306*** 0.364*** 0.364*** 0.348*** 0.309*** 0.292*** 0.303*** 0.305*** 0.302*** 0.325*** 0.265*** 0.277*** (0.0511) (0.0653) (0.0674) (0.0304) (0.0310) (0.0369) (0.0653) (0.0653) (0.0661) (0.0727) (0.0700) (0.0828) (0.0620) (0.0622) sza_100plus 0.254*** 0.186* 0.200** 0.301*** 0.301*** 0.299*** 0.186* 0.175* 0.180** 0.215** 0.197** 0.218*** 0.196** 0.155* (0.0776) (0.0887) (0.0860) (0.0718) (0.0710) (0.0725) (0.0887) (0.0838) (0.0789) (0.0762) (0.0815) (0.0750) (0.0789) (0.0759) age_6to9 -0.0689 -0.0635 -0.110*** -0.111** -0.0689 -0.0745 -0.0675 -0.0817 -0.0628 0.932* -0.0764 -0.0922 (0.0490) (0.0507) (0.0376) (0.0403) (0.0490) (0.0476) (0.0464) (0.0553) (0.0502) (0.511) (0.0461) (0.0720) age_10plus -0.0191 -0.0295 -0.0245 -0.0215 -0.0191 -0.00788 -0.0178 0.00310 0.00393 -0.365 0.00371 0.00929 (0.0307) (0.0325) (0.0238) (0.0226) (0.0307) (0.0335) (0.0316) (0.0242) (0.0277) (0.487) (0.0215) (0.0270) Norte 0.0536 (0.0900) Occidental -0.00417 (0.0875) CentroSur 0.118** (0.0491) young -0.172*** (0.0444) MinUtilConstr -0.0600 -0.0706 (0.0565) (0.0564) Commerce -0.0908 -0.0981 (0.0578) (0.0580) Services -0.0640 -0.0680 (0.0605) (0.0584) Utilities -0.0597 (0.0568) WholesaleRetail -0.0900 (0.0581) TransportStorageComm -0.00999 (0.0507) HotelsRestaurants -0.0644 (0.0536) OtherServices -0.0851 (0.0890) Tradable sector 0.569*** (0.133) 100 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) Share female workers -0.0507* (0.0261) Majority female workers -0.0182 (0.0509) Value Added per worker -log (previous period) -0.0754** (0.0355) Value Added per worker -log (previous period) quartile 2 -0.182* (0.103) Value Added per worker -log (previous period) quartile 3 -0.0896 (0.0897) Value Added per worker -log (previous period) quartile 4 -0.172* (0.0870) Constant 0.327*** 0.399*** 0.358*** 0.164*** 0.157** 0.161*** -0.171* 0.419*** 0.404*** 0.977*** 0.399** 1.133*** 1.160*** 0.506*** (0.102) (0.130) (0.114) (0.0542) (0.0550) (0.0544) (0.0920) (0.116) (0.122) (0.222) (0.142) (0.327) (0.374) (0.145) Observations 502 343 343 343 343 343 343 343 343 317 317 317 326 326 R-squared 0.285 0.269 0.251 0.108 0.108 0.110 0.269 0.278 0.270 0.331 0.314 0.336 0.333 0.344 Sector dummies YES YES YES NO NO NO YES YES YES YES YES YES YES YES Location dummies YES YES NO YES YES YES YES YES YES YES YES YES YES YES Year Dummies YES YES YES YES YES YES YES YES YES YES YES YES YES YES R2 0.285 0.269 0.251 0.108 0.108 0.110 0.269 0.278 0.270 0.331 0.314 0.336 0.333 0.344 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 101 Address: 1776 G St, NW, Washington, DC 20006 Website: http://www.worldbank.org/en/topic/jobsanddevelopment Twitter: @WBG_Jobs Blog: https://blogs.worldbank.org/jobs/