JOBS WORKING PAPER Issue No. 19 Firm Productivity and Employment in Paraguay 2010-2014 Elizabeth Ruppert Bulmer and Adrian Scutaru © 2018 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 and Adrian Scutaru. 2018. “Firm Productivity and Employment in Paraguay 2010-2014.” 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. ii Abstract Paraguay has achieved positive employment outcomes over the last decade, reflecting more jobs and improved average job quality, which together contributed to significant poverty reduction. Firms played a major role in creating these jobs, especially formal jobs. It is important to understand how firm performance has translated into employment, the nature of the resulting jobs, and the factors affecting these outcomes. Using firm-level datasets, this analysis explores the characteristics of firms that are growing or stagnating; the sectors into which new firms are entering or incumbent firms are expanding their operations; and the factors that may be helping or hindering firms to enter, expand, or increase their productivity. High rates of informality and evasion create a challenging private sector environment for firms, but even among formal firms, the prevailing structure is characterized by micro-firms that lack scale economies, are concentrated in non-tradables, have generally low productivity levels, and rely on unskilled employment. The analysis also finds a pervasive duality between micro-sized low-productivity firms on one end, and a small number of highly productive firms in concentrated markets on the other end. These results imply some important challenges for the continued development of a healthy, dynamic and inclusive private sector. iii Acknowledgments This report was written by Elizabeth Ruppert Bulmer (Lead Economist) and Adrian Scutaru (Consultant) of the Jobs Group, a unit of the World Bank’s Social Protection and Jobs Global Practice. The team is grateful for the comments provided by peer reviewers Mary Hallward- Driemeier (Senior Economic Advisor, Finance, Competitiveness and Innovation Global Practice) and Denis Medvedev (Lead Economist, Finance, Competitiveness and Innovation Global Practice). The report was written under the guidance of David Robalino (Manager, Jobs Group), Michal Rutkowski (Senior Director, Social Protection and Jobs Global Practice), Rafael Rofman (Program Leader for Argentina, Paraguay and Uruguay) and Jesko Hentschel (Country Director for Argentina, Paraguay and Uruguay). 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 analysis. The authors also owe thanks to the Secretaría Técnica de Planificación del Desarrollo Económico y Social and the Ministerio de Hacienda for the input and technical support offered. This assessment of firm performance and labor demand in Paraguay is part of the Let’s Work program in Paraguay, which 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). iv Contents Abstract ........................................................................................................................................................ iii Acknowledgments........................................................................................................................................ iv 1. Introduction ......................................................................................................................................... 1 2. Country Context ................................................................................................................................... 2 A. Macroeconomic Setting ................................................................................................................... 2 B. Recent Labor Market Performance.................................................................................................. 5 C. Regulatory and Institutional Context ............................................................................................... 6 3. Characteristics of Private Sector Firms ................................................................................................ 8 A. Data and Methodology .................................................................................................................... 8 B. Snapshot of Private Sector Firms in Paraguay ................................................................................. 9 4. Estimating Formal Firm Performance ................................................................................................ 17 A. Correlates of Productivity ........................................................................................................... 17 B. Correlates of Firm Size and Growth ............................................................................................ 23 C. Correlates of Wages .................................................................................................................... 26 5. Conclusions ........................................................................................................................................ 28 References .................................................................................................................................................. 31 Annex .......................................................................................................................................................... 32 v vi 1. Introduction Paraguay has achieved positive employment outcomes over the last decade, but these outcomes reflect heterogeneous effects among workers and firms. Job creation kept pace with rapid labor force growth, and average job quality improved, creating positive welfare effects. Firms played a major role in creating these jobs, especially formal jobs. It is important to understand how firm performance has translated into employment, and the degree to which this employment is good for development.1 Firms come in all shapes and sizes. They span different sectors of activity and range from very low to very high levels of productivity and competitiveness. Whereas a technology-intensive manufacturer of chemical products may employ highly skilled workers on formal contracts with generous salaries and benefits, at the other end of the spectrum we may find a non-salaried micro-entrepreneur providing local food services at very low profit margins and low productivity levels. Under the development lens, both types of employment are of value, but affect different segments of the labor market and create different dynamics for labor productivity growth and job growth. The factors behind firms’ labor demand and employment decisions are grounded in market conditions. In a conceptual framework of firm decision-making, firms can affect employment outcomes through three channels: starting new operations, expanding existing operations, or becoming more productive, which ultimately raises the skill level and earnings of workers. These spill over into more employment and/or better jobs in the economy. Firms may be motivated by a perceived business opportunity that has a positive rate of return. To seize the opportunity to start or expand operations requires adequate capital. Entrepreneurs also require know-how: not only technical knowledge, but perhaps more crucially the entrepreneurial capacity to implement their concept in the face of market realities. Multiple factors affect the role of firms in creating jobs, several of which are analyzed in this report. The analysis explores the characteristics of firms that are growing or stagnating; the sectors into which new firms are entering or incumbent firms are expanding their operations; and the factors that may be helping or hindering firms to enter, expand, or increase their productivity. By taking the firm’s perspective to understand the drivers of labor demand, the analysis will complement existing knowledge on the supply of labor, providing a fuller picture of jobs outcomes in Paraguay.2 The analysis in this report uses firm-level datasets3 to assess the distribution and main characteristics of formal and informal firms. It also looks at the correlates of formal firms’ 1 Jobs that are good for development are those that boost living standards, have higher levels of productivity, and enhance social cohesion through positive social externalities (World Bank 2012). 2 See “Paraguay Jobs Diagnostic: The Dynamic Transformation of Employment in Paraguay”, Ruppert Bulmer et al. (2017). 3 Datasets include a national census of firms conducted in 2011 (data in reference to 2010), a 2015/16 firm survey (data in reference to 2014), and firm registry data from 2013 to 2015. The national firm census counted all firms that could be identified in commercial spaces, and excludes household enterprises and farming in particular. Even though data coverage of firms is high, the firm census captures less than a third of total employment estimated by the household survey Encuesta Permanente de Hogares. See Annex for details. 1 labor demand and labor productivity across different sectors and regions of the economy to identify the challenges and potential obstacles to accelerating formal job creation and increasing labor productivity. Firms in Paraguay face serious challenges. The largest challenges emerging from the analysis presented below stem from the difficult private sector context of high informality, lack of scale economies due to small firm size, concentration of employment in services, and generally low productivity levels. Today’s private sector is not well-positioned to meet the expected rise in consumer demand for goods and services of increasing quality, if positive trends in household incomes and the emerging middle class continue. For Paraguayan firms and workers to benefit from a virtuous cycle of increased demand for more sophisticated inputs requiring more skilled labor in more productive and better-paying jobs, Paraguay will need a more dynamic private sector landscape where firms can enter, create employment, and upgrade productivity sufficiently to compete on the local and international stages. The remainder of this paper is structured as follows. Section 2 provides a snapshot of the macroeconomic, labor market and regulatory contexts within which firms are entering, operating and creating jobs. Section 3 presents descriptive statistics of Paraguay’s private sector with particular attention to the distribution of firms by size, age, formality status and productivity. Against this backdrop, Section 4 analyzes the correlates of firm performance among formal firms, testing a set of explanatory variables that include size, age, sector, gender composition of employment, and region. In addition to contemporaneous correlations tests, for a sub-sample panel of firms observed at two points in time, the analysis looks at the drivers of firms’ employment growth, productivity growth, and wage growth between 2010 and 2014. Section 5 concludes with the main findings. 2. Country Context A. Macroeconomic Setting Paraguay has experienced robust economic growth in the last decades, leading to important 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 2 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. Figure 1 Positive trends in GDP per capita and poverty reduction The structure of economic production is undergoing a profound transformation away from agriculture toward services and, to a lesser degree, industrial activities. Nevertheless, agriculture continues to play a central role. It accounts for 20 percent of both employment and GDP, and is a significant source of export earnings. More than three quarters of all exports are food and agricultural products, and another 22 percent is energy from hydroelectric generation (Figure 2). Commercial, capital-intensive agricultural production has boomed over the past 15 years. Although Paraguayan products have expanded into new markets, trade remains highly concentrated in relatively few products and markets (dominated by MERCOSUR, and especially Brazil). Paraguay’s highly concentrated export basket and fairly narrow non-agriculture activities are consistent with a high level of imports, even of foodstuffs. Paraguay’s agriculture sector has a dual nature: capital-intensive commodity exporters, and small-scale, labor-intensive family farming of traditional products. The modern agriculture sector employs a relatively small number of highly skilled workers, whereas traditional agriculture accounts for nearly a fifth of total employment—mostly self-employed farmers or unpaid family workers using low-productivity methods and producing for auto-consumption or local markets. The links between the modern and traditional agriculture sectors are minimal, as their relative contributions to growth and employment diverge. But neither engages in significant local transformation or value addition. 3 Figure 2 Export basket (2014) Source: Harvard University, Atlas of Economic Complexity (http://atlas.cid.harvard.edu). Paraguay’s last decade of solid economic performance was underpinned by sound macroeconomic policies. Annual GDP growth averaged nearly 5 percent, albeit with sizeable year-to-year fluctuations linked to agriculture-related volatility, while fiscal balance was maintained (averaging a fiscal surplus of 0.4 percent of GDP during 2004-2016) and the public debt level remained low (Table 1). Paraguay was little affected by the global financial crisis. The Central Bank’s monetary policy stance helped keep inflation under control; in fact, inflation has been below that of regional comparators since 2012, and is close to the OECD average.4 Exchange rate policies were sufficiently flexible to absorb external shocks, and foreign reserves have been maintained at prudent levels. Domestic credit as a share of GDP more than tripled since 2004, but the financial sector appears relatively sound.5 4 World Bank (2018) 5 See World Bank (2018) for a detailed description of recent macroeconomic developments. 4 Table 1 Key Macroeconomic and Fiscal Indicators 2010 2011 2012 2013 2014 2015 2016 Real GDP, percent change 13.1 4.3 -1.2 14.0 4.7 3.0 4.0 Consumer Inflation, percent change 7.2 4.9 4.0 3.7 4.2 3.1 3.9 Current Account, percent GDP 0.5 -1.0 -2.0 1.7 -0.4 -1.1 1.7 International Reserves, months of 3.9 5.0 4.6 5.3 7.2 6.8 7.4 future imports Consolidated Public Sector, percent GDP Revenue 20.9 22.6 23.6 22.1 22.8 24.0 23.7 Expenditure 20.4 21.1 25.3 23.5 23.5 25.4 24.9 Overall Balance 0.5 1.4 -1.6 -1.4 -0.7 -1.3 -1.1 Primary Balance 1.3 2.1 -1.0 -0.7 0.1 -0.3 0.0 Debt 15.3 12.4 16.2 17.0 19.7 24.0 24.6 Source: Paraguay authorities, IMF, World Bank B. Recent Labor Market Performance6 Rapid demographic growth has put considerable pressure on the labor market, but job creation has been sufficient to absorb the youth bulge to date. The labor force expanded by 2.6 percent annually over the past decade, while total employment grew 2.8 percent per year, equivalent to 63,000 net new jobs annually. Most added employment was in urban settings – concentrated in Greater Asuncion, consistent with the observed rapid urbanization. Looking to the future, Paraguay’s labor force is projected to add 970,000 workers between 2015 and 2030, requiring nearly 65,000 new jobs every year. Paraguay’s ongoing structural transformation is reflected in shifting employment patterns, where services are becoming increasingly dominant.7 Agricultural employment has contracted since 2001, especially among informal farmers, while manufacturing grew slightly, and service sector employment expanded rapidly. The majority of job growth since 2008 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). Whereas all sectors except agriculture experienced solid job growth, the sectoral distribution of labor productivity growth was mixed, and is reflected in the types of jobs created. Service-sector jobs tend to be less productive and pay lower wages, although public administration jobs are an exception. The increasingly urbanizing labor force is reflected by the migration of rural workers to better paying jobs in cities, although many of these 6 The discussion of the labor market relies on household survey data from the Encuesta Permenente de Hogares. 7 See Ruppert Bulmer et al. (2017) 5 jobs are in retail, construction, or other services, suggesting only modest gains in labor productivity. Formal (net) job creation outstripped informal job creation by four to one between 2010 and 2014, translating into improved average job quality. The informality rate declined from 77 to 71 percent of total employment during this four-year period, raising average wages and expanding access to social insurance. In fact, real wages increased across the board—for both formal workers (4 percent annual growth during 2010-2014) and informal workers (2.2 percent). Informal low-productivity employment still dominates, however. Self-employed farmers account for 13 percent of total employment, another 19 percent are self-employed in non-farm activities, and over a third are in informal wage work (Figure 3). Moreover, despite the diverse sectoral distribution of added jobs, most jobs in Paraguay are still found in agriculture, retail, and other services—three sectors with the lowest productivity, the lowest wages, and a high degree of informality. Figure 3 Employment by work status (2016) C. Regulatory and Institutional Context The business climate in Paraguay is burdensome, and certain aspects are 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. 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, among other things, with little change since 2010 6 (Figure 4). 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, e.g., by simplifying firm registration, reforming the insolvency regime, and creating a moveable assets collateral registry to ease credit access. Figure 4 Paraguay’s Doing Business rankings 2018 (out of 190 countries) Source: www.doingbusiness.org Paraguay’s labor regulations are characterized by certain rigidities and vulnerabilities that constrain firms’ competitiveness and growth.8 The tax on labor requires a 14 percent employer contribution and 9 percent employee contribution. For firms with low levels of productivity, these tax rates become a significant obstacle to creating formal jobs or declaring employees and paying social insurance contributions, a factor that drives the prevalence of informal wage and non-wage workers within formal firms. According to the World Bank Group’s 2017 Enterprise Survey of 364 registered firms, 10 percent report that labor regulations represent a major constraint to their firms’ operations.9 These regulations include, among others, minimum wages and dismissal rules. Paraguay’s minimum wage is not particularly binding due to the high rate of informality and evasion; on the other hand, it may be a contributing factor to high informality and low productivity. Severance pay is relatively generous in Paraguay, particularly for workers with more than 10 years of job tenure. Because dismissal costs fall directly on firms, hiring decisions may be distorted, or contract turnover may be artificially high. 8 Kuddo and Ruppert Bulmer (2017) 9 www.enterprisesurveys.org 7 There are broader business climate challenges that are structural in nature and thus harder to address. These relate to Paraguay’s small market size, poor connectivity to external markets, and weak institutional environment for protecting property rights, enforcing contracts, levying fair and transparent taxes, and guaranteeing enforcement of regulations. Paraguay’s small domestic market and land-locked setting in particular highlight the need for better trade links between local producers, foreign investors in export manufactures, and external consumers. Transport costs are high, and the time and cost of exporting exceed the LAC averages, rendering Paraguayan products less competitive in the international market.10 Data from the 2017 Enterprise Survey indicate that nearly three-fourths of firms compete against informal firms, and 30 percent report this as a major constraint to their operations and profitability. 3. Characteristics of Private Sector Firms A. Data and Methodology This report exploits several firm-level datasets to gain a detailed understanding of the private sector landscape in Paraguay (see Annex for a detailed description of the data). These firm-level data were collected by Paraguay’s national statistical office, the Dirección General de Estadística, Encuestas y Censos (DGEEC), between 2011 and 2016. The datasets include: a national census of all informal and formal firms with reference to 2010; a nationally representative survey of formal firms with reference to 2014; and a partial registry of formal firms, collected with reference to 2013, 2014 and 2015. We construct a small panel of formal firms that appear in both the 2010 census and 2014 survey, enabling direct comparisons over time. Given the high degree of informality in Paraguay’s economy, including the large swathes of the labor force working outside the structure of a firm, such as farmers, unpaid workers in family enterprises, and those self-employed in household enterprises, the firm census for 2010 does not capture 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. The analysis that follows provides a picture of private sector firm performance. With these data caveats in mind, the datasets include information on key aspects of firms’ hiring behavior and productivity, such as employment level, annual sales, wage bill, and value added. Firm-level employment is differentiated by gender and by remuneration method, i.e., paid, unpaid, and commission-based pay, whereas wages are reported at the firm level only. Additional information is collected on firm characteristics such as formality status, sector of production, firm age, and firm location. This information is used to understand the correlates of firms’ productivity and growth. The analysis defines unregistered firms as informal, and registered firms as formal, even when those registered firms are self-employed entrepreneurs and/or do not comply with regulations and tax rules. In this firm-level analysis, available data restrict us to the registration- 10 www.doingbusiness.org 8 based definition, although we later distinguish firm behavior on the basis of other relevant characteristics. Because the datasets lack variables relating to production inputs other than labor, we restrict our productivity measure to value added per worker. Data on the level of capital assets are available for only a small subset of firms, and there is no information collected on firms’ human capital, whether based on education or skill level or occupation. Restricting the productivity analysis to value added per worker rather than total factor productivity requires a careful interpretation of the productivity results that avoids broad conclusions relating to technical efficiency, for example, or that considers the role of price effects in driving value-added. The analysis that follows nevertheless provides important insights into recent firm performance and growth. B. Snapshot of Private Sector Firms in Paraguay Informality is widespread among firms captured in the firm census, but formal firms generate significant employment. Thirty-eight percent of firms identified in the 2011 Firm Census are informal, but 85 percent of employees work in formal registered firms (see Table 2). Even among registered firms, however, 60 percent are self-employed entrepreneurs without any paid employees. By some definitions, they too could be considered informal.11 Moreover, not all jobs within large registered firms are formal, and employers may underreport them: the household survey data shows that most wage employees lack permanent contracts with social insurance coverage. Despite these data caveats, the 51,000 formal firms that reported at least one paid employee account for two-thirds of all jobs captured in the firm census. Table 2 Firm Census 2011 Number of Employment Employment firms level share Informal unregistered (no RUC) 79,373 123,397 15% Formal registered (RUC) 131,669 675,756 85% o/w self-employed 80,212 138,731 17% o/w at least 1 paid employee 51,457 537,025 67% Total Census of Firms 211,042 799,153 100% Note: RUC is the registration number Registro Único de Contribuyente. Source: DGEEC Informal firms tend to be smaller and younger than formal firms. Whereas informal firms are typically smaller than formal firms, a non-trivial share of informal firms is large. The highest concentration of informal firms is 1-5 years old, but many have been in operation for 10, 20 or even 30 years (Figure 5). However, these older informal firms tend to remain small. 11 The recent Jobs Diagnostic report defines informal employment to include self-employment, farmers, unpaid family workers, and wage employees not covered by social security, inter alia (see Ruppert Bulmer et al. 2018). 9 Figure 5 Unregistered firms are mostly young Firm age Note: Excludes firms not reporting age (43 percent of formal firms and 16 percent of informal firms). Asuncion attracts more formal firms. A disproportionate share of formal economic activity takes place within Greater Asuncion, where only a quarter of firms are unregistered. The concentration of informal firms is much higher in the Eastern region, home to 29,000 informal (non-agriculture) firms compared the 17,000 in Greater Asuncion. Furthermore, the rates of informality are higher in the Eastern region (44 percent), the North-West (44 percent) and Central-South (43 percent, 25,000 informal firms). Informal firms tend to be less productive than formal firms, but average productivity is low in most formal firms as well (Figures 6 and 7). Whereas we might anticipate that informal firms have low productivity, the productivity levels of formal and informal firms are strikingly similar. Over half of formal firms, as well as the majority of informal firms, have average productivity levels below the minimum wage. Therefore, these firms cannot afford to comply with labor and other regulations. Figure 6 Informal firms are less productive, but not by much Note: Productivity is measured as a firm’s value added divided by its total employment. 10 Figure 7 Formal firms have low productivity, many below minimum wage Note: Formal firms only. Productivity (vertical axis) is measured as a firm’s value added divided by its total employment; average firm wage (horizontal axis) is measured as the firm’s wage bill divided by its total employment; the vertical line denotes the minimum wage. Most firms in Paraguay are micro-sized, but large firms create a lot of jobs. The vast majority of firms in the 2011 firm census have fewer than 10 employees (Figure 8). This is a pattern common in developing economies, even among formal firms (Figure 9). Whereas micro-firms outnumber all other firm-size categories, they generate only half of all jobs. Large firms – that is, those with more than 100 workers – account for a quarter of all firm-based jobs12 in the economy (Figure 10). 12 Excluding household enterprises and self-employed farmers. 11 Figure 8 Most firms have fewer than 10 employees (number of firms by number of employees) 200000 150000 100000 N. 50,000 0 Firm size 1-9 10-19 20-99 100+ Figure 9 Micro-firms Are Predominant in the Developing World, even among Formal Firms Note: Data reflects only formal firms with at least 1 paid employee. 12 Figure 10 Large firms employ a significant share of the workforce (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+ Young firms are under-represented in comparative terms, implying low entry rates.13 Information on firm age is missing for a significant share of firms, especially formal firms, and may be due to evasion.14 For firms reporting age, a large share of Paraguay’s firms is less than six years old (figure 11), but compared to other countries, Paraguay’s share of young firms is slightly low, and the employment share of these young firms is even lower (figure 12). This may signal difficulties with market entry for new firms and/or market exit for older but unproductive firms. Low entry rates are corroborated by the panel registry15 of firms collected in 2013, 2014 and 2015; this data shows that the share of young firms declined from 14 to 8 percent in the space of three years, and translated into a decline in the share of new jobs added by entering firms. 13 A young firm is defined as 1-5 years old. A new entrant is defined as 1 year old. 14 Data quality issues are discussed in the Annex. Data on firm age is missing for 43 percent of formal firms (high compared to other countries) and 16 percent of informal firms (consistent with other countries). The composition of formal firms for which age is missing is similar to the overall population of firms, such that their omission does not bias the analysis. 15 Refers to the pre-census of firms collected by DGEEC. 13 Figure 11 Distribution of firms by age (years) 50,000 40,000 30,000 N. 20,000 10,000 0 Firm age 1 2-5 6-9 10-19 20-29 30+ Note: Excludes firms not reporting age. Figure 12 International comparison of proportion of young formal firms: Paraguay ranks slightly low Note: Shares of formal firms reporting age and with at least one paid employee Source: Authors’ calculations. Most firms and jobs are in the commerce sector. Nearly two-thirds of firms and half of all employment in private sector firms are in commerce activities (Figures 13 and 14). Seven in ten informal firms are in the commerce sector, while two in ten are in services (half of which are hotels or restaurants). Even among formal firms with at least one paid employee, commerce sector firms account for over half, and provide 40 percent of wage employment. Commerce sector firms and especially other services are small on average (6 and 3 paid employees, respectively), compared 14 to mining firms (40 employees on average), agro-processing firms (23 employees) and construction firms (14 employees).16 Services and manufacturing firms’ size varies by subsector, but both sectors are important sources of jobs (respectively accounting for 29 percent and 24 percent of wage employment, and 26 percent and 19 percent of total employment). Figure 13 Figure 14 Commerce sector firms dominate Half of jobs are in commerce (share of firms) (share of employment) 1.33% 3.78% 11.87% 23.81% 26.06% 19.32% 62.99% 50.83% MinUtilConstr Manufacture MinUtilConstr Manufacture Commerce Services Commerce Services Commerce sector firms appear to be the most agile in starting operations, including among formal firms with at least one paid employee (figure 15). Three-fifths of new firms in 2010 started commerce operations, compared to 20 percent in the services sector, and 16 percent in manufacturing. New entrants in commerce also created the most jobs among total new firms, although manufacturing was a close second. The largest number of firms entered in the Eastern and Central South regions, followed by Greater Asuncion. Figure 15 Most new entrants were in commerce (sector composition of firms entering in 2010 with at least one paid employee, %) 16 Average firm sizes are for formal firms with at least 1 paid employee. 15 Women are concentrated in less productive activities, even within the formal sector. Evidence from the household survey Encuesta Permanente de Hogares indicates that women are much more likely than men to work in commerce, hotels and restaurants and other services, all sectors with low average productivity.17 These findings are corroborated by the firm census data, which shows that most jobs are held by men in formal firms (although the job itself may be informal, defined as lacking social insurance coverage), and firms in the mining, utilities and construction sector and in manufacturing are particularly dominated by men (Figure 16). Women, by contrast, are more likely to work in commerce and services. Figure 16 Unequal gender distribution across sectors Note: MinUtilConstr denotes mining, utilities and construction. 17 Ruppert Bulmer et al. (2017) 16 4. Estimating Formal Firm Performance18 The descriptive statistics presented above belie significant heterogeneity in private sector activity across sectors, types of production, capital-intensity, and target markets (i.e., domestic or external), among others. Differences in firm characteristics may affect the performance and productivity of individual firms. We therefore use regression analysis to test the statistical significance of a set of key variables in explaining the following dependent variables: (a) firm productivity, (b) firm size, and (c) firm-level wages. The analysis uses the 2011 firm census data on formal registered firms with at least one paid employee to estimate contemporaneous correlations.19 The potential explanatory variables considered include firm size, age, sector of activity, gender composition of the work force, capital intensity, and region, among others. We create a smaller, panel dataset by merging data from the 2011 census and the 2015/2016 firm survey for firms that are observed in both points in time, namely 2010 and 2014. This panel data is used to test for the main drivers of firms’ productivity growth, firms’ employment growth and firm-level wage growth over that period. A. Correlates of Productivity Firm size matters for productivity. The positive, significant correlation between value-added per worker and firm size means that larger firms are more productive than smaller firms (Figure 17; Annex Table 1). This correlation is comparatively strong in Paraguay, relative to other countries where a similar analysis was done; cross-country regressions find an especially large positive link between productivity and firm size in Paraguay, controlling for other factors such as age, sector and location (Figure 18).20 Given the evidence that smaller firms are significantly less productive, the high share of micro firms in Paraguay and the very modest share of large firms (more than 100 employees) raises concern. 18 The remainder of the analysis applies to formal firms with at least one paid employee, employment refers to paid employment only, and productivity is measured as value added per paid employee. Self-employed entrepreneurs are excluded. 19 Note that the regression analysis does not establish any causal relationship, only correlation, 20 Note that in Afghanistan and Uganda, smaller firms are more productive, while in Moldova and Peru, there is no statistically significant correlation between productivity and firm size. 17 Figure 17 Estimated correlates of firm value added per paid employee (coefficient values) Note: Detailed regression results are in Annex Table 1, specification 4. Reference sector is manufacturing. Reference region is Asunción Figure 18 Cross-country regression analysis on the link between productivity and employment (2010) Estimated Coefficient: Value-Added per Worker Afghanistan Cote dIvoire NS Moldova Countries NS Peru PRY Paraguay Uganda Vietnam -.2 -.1 0 .1 .2 Employment (log) Note: Reported coefficient values come from OLS regressions of the log of real value-added per worker on independent variable firm employment level and controls for age, sector, ownership, and firm location, clustered into 5 sector categories and 3 region categories. Coefficients are significant at least at 95% (confidence intervals shown as horizontal bars); NS indicates not significant at 90%. Source: Authors’ calculations. 18 The positive correlation between firm productivity and firm size is even more pronounced among the most productive firms. That is, firms in the most productive quartile have significantly more employees relative to the lower quartiles (Figure 19; Annex Table 2). This suggests the possibility of a threshold effect above which firms can compete more easily and expand their employment. The productivity gap between micro and large firms is comparatively wide in Paraguay vis-à-vis other countries (Figure 20). The high degree of market concentration in Paraguay likely plays a key role in squeezing out smaller players. In many 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 percent (see Annex Table 3). Even in the retail sector, where entry is relatively easy, sector concentration is unexpectedly high. Figure 19 The most productive firms are the largest by far Source: Authors’ calculations. 19 Figure 20 International comparison of productivity gap between micro and large firms (output per worker, ratio of micro over large) Note: Output is measured by sales. Source: Authors’ calculations. Productivity varies by sector. Some sectors are correlated with high productivity, others with low productivity. Construction, retail, financial services and transport and communications have a positive link to value-added per worker, controlling for firm size (Annex Table 1).21 Food and beverage manufacturing, hotels and restaurants, utilities and especially apparel manufacturing are negatively correlated with value-added per worker. So are firms in tradable manufacturing sectors compared to those in non-tradable manufacturing, again controlling for firm size. These results can be compared with the observed differentials in average productivity across subsectors, where apparel, other services (for example, security and landscaping), and hotels and restaurants are among the least productive (Figure 21). 21 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. Other distortions may arise from 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. The data do not allow us to estimate whether any particular sectors have more unreported paid workers. 20 Figure 21 Comparing value-added per worker across 2-digit industries Note: Reference sector is apparel. Regression on pooled 2010 and 2014 data. Data labels are the ISIC Rev. 4 two-digit codes. Categories are 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. Productivity differences between firms in different sectors actually declined between 2010 and 2014, a period of rapid formal job creation. Firms in the transport and communications sector, which has above-average productivity, experienced relatively slow productivity growth. At the same time, the hotel and restaurant sector had significantly faster productivity growth during the period compared to firms of a similar size and age in other sectors (Annex Table 4). These trends, which reflect a reversion toward the mean value-added per worker, might differ if we were able to observe other factors of production that allowed us to measure total factor productivity controlling for physical and human capital inputs, for example. Firms with more capital assets are more productive, based on the very small sample of firms reporting capital assets. Internet use is one example of fixed capital for which there is data in the firm census (for 24,000 medium and large firms). We therefore estimate the role that access to basic internet technology plays in enhancing productivity, and find that those that use the internet to receive purchase orders are significantly more productive than those that do not (Annex Table 21 5).22 But we also find that internet use in 2010 has no predictive power of higher productivity in 2014 (Annex Table 6). This finding suggests that, whereas more productive firms may use the internet in their sales operations, this usage does not make a significant difference in firms’ ability to raise their productivity. As firms age, they become slightly more productive; young firms are less productive than those in operation for more than five years (Annex Table 1). This trend is consistent with the productivity distribution of firms shown in Figure 22, and may be influenced by the fact that young firms are less able to invest in capital assets.23 When we test for the effects of firm age on productivity growth, we find no impact; that is, older firms did not have faster productivity growth between 2010 and 2014 (Annex Table 4). Figure 22 Productivity distributions vary little by firm age (years) .4 .3 density .2 .1 0 4 6 8 10 12 14 Value Added per worker -real LCU (log) 1-5 6-9 10+ Firms in Greater Asunción are significantly more productive compared to those in the Central-South region, even controlling for sector and firm size. Productivity of firms operating in Asuncion is about 20 percent higher than firms operating in the Central-South region, although this could be driven by higher prices rather than greater technical efficiency. Positive agglomeration effects around the capital – together with the higher average wages paid in Greater Asuncion (discussed below) – are consistent with the rapid urbanization and urban-based job creation observed over the past decade.24 There is no statistical difference in productivity between firms in the capital and those in the North-West or Eastern regions, however. 22 Firms that use the internet to receive purchase orders are also larger (Annex Table 5). 23 Defining productivity as value-added per worker gives a larger weight to capital assets in the productivity measure. 24 Ruppert Bulmer et al. (2018). 22 Female-dominated firms are less productive. Women tend to work in low productivity jobs that are poorly remunerated, and the gender earnings gap is wide.25 Part of this gap is due to women’s lower likelihood of holding a formal job, but a significant part is due to gender differences in sector of work. The firm-level data provides complementary evidence of sectoral segregation under which women are relatively absent from more productive sectors. Firms that have a majority- female work force are less productive than majority-male firms (controlling for other factors). This is partly because women tend to work in smaller firms (Annex Table 2), majority-female firms tend to be smaller on average (except for apparel manufacturing, which is low-paid), and female employees are disproportionately represented in the commerce sector and hotels and restaurants. B. Correlates of Firm Size and Growth Younger firms are smaller than older firms and grow more slowly compared to those over age 5 (Figure 23; Annex Tables 2 and 7). Firms under age 6 experienced significantly slower growth in number of employees than older firms. This may be explained by the fact that young firms are less productive and therefore struggle to compete and catch up to larger, more established firms through job creation. Paraguay ranks low when comparing its firm size-firm age distribution with those of other developing countries (Figure 24). The low firm entry rates in Paraguay compared to other economies26 may be a compounding factor behind the low share of jobs within young firms. This would arise if, for example, potentially dynamic firms cannot enter the market. Figure 23 Estimated correlates of firms’ employment growth (coefficient values) Note: For details, see Annex Table 7, specification 5. Reference sector is manufacturing. Significance: *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ calculations. 25 Mincer regressions controlling for a number of factors estimate a gender wage gap equal to 25 percent when measured using hourly wages of full-time wage employees, and 43 percent when measured using monthly wages of all workers including informal and self-employed. See Ruppert Bulmer et al. (2017). 26 In 2010, Vietnam had an entry rate of 23 percent, 8 percent in Peru, 5 percent in Moldova, and 3 percent in Paraguay. 23 Figure 24 Average firm employment by age (employment indexed to 1 when firm age < 5) Note: This figure plots contemporaneous size-age combinations, not dynamic changes over time. Source: Authors’ calculations. Larger firms (50 or more employees) grow more quickly. Considering the dynamic effects of firm size on employment growth using the panel data, firms with 50 or more workers had significantly higher employment growth than smaller firms – on the order of 30 percent higher (Annex Table 7). Cross-country comparisons suggest that Paraguay has a relatively high proportion of micro firms (1-9 employees) with an average age over 10 years old compared to other countries (Figure 25).27 27 Twenty-three percent of firms in Paraguay are micro firms compared to only 7 percent in Vietnam, and 14 percent in Kosovo. 24 Figure 25 Micro firms struggle to grow over time: international comparison 7 10 14 14 17 18 18 18 19 22 23 27 28 35 36 39 48 100 93 90 86 86 83 82 82 82 81 78 80 77 73 72 65 64 61 60 52 Share 40 20 0 o ola so y am ue ov fa b ia is t an ir e ta n on e ua ov a da sh rd e Pe ru fr i ca et n biq s g m Iv o nis Le ag old an de Ve Vi am Ko An ina Za jik d ha ar Ug gla hA rk Ta e rra P M n bo ut oz Bu Co t f g e a a So M A Si B C % Micro Firms Age<10 % Micro Firms Age 10+ Source: Authors’ calculations. Firms that have high value-added per worker or are more capital-intensive add fewer jobs. We find that firms with higher initial productivity had slower job growth (Annex Table 7). Firms that are more capital-intensive are less likely to create new jobs compared to their more labor- intensive counterparts. These results are consistent with the economics literature that production technologies that rely on capital cannot easily accommodate more labor without adjusting capital as well. Patterns in firm size vary across sectors. Firms that belong to the commerce or services sectors tend to be significantly smaller than firms operating in the manufacturing sector (controlling for other factors including age, region and sales). But there is little link between sector and rate of job growth. Firms with a higher proportion of female employees tend to be smaller, even when we control for sector, among other things. And firms with a majority female workforce are 11 percent smaller than those with a majority male workforce, other things being equal (Annex Table 2). 25 C. Correlates of Wages28 Wages rise with firm age. Older firms pay slightly higher wages (Figure 26; Annex Table 8). But firm age is ruled out as a determinant of higher wage growth (except for firms over 30 years old, which experienced 30 percent faster growth in wages between 2010-2014 compared to young firms (Annex Table 9)). The relationship between firm wages and firm size is mostly positive. Average labor cost per paid worker rises with firm size (continuous variable), but falls for firms with more than 10 paid employees when controlling for sector and sales level. But when we add controls for productivity level, larger firms pay more. It is notable that very large firms (250 or more employees) experienced much higher wage growth (about 11 percent higher) compared to micro firms (1-9 employees). Firms with 10-19 employees and those with 20-49 employees, by contrast, experienced somewhat slower wage growth than micro firms (6 percent and 12 percent slower, respectively).29 Figure 26 Estimated correlates of average firm wage per paid employee (coefficient values) Note: For details, see Annex Table 8, specification 1. Reference sector is food and beverage manufacturing. Reference region is Asunción. Significance: *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ calculations. 28 A firm’s average wage is calculated as its wage bill divided by total paid employees. 29 Note that these firm-size results may be slightly biased, given that micro-sized firms (i.e., those with 1-9 paid employees) are not very well captured in the firm survey or the panel (see the data discussion in Annex) 26 Wages are positively correlated with productivity and output, but wage growth is negatively correlated with productivity (Annex Tables 8 and 9). Firms with higher revenues tend to pay higher wages. The positive wage returns rise further for firms in the top productivity quartiles, a pattern similar to that observed in other developing economies (Figure 27). Economic theory suggests that firms that produce goods or services of higher value using more skilled workers and more sophisticated capital and/or human capital inputs will pay higher wages than other firms. The evidence for Paraguay provides some support for this claim (although we cannot rule out that higher productivity and wages are partly driven by higher prices). Despite higher wages, however, more productive firms experienced slightly slower wage growth during the 2010-2014 period. Figure 27 Labor cost per worker rises with firm productivity (estimated coefficient values) Note: Labor cost per worker is measured in 2010 USD. Reported coefficient values come from OLS regressions of the log of real labor cost per paid worker on the following dependent variables: age, sector, firm size category, ownership, and firm location, clustered based on 5 sector categories, 3 region categories and 3 size categories. Standard errors are robust, UGA indicates Uganda, CIV indicates Cote d’Ivoire, PER indicates Peru, MDA indicates Moldova, VNM indicates Vietnam, and AFG indicates Afghanistan, Source: Authors’ calculations. Regional and sectoral wage effects are significant. Wages are higher in Greater Asuncion compared to all other regions, in the range of 8 to 29 percent higher, controlling for other factors. Sector of work also matters for wages. Metals and machinery manufacturing, financial services, 27 construction, and transport and communications have the highest wages, while hotel and restaurant jobs pay the least (29 percent less than jobs in food and beverage manufacturing). Majority female firms are characterized by much lower wages, other things being equal. With respect to wage growth, only the utilities and hotel and restaurant sectors show significantly slower wage growth than food and beverages; we observe no other statistically significant sector effects. When only tradable versus non-tradable sectors are considered, wages in the tradables sector grew much faster than in non-tradables, but started from a much lower base. Market concentration had no significant impact. 5. Conclusions The main findings that emerge from analyzing the firm-level datasets are as follows. • Nearly two-fifths of Paraguay’s private sector firms are informal. And even among formal registered firms, over three-fifths are self-employed entrepreneurs or family firms with no paid employees. • Firm productivity levels and wages are generally low, including among formal firms, except for a relatively small number of high-productivity firms. This means that registering as formal has little impact on firm performance. Many firms cannot afford to comply with regulations, particularly the large share whose labor productivity is below the minimum wage. • Over 90 percent of firms have fewer than 10 employees (micro firms). Being small may be inherently challenging in terms of scale and ability to compete with larger and more capital- intensive firms. • The productivity gap between micro and large firms in Paraguay is especially wide by international standards. This suggests the possibility of threshold effects and/or uncompetitive, highly concentrated markets that crowd out smaller players. Larger firms are more productive, and medium and large firms created the most jobs between 2010 and 2014 compared to smaller firms. • Despite the dominance of micro firms, nearly a quarter of firm-based employment in Paraguay is in large firms. • Firm entry rates in the formal sector are low. Commerce sector (non-tradable) firms are the most prevalent and have the highest entry rates, but they tend to be small, pay modest wages and have low capital intensity. Whereas the commerce sector is effective at absorbing labor force entrants, its concentration in low-skilled labor-intensive work limits the scope for significant technological innovation and productivity growth. • After firm size, the most significant correlate of labor productivity (measured by value- added per worker) is firm age. Both correlates of low productivity – small size and young age – may be influenced by the fact that these firms are less able to invest in capital assets, given our productivity definition as value-added per worker, which gives a larger weight 28 to capital assets than the alternative definition of total factor productivity. In addition to being less productive, young firms are also afflicted by slower growth. Older firms (especially over age 30) grow significantly faster. • Location in Greater Asunción also matters. Firms in Greater Asuncion have higher value- added per worker, pay higher wages, and likely benefit from agglomeration effects that increase the pool of skilled labor and boost productivity. But we cannot definitively rule out that the higher productivity observed in Asuncion-based firms is driven by higher prices rather than greater technical efficiency. • Women tend to be concentrated in the commerce and services sectors, especially the low- productivity sub-sectors of retail, hotels and restaurants, and other services. Women also dominate apparel manufacturing, one of the least productive and lowest paying manufacturing activities. Firms with more than 50 percent female workers tend to be smaller, less productive and pay significantly lower wages than majority-male firms, controlling for other factors. These findings point to some important challenges for the continued development of a healthy, dynamic and inclusive private sector. The patterns of informality and firm entry, job creation and productivity reflect a pervasive duality between micro-sized low-productivity firms on one end, and a small number of highly productive firms with controlling stakes in concentrated markets on the other end. Gender segregation by sector – where women are more likely to engage in low- productivity employment – contributes to gender gaps in job quality and earnings. And the widening urban-rural divide between metropolitan Asuncion and the rest of the country where once-dominant agriculture is losing its appeal for migrating youth is another factor feeding economic dualism. Despite solid GDP performance, the limited diversification of the economy and high rates of informality and evasion create a challenging private sector environment for firms. The prevailing firm structure is characterized by micro-firms that lack scale economies, are concentrated in non- tradables, have generally low productivity levels, and rely on unskilled employment. Today’s private sector is not well-positioned to meet the expected rise in consumer demand for goods and services of increasing quality. In order for Paraguayan firms and workers to benefit from a virtuous cycle of increased demand for more sophisticated inputs requiring more skilled labor in more productive and better-paying jobs, Paraguay will need a more dynamic private sector landscape where firms can enter, create employment, and upgrade productivity sufficiently to compete on the local and international stages. Although large firms have the capacity to create more employment, 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 indirect job creation and ultimately constrain economic growth. This analysis provides new insights into firm performance and labor demand in Paraguay, but knowledge gaps remain relating to the specific drivers of low firm entry, slow firm growth and limited productivity gains. Policies certainly play a role, but which policies and to what extent they 29 matter are not well understood. Government action to review its competition policy framework and ensure a level playing field through neutral tax and regulatory policies would be important first steps. Future policy considerations should include measures to accelerate firm entry, reduce barriers to formalization, raise firm productivity of micro-entrepreneurs as well as SMEs with growth potential, and help productive firms gain access to markets (which ultimately will create more employment). But effective policy design for alleviating the binding constraints to firm performance and growth will require more information regarding firms’ production structure, use of physical and human capital and technology, and the associated costs of production inputs. Information gaps could be filled in part by strengthening existing survey instruments and data quality by (i) collecting more information on firm-level variables (e.g., production inputs, labor quality, skill level and/or occupation, wages for each worker or average wage per occupation), (ii) expanding sectoral coverage to include all sectors, and (iii) more careful tracking of firms over time to capture firm exit and better capture firm entry. Whereas firm-level data provides only a partial picture of the economy and labor force, it nevertheless provides an important perspective of the private sector’s operation and prospects for growth and job creation. It would be useful for future data collection efforts to give specific consideration to how the household surveys and firm surveys could complement each other to provide a more comprehensive and integrated reflection of output and employment in Paraguay. 30 References Kuddo, A. and E. Ruppert Bulmer. 2017. “An Assessment of Labor Regulations in Paraguay”, Jobs Group, World Bank (mimeo). Ruppert Bulmer, E., S. Watson, D. de Padua and A. Garlati. 2017. “Jobs Diagnostic Paraguay: The Dynamic Transformation of Employment in Paraguay”, Jobs Series No. 10, Jobs Group, Washington, DC: World Bank. World Bank. 2012. World Development Report 2013: Jobs. Washington, DC: International Bank for Reconstruction and Development. World Bank. 2018. “Paraguay Systematic Country Diagnostic”, Washington, DC: World Bank Group. 31 Annex Data Sources for the Analysis The data sources used in this analysis provide information on private sector firms and employment in Paraguay from 2010 to 2015. Each dataset was collected by Paraguay’s Dirección General de Estadística, Encuestas y Censos (DGEEC), and provides information at the level of the firm. These datasets include information on key aspects of firms’ hiring behavior and performance, such as employment level, annual sales, wage bill, 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 firm 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 or skill level or occupation. The datasets are not perfectly comparable; the analysis therefore uses each source separately, and provides caveats when making comparisons. 1. The 2011 Firm Census – Censo Económico – is a national census that includes data on firm characteristics such as employment, sales and value-added as of 2010. The census covers all regions of Paraguay, and identifies 211,042 firms employing 799,153 workers, including paid, unpaid and commissioned30 employees. The census includes both formal and informal firms, as well as the 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 430,799 workers (recall Table 1). The firm size thresholds used 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. The analysis defines new firm-size categories to facilitate international comparisons (i.e., 1-9 paid employees; 10-19; 20-99, 100 and over). Quality issues: Whereas response rates were mostly strong, two key variables have large numbers of “missings”: 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 distribution of firms not reporting age do not appear to be biased. Forty-nine percent of firms with at least one paid employee and not reporting age are in the commerce sector, and account for 42 percent of employment. In the total sample of firms (i.e., those both reporting and not reporting age), commerce sector shares are 55 percent of firms and 43 percent of employment. Similarly, nearly half of the non-reporting firms are located in Greater Asuncion, close to the overall population weight. • 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. 30 Tercerizado or comisionista. 32 • 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. We omit the variable from the regression analysis, due to the sharply reduced sample size, but we test the variable for significance as a robustness check to see how its inclusion alters the regression estimates. 2. The 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, but for manufacturing, commerce and non-financial services, it is representative at the 2-digit level. 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. Quality issues: 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. 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. Quality issues: Firm entry and exit are only partly observed, because the dataset reflects a large sample rather than a complete census. 4. 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 Census and the 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 Census and the Survey, although the panel firms are slightly larger, marginally more productive and pay modestly higher wages (Figure A.1). 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. 33 Figure A.1 Comparing Data Across the Census, Survey and Panel 34 Figure A.1 (continued) Comparing Data Across the Firm Census, Firm Survey and Panel Note: Only formal firms with at least one paid employee are plotted. The panel is also highly representative vis-à-vis the sectoral composition of employment and firms within manufacturing, commerce and non-financial services (Figure A.2). Figure A.2 Comparing Sector Composition between the Firm Census and Firm Survey Note: Only formal firms with at least one paid employee. The number of micro-sized firms (i.e., those with 1-9 paid employees) is not very well captured in the firm survey or the panel, creating a risk that the growth regression results are somewhat biased with respect to firm size categories. 35 Annex Table 1: 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) 36 Continued (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 37 Annex Table 2: 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) 38 Continued (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) 39 Continued (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 40 Annex Table 3: High Market Concentration in Certain Industry and Services Sub-Sectors (ranking the top 10 sub-sectors by sales; market share of the top 4 firms in each sub-sector) Sub-Sector Sub-Sector Market Share of Employment/Total Sales/Total Top 4 Firms in Sector Sector Employment Sector Sales Sector Industry 3510 Electricity generation, transmission, distribution 7% 24% 100% 1010 Meat processing and preservation 10% 18% 62% 1200 Manufacture of tobacco products 2% 5% 77% 1113 Manufacture of malt liquors, malt 1% 4% 100% 1061 Manufacture of grain mill products 2% 3% 66% 2410 Manufacture of basic iron and steel 1% 3% 99% 1040 Manufacture of vegetable and animal oils and fats 1% 3% 77% 1129 Manufacture of non-alcoholic beverages 2% 2% 91% 2100 Manf. of basic pharmaceuticals (prod.&prep) 4% 2% 61% 4210 Construction of roads and railways 4% 2% 37% Services 6419 Other monetary intermediation 6% 14% 52% 4620 Wholesale agr. raw materials, live animals 3% 11% 54% 4711 Retail non-spec. with mainly food, bev. or tobacco 10% 8% 42% 4730 Retail automotive fuel in specialized stores 3% 8% 43% 4630 Wholesale food, beverages, tobacco 4% 6% 18% 4741 Retail computers, software, telecomms equipment 2% 5% 30% 4652 Wholesale electronic and telecomms equipment 1% 4% 48% 6100 Telecommunications 4% 3% 78% 4651 Wholesale computers, peripheral equip., software 0% 3% 51% 4510 Sale of motor vehicles 2% 3% 36% Note: Employment (sales) shares of total indutry employment (sales) or total services employment (sales). Industry includes manufacturing, utilities, mining and construction. 41 Annex Table 4: OLS Regression: Growth in Value-Added per Worker (change in VA per paid employee per firm, 2010-2014), panel dataset (1) (2) Size 10-19 0.309 0.337 (0.243) (0.250) Size 20-49 0.158 0.193 (0.184) (0.192) Size 50-99 -0.523*** -0.292 (0.0946) (0.200) Size 100+ -0.498* -0.453 (0.250) (0.268) age_6to9 -0.160 -0.136 (0.123) (0.156) age_10plus -0.00493 -0.0147 (0.119) (0.115) Utilities -0.0840 (0.0710) WholesaleRetail -0.0906 (0.0644) TransportStorageComm -0.451*** (0.0970) HotelsRestaurants 0.557*** (0.150) OtherServices 0.0425 (0.216) MinUtilConstr -0.0837 (0.0718) Commerce -0.0885 (0.0654) Services 0.128 (0.0796) Constant -0.229* -0.240* (0.129) (0.127) Observations 320 320 R-squared 0.125 0.163 Sector dummies NO NO Location dummies YES YES Year Dummies YES YES R2 0.125 0.163 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 42 Annex Table 5: OLS Regression: Impact of Internet Use on Log Value-Added per Worker (2010) (Sample restricted to firms with 50+ employees) Value-Added per Worker Value-Added (log) - per Worker VARIABLES Heckman (log) - OLS Age 0.0123*** 0.0130*** (0.00164) (0.00164) Age squared -0.000190*** -0.000195*** (4.14e-05) (4.22e-05) normalised herfindahl index by sector and time rS -0.247*** -0.219** (0.0853) (0.0848) Internet usage for sales 1.114*** 0.441*** (0.140) (0.0461) Constant 10.30*** 10.27*** (0.0514) (0.0565) Observations 24,271 24,271 R-squared 0.159 0.147 Sector dummies YES YES Location dummies YES YES R2 0.159 0.147 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 43 Annex Table 6: OLS Regression: Impact of Internet Use in 2010 on Log Employment, Log Value-Added per Worker and Log Wages in 2014 (Sample restricted to firms with 50+ employees) Value-Added Value-Added Employment Employment per Worker per Worker Labor Costs Labor Costs VARIABLES (log) (log) (log) (log) (log) (log) Age 0.0152 0.0457*** -0.00842 -0.000427 0.00578 0.00650 (0.0101) (0.00991) (0.00649) (0.0132) (0.0136) (0.0204) Age Squared -2.57e-05 -0.000411*** 7.23e-05 -3.05e-05 -3.99e-05 5.20e-05 (0.000114) (0.000102) (6.35e-05) (0.000121) (0.000128) (0.000234) Used Internet to Receive orders to purchase products in 2010 1.230*** 0.830*** 0.0260 0.112 0.0198 -0.0557 (0.281) (0.222) (0.174) (0.224) (0.0858) (0.0871) normalised herfindahl index by sector and time rS -0.711* -0.836 0.463** -0.776 0.109 0.430 (0.369) (0.637) (0.192) (0.609) (0.150) (0.425) Capital -real (log) 0.208*** 0.111** 0.0760*** (0.0633) (0.0425) (0.0241) Value-Added (log) Quartile 2 -0.121 0.229* (0.112) (0.119) Value-Added (log) Quartile 3 0.117 0.233 (0.136) (0.153) Value-Added (log) Quartile 4 0.108 0.0494 (0.166) (0.264) Employment (log) -0.281 -0.596* -0.0254 -0.0770 (0.179) (0.300) (0.0960) (0.161) Employment (log) - squared 0.0474** 0.0675** 0.00702 0.00565 (0.0221) (0.0310) (0.0122) (0.0208) Constant 3.077*** -0.736 11.39*** 10.51*** 9.858*** 7.876*** (0.361) (1.476) (0.446) (0.970) (0.315) (0.691) Observations 673 437 673 437 744 486 R-squared 0.414 0.607 0.404 0.481 0.362 0.424 Sector dummies YES YES YES YES YES YES Location dummies YES YES YES YES YES YES R2 0.414 0.607 0.404 0.481 0.362 0.424 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 44 Annex Table 7: 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) 45 Continued (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 46 Annex Table 8: 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) 47 Continued (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) 48 Continued (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 49 Annex Table 9: OLS Regression: Wage Growth (change in firm wage bill per paid worker, 2010-2014), panel dataset (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) sza_10to19 -0.0760** -0.0615* -0.0615* -0.0536 -0.0658** -0.0745** -0.0228* -0.0621*** -0.0233 -0.0513* -0.0553 (0.0348) (0.0296) (0.0296) (0.0323) (0.0243) (0.0281) (0.0129) (0.0183) (0.0189) (0.0264) (0.0360) sza_20to49 -0.102 -0.149* -0.149* -0.116 -0.111* -0.119* -0.116 -0.131 -0.116 -0.175* -0.122 (0.0795) (0.0718) (0.0718) (0.0785) (0.0632) (0.0636) (0.0774) (0.0865) (0.0760) (0.0897) (0.0748) sza_50to99 -0.0340 -7.69e-05 -7.69e-05 0.00391 0.0648* 0.0230 0.00909 0.00592 0.00915 0.0226 0.00167 (0.0284) (0.0453) (0.0453) (0.0433) (0.0346) (0.0465) (0.0520) (0.0465) (0.0370) (0.0495) (0.0445) sza_100plus 0.0406 0.115 0.115 0.112 0.102** 0.0929* 0.132** 0.0907 0.128* 0.0759 0.113 (0.0414) (0.0716) (0.0716) (0.0723) (0.0454) (0.0487) (0.0624) (0.0630) (0.0674) (0.0609) (0.0722) age_6to9 0.0277 0.0277 0.0383 -0.00645 -0.00875 -0.0173 0.00375 -0.00248 -0.0101 (0.0678) (0.0678) (0.0683) (0.0470) (0.0510) (0.0513) (0.0445) (0.0544) (0.0379) age_10plus -0.00673 -0.00673 -0.0173 -0.0244** -0.0218* -0.0166 -0.0192 -0.00987 -0.00721 (0.0164) (0.0164) (0.0203) (0.0110) (0.0115) (0.0139) (0.0182) (0.0135) (0.00919) Norte -0.0602 -0.0672 (0.0428) (0.0409) Occidental 0.0170 0.0169 (0.0229) (0.0235) CentroSur 0.0339 0.0374 (0.0347) (0.0376) MinUtilConstr -0.152*** (0.0383) Commerce -0.0134 (0.0301) Services -0.0746** (0.0339) Utilities -0.152*** (0.0394) WholesaleRetail -0.0129 (0.0309) TransportStorageComm 0.0310 (0.0768) HotelsRestaurants -0.182*** (0.0308) OtherServices -0.0389 (0.0307) 50 Continued (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Sales/worker log (previous period) -0.0610*** (0.00850) Value Added/worker log (prev. period) -0.0763*** (0.0181) Value Added/worker log (prev. period) quartile 2 -0.205*** (0.0518) Value Added/worker log (prev. period) quartile 3 -0.176*** (0.0571) Value Added/worker log (prev. period) quartile 4 -0.199*** (0.0369) Constant 0.115** 0.138*** 0.138*** 0.128*** 0.124*** 0.126*** 0.818*** 0.243*** 0.935*** 0.322*** 0.119*** (0.0417) (0.0452) (0.0452) (0.0365) (0.0263) (0.0278) (0.0791) (0.0659) (0.168) (0.0532) (0.0374) Observations 502 343 343 343 343 343 317 317 326 326 343 R-squared 0.264 0.296 0.296 0.215 0.127 0.173 0.363 0.374 0.372 0.425 0.216 Sector dummies YES YES YES YES NO NO YES YES YES YES YES Location dummies YES YES YES NO YES YES YES YES YES YES YES Year Dummies YES YES YES YES YES YES YES YES YES YES YES R2 0.264 0.296 0.296 0.215 0.127 0.173 0.363 0.374 0.372 0.425 0.216 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 51 52 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/