Policy Research Working Paper 10739 The Role of Firm Dynamics in Aggregate Productivity, Job Flows, and Wage Inequality in Ecuador Fausto Patiño Peña Esteban Ferro Macroeconomics, Trade and Investment Global Practice & Finance, Competitiveness and Innovation Global Practice March 2024 Policy Research Working Paper 10739 Abstract This paper examines the role of firm dynamics in aggregate than the bottom quintile, after controlling for firm size and total factor productivity, job flows, and wage inequality in age. The findings also provide evidence of credit misalloca- Ecuador. Utilizing a comprehensive employer-employee tion across firms. Additionally, industries with higher job dataset, the paper documents firm dynamics and job flow mobility, credit access, and competition and lower non- patterns that are consistent with the presence of market wage labor costs, minimum wage incidence, and zombie distortions. Also, the paper identifies factor misallocation as firms demonstrate higher allocative efficiency. Moreover, the main contributor to Ecuador’s total factor productivity worker-level regressions indicate that misallocation drivers deceleration. Given these trends, the paper explores alloc- explain up to 41 percent of wage inequality, with non-wage ative inefficiency drivers through firm- and industry-level labor costs and product market frictions as distortions driv- regressions. Firms in the top productivity quintile face ing this inequality. distortive non-wage labor costs that are 3.7 times higher This paper is a product of the Macroeconomics, Trade and Investment Global Practice and the Finance, Competitiveness and Innovation Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at fpatinopena@worldbank.org and eferro@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team The Role of Firm Dynamics in Aggregate Productivity, Job Flows, and Wage Inequality in Ecuador* Fausto Patiño Peña and Esteban Ferro++ JEL Codes: D2, J21, J3, L11, L25, O47 Keywords: Firm Performance, Aggregate Productivity, Job Flows, Wage Inequality * For helpful comments we thank Besart Avdiu, Elwyn Davies, Christian Gonzalez, Mariana De La Paz Pereira Lopez, and Julio Velasco.  fpatinopena@worldbank.org. ++ eferro@worldbank.org. 1 Introduction This paper studies the role of firm dynamics in aggregate total factor productivity (TFP), job flows, and wage inequality in Ecuador. A large strand of literature has documented that policy-related drivers and structural frictions impact firm performance with significant effects for aggregate TFP (Restuccia and Rogerson [2008, 2013], Hsieh and Klenow [2009], Bartelsman et al. [2013], among others). Recent literature has found that distortions that impact firm-level dynamics are also important for worker-level outcomes, including employment flows and wage inequality (Haltiwanger et al. [2013, 2016], Decker et al. [2014], Barth et al. [2016], Manning [2021], Criscuolo [2020, 2021, 2022], among others). To examine how firm performance drivers affect aggregate TFP and, simultaneously, impact job flows and wage inequality, the paper leverages a comprehensive employer-employee panel dataset encompassing formal firms in Ecuador. At the firm level, this paper documents a substantial overlap in the productivity distributions of incumbent firms versus entrants and survivor firms versus exiters. Additionally, firm productivity growth is negligible across firms’ life cycle. These firm dynamics suggest underlying market frictions, patterns that are corroborated by macro-level empirical trends. By decomposing aggregate TFP growth, the paper identifies factor misallocation across firms as the predominant channel contributing to Ecuador's aggregate TFP deceleration. Moreover, the distributions of firms by size and age and job flow trends are consistent with persistent allocative frictions, as larger and more productive firms are underrepresented in the economy, new jobs are predominantly created by small, unproductive entrants, and existing jobs are primarily destroyed by old, unproductive firms that continue operating instead of exiting. To reconcile these firm-level and aggregate trends, the paper investigates structural and policy drivers of allocative inefficiencies. Regression analysis at the firm level shows that firms in the top quintile of the productivity distribution face distortive non-wage labor costs that are 3.7 times higher than the bottom productivity quintile, after controlling for firm size and age. Also, the paper finds that credit is allocated to both productive and unproductive large firms, while productive small firms face higher credit constraints. Moreover, industry-level regressions indicate that sectors with higher job mobility, lower non-wage labor costs, less exposure to the minimum wage policy, improved credit access, fewer zombie firms, and greater market competition exhibit higher allocative efficiency. Furthermore, the paper finds that drivers contributing to resource misallocation account for up to 41 percent of aggregate wage inequality. Worker- level regressions indicate that non-wage labor costs and product market frictions are allocative distortions driving aggregate wage inequality, while the minimum wage policy dissipates this inequality despite generating misallocation. Between 2012 and 2020, the sample period studied in this paper, Ecuador witnessed economic stagnation, as growth driven by factor accumulation was offset by a significant decline in aggregate TFP, starting in 2014. This suggests that aggregate TFP growth was constrained by the presence of institutional and regulatory distortions that kept factor inputs from being efficiently distributed across firms and sectors. These distortions also slowed down firm-level productivity growth and impeded healthy firm entry and exit dynamics. This decline in aggregate TFP coincided with a change in the trend of income inequality, whose previous decreasing pattern plateaued after 2014 and slowly began increasing in 2018. By 2020, Ecuador was the third-most unequal country in South America. These patterns posit the question of whether the same drivers that constrain aggregate TFP performance are also linked to labor market outcomes and aggregate inequality trends. Market distortions deter aggregate TFP performance by constraining firm outcomes and dynamics. In particular, these distortions diminish aggregate TFP, impeding the efficient allocation of capital and labor across firms (i.e., the allocative efficiency channel) and sectors (i.e., the structural transformation channel), restricting the growth of firm-level technical efficiency (i.e., the technical efficiency channel), and yielding unfavorable dynamics in firm entry and exit (i.e., the market selection channel). These market frictions, which restrain firm outcomes and depress aggregate TFP, also have effects on workers by distorting labor market quantities and prices. In an economy without distortions, heterogeneity in firm-level productivity results in differences in firm-level employment, with more productive firms employing a larger workforce 2 compared to less productive firms. As markets are efficient, marginal products of labor equalize across firms, resulting in equal wages for workers with similar characteristics. However, when input and output markets face structural or regulatory distortions, marginal products of labor no longer equalize across firms, which leads to wage inequality among similar workers for two reasons. First, market distortions generate labor market rents that can be shared between firms and workers. Differences in workers' bargaining power across firms determine the sharing of these rents, leading to wage differences. Second, market frictions distort firms’ marginal costs, generating dispersion in marginal revenue products. As a result, similar workers will receive different wages due to the distorted performance of firms. Moreover, labor markets adjust through wages as they cannot fully adjust through quantities. Hence, market distortions impact job flows and wage inequality by affecting firm dynamics. To empirically assess the role of firm dynamics in aggregate TFP, job flows, and wage inequality in Ecuador, this paper utilizes a comprehensive employer-employee matched panel dataset constructed from two firm- level datasets and one worker-level database. The firm-level datasets provide variables, including firms’ value-added, capital, labor, and materials, which are relevant for estimating firm-level (revenue) productivity using the control function approach to correct for endogeneity (Ackerberg et al. [2015]). The employee database contributes information on wages and human capital characteristics, essential for identifying the determinants that drive aggregate wage inequality. This paper first documents firm-level patterns, which are indicative of underlying market distortions. As is standard in the literature, the paper finds an increasing relationship between different measures of productivity (i.e., revenue TFP, labor productivity, and sales per worker) and size. However, the paper also finds that older firms are not necessarily more productive than younger ones, signaling constraints on firm- level productivity growth. Comparing the productivity of surviving firms relative to exiters, the former group is, on average, more productive than the latter. Still, a substantial overlap in their productivity distributions implies inefficiencies in exit dynamics (Levy, 2018). Similarly, incumbent firms exhibit higher productivity levels than entrants, but again, an overlap in their productivity distributions suggests inefficient entry patterns. Furthermore, examining firms' life cycle patterns reveals additional evidence of market distortions. The median firm in Ecuador experiences initial modest growth in technical efficiency during the early years of operation but encounters stagnation and decline in later stages, underscoring the inability of firms to catalyze and sustain productivity growth as they age. At an aggregate level, the paper documents a decline in aggregate TFP over the sample period. Employing the Melitz and Polanec (2015) approach to decompose aggregate TFP growth, the paper identifies the allocative efficiency channel as the component driving aggregate TFP performance, stemming from persistent factor misallocation across firms. Sectoral factor reallocation is also inefficient, especially during economic downturns. Following adverse external shocks, the combined effect of market distortions with broader macro-structural challenges proves notably detrimental, as it amplifies these economic downturns by reallocating resources towards less productive firms (allocative efficiency channel) and sectors (structural transformation channel), rather than serving as cleansing mechanisms that shift resources more efficiently. Technical efficiency growth in Ecuador has not been able to offset the negative impact of the allocation channels on aggregate TFP trends. The modest contributions of the technical efficiency channel to aggregate TFP growth align with the poor life cycle patterns observed in firm-level productivity growth. Last, the market selection channel (entry plus exit) contributes marginally to aggregate productivity changes due to the entrance of many unproductive firms and the low survival of high-productivity entities. The paper finds that the poor contributions of the various productivity growth channels are linked to persisting distortions in factor and product markets. To formalize the relationship between allocative efficiency and the aforementioned market distortions, the paper conducts firm- and industry-level regressions, examining various measures of policy-related distortions. At the firm level, distortions in labor and capital markets are studied. The fluidity of Ecuador's labor market is hindered by regulatory frameworks that impose distortionary regulations, leading to escalated labor costs for firms. In comparison to other South American countries, Ecuador has higher firing costs. Also, its labor laws impose additional 3 employment costs on firms, such as compulsory worker participation in firm dividends. As a result, larger and more productive firms face higher non-wage labor costs, leading to an inefficient allocation of labor towards less productive enterprises. The paper finds that firms in the top productivity quintile have non- wage labor costs per worker that are 3.7 times larger than firms in the bottom productivity quintile, after controlling for firm size and age. Credit access patterns indicate misallocation of capital in Ecuador. Banking credit is inefficiently allocated among larger firms, as those with access to credit from banks are 3 percent less productive than their credit-constrained large counterparts, although this negative differential is not significant. Conditional on having access to credit, smaller firms are more productive than larger firms, contrasting the productivity-size relationship of the entire firm sample. Micro enterprises with access to credit are 20 percent more productive than large firms with credit access. These findings suggest an inefficient allocation of capital between firm size groups, with credit being allocated to both productive and unproductive large firms, while productive small firms face limited access to capital due to higher interest rates and collateral constraints. These constraints are also less binding for older, unproductive firms, leading to inefficient hoarding of capital. In addition to generating resource misallocation, these labor and credit market distortions dampen firm productivity growth as fewer resources are available for innovation. Furthermore, these distortions in factor markets, combined with weak institutional frameworks, such as inadequate insolvency regimes, create disincentives for low-productivity firms to exit while promoting the entry of poor performing new firms. At the industry level, the paper finds that sectors exhibiting higher job mobility, lower non-wage labor costs, less exposure to the minimum wage policy, improved credit access, fewer zombie firms, and greater market competition exhibit higher levels of allocative efficiency. Beyond yielding underwhelming aggregate TFP outcomes, market distortions have also generated firm distributional patterns and job flow trends that further underscore Ecuador's subpar economic performance. Large firms (those with more than 100 workers) represent less than 0.5 percent of formal firms but employ over 40 percent of the workforce. In contrast, micro firms (those with fewer than five workers) constitute over 90 percent of firms and contribute around 30 percent of total employment. Compared to the United States, Ecuador exhibits a substantially lower share of large firms and a higher share of micro firms, a characteristic pattern in developing economies attributed to market inefficiencies (Hsieh and Olken, 2014; Krueger, 2013; Banerjee and Duflo, 2005, 2011; among others). Moreover, there has been a considerable shift in economic activity from younger to older firms during the sample period. While this shift could be positive for aggregate TFP if firms experienced productivity growth as they aged, this is not the case in Ecuador. Market frictions have deterred the entry of productive firms and hindered the exit of unproductive older enterprises, resulting in an aging population of firms without accompanying firm productivity growth. These market frictions also drive aggregate job flow trends, with small, unproductive entrants creating most new jobs, while existing jobs are destroyed by old, unproductive firms that persist in the market. The underlying structural and policy drivers that influence Ecuador’s poor firm-level outcomes and resource misallocation also play a crucial role in explaining aggregate wage inequality. In Ecuador, variations in aggregate wage inequality are primarily driven by differences in wages across firms for similar workers (between-firm wage inequality) rather than differences in wages within firms for different workers (within-firm wage inequality). Between-firm wage inequality primarily stems from misallocation-related rents generated by policy-driven distortions (i.e., non-wage labor costs, firing costs, and institutional constraints on labor mobility). Specifically, distortionary policies and constraints affecting firms’ performance potentially account for up to 41 percent of aggregate wage inequality in Ecuador. Additionally, estimates of the passthrough of revenue productivity to wages suggest that formal workers have low bargaining power and that the distortions leading to labor misallocation are also associated with wage inequality. As explained by Criscuolo et al. (2020, 2021, 2022), higher levels of pass-through are an indication of both the higher bargaining power of workers and resource misallocation at the firm level. In Ecuador, a 10 percent increase in revenue productivity yields a 0.8 percent increase in wages, which is lower than most OECD countries (Criscuolo, 2021). This can be attributed to the limited availability of formal sector jobs, which reduces workers' bargaining power. Moreover, this paper finds that non-wage labor costs and product market competition not only contribute to misallocation and aggregate productivity 4 losses but also to higher wage inequality. In contrast, minimum wage policies reduce wage inequality in Ecuador, as observed in other countries, while also generating allocative inefficiencies. This paper contributes to three strands of literature. First, it adds to the work on aggregate TFP and its relation to firm outcomes and dynamics. Seminal work by Restuccia and Rogerson (2008, 2013), Hsieh and Klenow (2009), Bartelsman et al. (2013), among others, have attributed differences in aggregate TFP to resource misallocation across firms. Papers including Hsieh and Klenow (2009,2014), Bartelsman et al. (2013), Midrigan and Xu (2014), Gopinath et al. (2017), and Restuccia and Rogerson (2017) have connected resource misallocation to specific market distortions constraining firm dynamics. Additionally, studies by Krueger (2013) and Hsieh and Olken (2014) have highlighted that these distortions not only deplete aggregate TFP but also yield anemic firm distributional patterns. This paper provides additional empirical evidence on drivers of resource misallocation and their contribution to poor aggregate TFP outcomes and lackluster firm distributional patterns in the case of Ecuador. Second, this paper contributes to the literature on firm dynamics and job flows. Papers, such as Haltiwanger et al. (2013, 2016) and Decker et al. (2014), have underscored the significance of young firms in accounting for most of the job flows and employment reallocation in the U.S. Their work complements Dunne et al. (1989), Davis and Haltiwanger (1990, 1992), Davis et al. (1996), Foster et al. (2001), and Becker et al. (2007), which find that employment reallocation is accompanied by output growth and capital reallocation. This paper finds that, in the context of a developing economy characterized by highly distorted markets, job creation by entrants is inefficient, and job destruction is largely explained by unproductive, older firms that continue to operate. Lastly, this paper contributes to the literature on the pivotal role of firm outcomes in explaining wage inequality. Papers such as Faggio et al. (2010), Barth et al. (2016), Berlingieri et al. (2017), Manning (2021), and Criscuolo (2020, 2021, 2022), among others, have documented that firms are important for explaining wage variation among workers. Furthermore, Criscuolo (2020, 2021, 2022) have studied potential drivers of misallocation that are linked to wage inequality outcomes for OECD countries. This paper builds upon this work by simultaneously linking drivers of misallocation to wage inequality in a developing economy, underscoring the relevance of non-wage labor costs and product market distortions for wage dispersion. The paper is organized as follows: Section 2 describes the dataset leveraged for the different empirical analyses. Section 3 provides aggregate motivating facts about aggregate productivity and inequality trends in Ecuador. Section 4 documents productivity, market selection, and life cycle patterns of Ecuadorian firms. Section 5 studies the main channels and drivers of aggregate productivity growth in Ecuador. Sections 6 and 7 characterize firm distributional and job flow patterns, respectively. Section 8 assesses the role of firm dynamics on wage inequality in Ecuador. Last, Section 9 concludes. 2 Data This paper utilizes two firm-level datasets and one employee-level database to construct a comprehensive employer-employee matched dataset, facilitating a simultaneous examination of firm and labor outcomes. The first source is the Directory of Companies and Establishments (Directorio de Empresas y Establecimientos [DIEE]), a public database created by the Ecuadorian Statistics Institute (Instituto Nacional de Estadística y Censos [INEC]), which incorporates firm administrative data from the national revenue service (Servicio de Rentas Internas [SRI]) and firms’ employment data from the social security institution (Instituto Ecuatoriano de Seguridad Social [IESS]). The DIEE covers information for all formal firms registered with the SRI since 2012, totaling around 800,000 firms. However, it has limitations, providing only limited firm performance variables (sales and employment) and lacking data necessary for estimating value added and productivity, such as firms' costs. The DIEE data is complemented with non-public firm administrative data obtained from the SRI, which encompasses the entire universe of formal firms in Ecuador. This second source includes key variables for estimating firm TFP, such as sales, gross production, costs, fixed assets, investment, employment, and materials at the firm level. The dataset also comprises relevant firm characteristic variables, including economic sector of activity, age, and geographical location, among others. The third source, the Labor and 5 Business Dynamics Laboratory (Laboratorio de Dinámica Laboral y Empresarial [LDLE]), is constructed by INEC using information from IESS, SRI, and other government institutions. It contains worker-level data for over 70,000 firms that meet two criteria: reported sales to SRI and registered employment with IESS (formal employment). Combining these three data sources yields a rich employer-employee database useful for analyzing firm dynamics, aggregate productivity, and labor demand patterns for the 2012–2020 period.1 The main firm performance variable of interest in this paper is firm revenue based TFP. The control function approach developed by Ackerberg et al. (2015) is used to estimate revenue TFP. The variables necessary for estimating revenue TFP are value-added, capital, labor, and materials (which is used as the proxy variable for the control function approach) at the firm-level. The paper follows the methodology of Avellan and Ferro (2017) to construct these variables using the Ecuadorian firm administrative data, adhering to standard measurements in the literature. Value-added is calculated as firm gross output (sales adjusted for product inventories) minus intermediate consumption of materials and services. Intermediate consumption comprises operational expenditures and production costs. Capital is measured based on the value of fixed assets reported in firms' financial statements, lagged by one year. Labor is measured as the number of workers employed by a firm. Materials are computed as the sum of raw materials, transport costs, gasoline costs, and utilities costs. Monetary variables are transformed into constant prices using sector deflators from the Ecuadorian input-output matrix. Consequently, firm gross production, intermediate consumption, and value-added are deflated using their respective sector deflator indexes. Similarly, materials are deflated using the sector deflators for intermediate consumption, while capital is deflated using the sector deflators for gross fixed capital formation. Lastly, there are two significant firm characteristic variables relevant in the analyses of the paper: sector and age. The firm's sector of activity corresponds to the International Standard Industrial Classification (ISIC) sector classification. The age of a firm is calculated as the difference between the observation year and the year of the firm's commencement of operations. Variables pertaining to workers are measured in a standard manner to facilitate the wage inequality analyses. Worker-level information is available at a monthly frequency in the LDLE. Following relevant papers in the literature (Barth et. al., Criscuolo et al. [2020, 2021, 2022]), only workers that are fully employed for a calendar year are kept in the estimation sample. Wages are constructed as a worker’s daily wage rate from reported monthly wages and the number of days worked in a month. These are deflated by the consumer price index to obtain real wages. To match to the firm-level data, which is at an annual frequency, yearly real wage rates are estimated as the average across months of the real daily wage rate. The paper also utilizes the following variables associated with workers’ job characteristics and human capital: part-time employment, education, age, gender, and ethnicity. A dummy for part-time employment is constructed using the number of days worked in a month. Workers with less than 15 days of employment per month are classified as having part-time employment.2 For education, a bivariate variable is created to identify if a worker has obtained at least a high school diploma. Last, workers’ age, gender, and ethnicity are provided by the LDLE. 3 Motivation Facts: Aggregate Productivity and Inequality Trends in Ecuador A concern in Latin American economies is that growth has been unsatisfactory and driven by poor productivity performance, as documented by Busso et al. (2013). Ecuador is no exception to this trend. As depicted in Figure 1, between 2003 and 2014, Ecuador's GDP grew at an average annual rate of 4.7 percent, largely driven by the growth of physical capital stock. However, economic growth has been slightly 1 It is important to acknowledge certain limitations of the data sources employed in this paper. Firstly, the universe of firms and workers considered only pertains to formal economic activity. Additionally, due to variations in regulations governing the submission of tax returns by firms, many entities do not report information necessary for measuring relevant variables for the paper. Given this, the sample size can vary across the exercises carried out in this paper, as the number of observations is maximized in relation to the variable of interest in each exercise. To mitigate sample attrition, values are imputed for observations with missing data for the variables used in the productivity estimation method. 2 To transform number of days worked per month to an annual measure, the paper considers a worker's most frequently reported number of days worked per month within a year as their annualized measure. 6 negative since 2014, primarily due to an average annual decline of 3.2 percent in TFP. Aggregate TFP growth is directly linked to the dynamics of firms operating in the economy through three main channels: the reallocation of factor inputs across operating firms and sectors (i.e., allocative efficiency channel), changes in firm capabilities (i.e., technical efficiency channel), and market selection through the entry and exit of firms (i.e., market selection channel). First, aggregate TFP grows when there is a reallocation of factors from less productive firms or sectors to more productive ones. Second, improvements in firm capabilities through higher technology adoption, innovation, and managerial skills also lead to higher aggregate TFP, since these enhancements enable firms to generate greater output with the same level of capital and labor inputs. Lastly, the market entry of high-productivity enterprises or the exit of low- productivity firms also drive aggregate productivity growth because more productive firms would operate in the market, on average, and factor inputs would be allocated to these higher-performing entities. Therefore, studying the patterns of firm dynamics associated with these channels is crucial for identifying the potential drivers of Ecuador’s decline in aggregate productivity.3 Figure 1: Drivers of GDP Growth in Ecuador Source: Author’s Elaboration, WDI, Banco Central del Ecuador, INEC. Note: This graph decomposes GDP growth into four components: growth in capital stock, growth in labor, growth in human capital, and growth in TFP. The first two columns compare the growth decomposition before and after 2014. The third to fifth columns compare the growth decomposition for three set of years that coincide to the firm-level data leveraged in the paper. The impact of firm dynamics extends beyond aggregate productivity and can have pervasive effects on workers' earnings and income inequality. Recent studies by Criscuolo et al. (2020, 2021, 2022) and others have established a connection between poor aggregate productivity performance in OECD countries and the rise in income inequality. This research underscores the crucial role of market distortions that impede the efficient functioning of markets, including labor markets, in driving both aggregate productivity and income inequality outcomes. Figure 2 reveals correlated trends in aggregate productivity and inequality in Ecuador. The country experienced a period of significant aggregate productivity growth from 2003 to 2014, accompanied by a substantial decrease in income inequality measured by the Gini Index. However, after 2014, there was a sharp decline in aggregate productivity, while income inequality’s previous declining trend plateaued until 2018 and began to slowly rise thereafter until 2020. By 2020, Ecuador had the third- highest Gini coefficient in South America, surpassed only by Colombia and Brazil. This suggests that firm dynamics in Ecuador may also play an important role in explaining income inequality patterns, in addition to aggregate productivity performance. 3 For a further discussion on the channels and drivers of aggregate productivity in emerging economies, please refer to Cusolito and Maloney (2018). 7 Figure 2: Aggregate Productivity vs. Aggregate Inequality Trends in Ecuador Source: Author’s Elaboration, WDI, Banco Central del Ecuador, INEC. Note: This figure reports the trend of aggregate productivity (TFP) in Ecuador and the Gini Coefficient, as well as the 2-period moving average for both variables. 4 Firm Dynamics in Ecuador4 To study firm-level productivity dynamics, this paper assumes that firms possess a standard Cobb-Douglas production technology in which capital, , , and labor, , , inputs are transformed into a final output good, , : , = , + , + , + , , (1) where i, , and denote firms, sectors, and years, respectively. Equation (1) presents the log transformation of firms’ technology, where lowercase variables denote natural logs. The term ω , is the revenue TFP of firm , which captures how good firm is at converting capital and labor inputs into final output (i.e., firms’ technical efficiency).5 Any idiosyncratic shock to production is captured by , . Parameters and are the output elasticities of labor and capital, respectively.6 To estimate firm revenue TFP and the output elasticities of factor inputs, this paper leverages the control function estimation approach developed by Ackerberg et al. (2015).7 In the following sections, revenue TFP is studied as the primary variable of firm- level performance. Labor productivity (value-added per worker) and sales per worker are also examined. 4 The remainder of the paper analyzes the impact of firm dynamics on aggregate productivity, labor demand, and wage inequality for only formal firms, given the unavailability of firm-level data for the informal sector. Nevertheless, the findings for formal firms hold significant implications for the informal sector and the broader economy. This is because the formal sector accounts for most of the economic activity in terms of production. According to recent studies for Ecuador (Mejía et al. [2023], Ruesga et al. [2020], Aguilar & Sarmiento [2012], and López et al. [2003]), the informal sector contributes with between 20 and 40 percent of total production. Although this contribution is considerable, the informal sector accounts for an even larger share of employment, at around 55 percent. However, this merely underscores the lower productivity of the informal sector compared to the formal one, as it involves more workers generating relatively less output. Hence, access to data on informal sector firms would likely bolster any negative findings concerning aggregate economic performance. 5 While firm revenue TFP has been widely employed as a proxy for firms’ technical efficiency (i.e., quantity-based TFP), it is important to acknowledge its limitations in doing so. As explained in Section 2, all variables used in estimating revenue productivity are converted to physical quantities using industry-level price indexes. However, since the data utilized in this study does not include firm-level prices, it is not possible to disentangle quantity-based TFP from firm prices even after deflating by industry-level price indexes. Consequently, estimates of revenue productivity also reflect residual firm-level price effects, which could account for various factors such as market distortions manifested in heterogeneous firm input costs, variations in product quality among firms, and market power considerations (Cusolito and Maloney, 2018). Furthermore, like technical efficiency, these factors may be correlated with policy changes, which could lead to inaccurate policy prescriptions. Due to the impossibility of directly estimating quantity-based TFP, revenue TFP is used as a proxy for technical efficiency. Foster et al. (2008), Eslava and Haltiwanger (2016), and Iacovone and Patiño Peña (2021) report correlations between quantity-based and revenue-based TFP of 0.75, 0.69, and 0.75, respectively, indicating that despite contamination, revenue TFP exhibits a high correlation with technical efficiency. 6 The output elasticities of labor and capital are indexed by sector , which assumes that firms’ technology is the same within sectors but differs across sectors. 7 Estimating output elasticities and firm revenue productivity using an ordinary least squares (OLS) regression of output, , on observable capital, , , and labor, , , is problematic as the unobservable revenue productivity of the firm, ω , , is correlated with the demand of the flexible labor 8 4.1 Relationship of Firm Productivity with Firm Size and Firm Age The top three panels of Figure 3 illustrate the revenue TFP, labor productivity (value-added per worker), and sales-per-worker premiums for different firm size groups compared to micro enterprises. 8 These productivity premiums increase with firm size, indicating that larger firms are more productive. Allocating more capital and labor to larger firms could enhance aggregate productivity, as these resources would be utilized by more productive firms. The findings of Figure 3 also align with the aggregate results of Figure 1, showing that economic performance, at the aggregate and micro levels, is associated with capital accumulation rather than TFP. This is evident when comparing the size premiums of labor productivity and revenue TFP.9 Large firms have labor productivity premiums that are twice as high as those of medium and small size groups, while their revenue TFP premium is only about 20 percent higher. Hence, as firms increase in size, their revenue generation relies more on factor accumulation than technical efficiency. Additionally, this suggests potential constraints on firm-level productivity growth as they age. The bottom panels of Figure 3 support this, as firms aged between 5 and 14 years exhibit small revenue TFP premiums compared to younger firms, which eventually dissipate for firms aged 15 years or older.10 Figure 3: Firm Productivity Premiums by Size and Age Source: Author’s Elaboration, INEC. Note: The top three panels of this figure report the revenue productivity, sales, and value-added premiums of firm size for the 2012 – 2020 period. The control group is the micro firms size group. The bottom three panels of this figure report the revenue productivity, sales, and value-added premiums of firm age for the 2012 – 2020 period. The control group is the less than five years old age group. Regressions control for industry and year fixed effects. input. That is, firms choose labor, , , in period after they observe their revenue productivity shock, ω . The same is not the case for capital, , , as its level in period is chosen in period − 1. The control function approach is the most widely used production function estimation method that corrects for this endogeneity issue. This approach was first developed by Olley and Pakes (1996), and has been improved upon by other seminal papers in the literature, such as Levinsohn and Petrin (2003), Wooldridge (2009), and Ackerberg et al. (2015). The methodology developed by Ackerberg et al. allows to correct for endogeneity in the estimation of productivity, using materials as an instrumental variable, as well as correcting for entry and exit. 8 This paper classifies firms into 4 size categories following the size groupings of the World Bank Enterprise Survey: 1) micro firms have less than 5 workers, including self-employed workers, 2) small firms have between 5 and 19 workers, 3) medium-sized firms have between 20 and 99 workers, and 4) large firms have more than 100 workers. 9 Labor productivity can be decomposed into capital deepening (capital-labor ratio) and revenue TFP, to determine whether labor productivity levels or its changes are explained by the accumulation of factors (capital deepening) or technical efficiency. A similar decomposition can be made for sales per worker, by also including materials as another component in the decomposition in addition to capital deepening and revenue TFP. 10 This paper classifies firms into 5 age categories: 1) less than five years of age, 2) between five and nine years of age, 3) between 10 and 14 years of age, 4) between 15 and 19 years of age, and 5) 20 or more years of age. 9 4.2 Market Selection Patterns of Firms Standard theory of firm dynamics (such as Hopenhayn [1992] and Jovanovic [1982]) suggests that, in markets without imperfections, survivor firms are more productive than exiters since efficient market selection results in the exit of firms with productivity levels that are too low to continue operating. Thus, efficient market selection yields that the productivity distribution of survivor firms would be significantly different than the exiters’ productivity distribution, as the former’s distribution would be shifted to the right of the latter’s. Figure 4 and Figure A 1 illustrate that the revenue TFP distributions of survivor firms are slightly shifted to the right of exiter firms, indicating that, on average, survivors are more productive than exiters.11 However, there is considerable overlap between the two distributions, suggesting that market selection is inefficient in eliminating unproductive firms and preventing productive ones from exiting. 12 Moreover, the productivity gap between survivor and exiter firms has diminished over time due to two distributional shifts. First, the revenue productivity distribution of exiter firms became more left-skewed, suggesting a higher concentration of high productivity exiters. Conversely, the distribution of survivor firms became less left-skewed, indicating a lower concentration of higher productivity firms operating in the market. Figure 4: Revenue TFP Distributions: Exiters vs. Survivors Source: Authors’ Elaboration, INEC. Note: This figure studies the distributions of revenue productivity for the 2012 – 2020 period. In this figure, the distributions of survivors and exiters for 2013, 2016, and 2019 are plotted to analyze changes across time for these distributions. Given that exiters are defined with respect to period + 1, the distribution for 2020 cannot be constructed, as the firms that exited between 2020 and 2021 are not yet observed. According to Bartelsman and Doms (2000) and Foster et al. (2001), efficient markets generally exhibit higher average productivity among incumbents due to their market experience. While there is significant overlap between the distributions of incumbents and entrants, Figure 5 and Figure A 2 illustrate that incumbent firms have a rightward shift in their revenue productivity distribution compared to entrants. 13 However, over time, this shift has decreased, indicating a rise in the number of high-productivity entrants. Notably, during the Covid-19 pandemic (2020), the distribution of entrants shows a bimodal pattern, suggesting higher entry of both unproductive entrepreneurs seeking opportunities amid reduced labor market prospects and productive entrepreneurs that have benefited from new business opportunities and 11 In this analysis, a firm is defined as an exiter if it is present in the sample of formal firms in year but not in year + 1, while survivor firms are present in both years. 12 Levy (2018) also finds that there is significant overlap in the distributions of survivor and exiter firms and associates this pattern to existence of distortionary market frictions. 13 In this analysis, entrant firms are defined as those not present in the sample of formal firms in year t-1 but present in year t, while incumbent firms are present in both years. 10 improved productive processes that enhance their business models (i.e., online delivery, digital payment systems). Figure 5: Revenue TFP Distributions: Entrants vs. Incumbents Source: Authors’ Elaboration, INEC. Note: This figure studies the distributions of revenue productivity for the 2012 – 2020 period. In this figure, the distributions of incumbents and entrants are plotted for 2013, 2016, 2019, and 2020 to analyze changes across time for these distributions. 4.3 Life Cycle Trends of Firms The top panels of Figure 6 illustrate the life cycle patterns of revenue TFP and labor productivity for the average, median, P10, and P90 firms in Ecuador. TFP for the median and mean firms shows an increase between the ages of five and nine years compared to the first age group (less than five years old), but beyond that, firm TFP growth becomes negative. Notably, firms aged 20 years or older exhibit lower TFP than those in their early operating years. These findings complement those of Figure 3, which indicate a decrease in revenue TFP premiums for older age groups of more than nine years and even negative premiums for firms aged 20 years or more (relative to firms less than five years old). These trends imply the presence of barriers that impede firm TFP growth in Ecuador. Hsieh and Klenow (2014) find that firms in emerging economies, such as China, Mexico, and India, experience lower rates of TFP growth across the life cycle compared to firms in developed countries due to reduced incentives for process improvement, product quality enhancement, and expansion into foreign markets. The productivity life cycle trends in Ecuador are even more concerning, as TFP growth ceases after nine years of operation and declines in later stages. Although the life cycle trends of labor productivity are not as pessimistic as those of TFP, labor productivity stagnates during the middle years of the life cycle and declines for firms aged 20 years or more.14 While the P10 firm demonstrates TFP growth throughout its life cycle until age 20, the P90 firm experiences declines in TFP at every stage. Therefore, firms on the left side of the productivity distribution upon entry achieve catch-up in terms of technical efficiency. However, this catch-up is also explained by high- productivity firms becoming less and less productive. This suggests that frictions hindering TFP growth disproportionately affect initially high-productivity firms in Ecuador, imposing greater constraints on their adoption of advanced or innovative technologies. The trends observed in labor productivity for the P10 and P90 firms align with the patterns in TFP, further emphasizing the differential impact of market frictions on productivity growth, particularly for firms at the higher end of the productivity distribution. 14 The top panels of Figure A 3 demonstrate that the life cycle trends of revenue TFP and labor productivity for the sample of surviving firms mirror those of the entire sample, indicating that the patterns observed in Figure 6 are not driven by inefficient market selection. 11 Figure 6: Life Cycle Trends of Firms Source: INEC. Authors’ Elaboration. Note: This figure studies the life cycle patterns of firms in Ecuador for the 2012 – 2020 period. To plot the graphs in this figure, observations are pooled across all years, and then the P10, median, mean, and P90 are identified for each variable (revenue TFP, labor productivity, sales, capital, employment, and wage bill) for five age groups in our sample. The five age groups are: 1) less than five years of age, 2) between five and nine years of age, 3) between 10 and 14 years of age, 4) between 15 and 19 years of age, and 5) 20 or more years of age. The middle and bottom panels of Figure 6 present the life cycle patterns of sales, capital, employment, and wage bill for formal firms in Ecuador. The median and mean firms exhibit significant sales growth in the first five years of operation, followed by continued but slower growth in subsequent years. In contrast to TFP patterns, higher sales are consistently associated with older firms. The life cycle trends for employment and capital display similar patterns for the median and mean firms, indicating a continuous accumulation of production factors as firms age. These trends indicate dynamic patterns of misallocation. In efficient factor input markets, the stagnating and then declining trend of firm productivity across the life cycle would typically be accompanied by a similar trend in capital and labor as firms age. However, in Ecuador, the 12 opposite occurs, suggesting that factor input growth is driven by market imperfections that encourage inefficient factor hoarding and hinder the efficient reallocation of resources from older, less productive firms to younger, more productive ones. This further implies that labor productivity growth stems from factor accumulation (i.e., capital deepening) rather than improvements in firm capabilities. 5 The Role of Firm Dynamics in Aggregate Productivity 5.1 Decomposition of Aggregate Productivity Growth This section studies the channels of Ecuadorian aggregate productivity growth between 2012 and 2020. 15 For this, aggregate productivity growth is decomposed into five components, as in Melitz and Polanec (2015): the within, between, entry, exit, and structural transformation components: ∆Ω = , ∆ + , ∆ + , , + , , + ∆, Ω, . (2) , , This decomposition is valuable for identifying aggregate productivity growth drivers. For instance, if aggregate productivity growth is driven by the within component, it suggests that surviving firms have improved their technical capabilities over time through technological adoption, productive investments, or enhanced management practices. Growth driven by the between component indicates more efficient allocation of capital and labor in factor markets, potentially due to reduced market frictions or less distortive regulations. Aggregate productivity growth attributed to the entry and exit components suggests favorable market conditions that enable the entry of more productive new firms and the exit of less productive establishments. Lastly, aggregate productivity growth resulting from the structural transformation component indicates improved macroeconomic conditions, leading to the shift of economic activity from less productive sectors to more productive ones. Appendix 1 provides a further discussion on the derivation of this decomposition. Figure 7 shows that, between 2012 and 2020, the year-to-year growth of aggregate TFP in Ecuador has been consistently poor. Only one year, 2017, exhibited a positive growth of 0.3 percent in aggregate productivity compared to the previous year. Conversely, the years 2015 (-11.8 percent), 2018 (-10.8 percent), and 2020 (-6.9 percent) witnessed the largest drops in aggregate productivity. As a result, by 2020, aggregate productivity had declined by nearly 33 percent compared to 2012. To account for the volatility in the performance of the oil industries, characterized by significant shifts in prices and quantities, the aggregate productivity growth decomposition analysis is carried out for non-oil economic formal sector activities. 16 Figure 7 also illustrates the decomposition of aggregate productivity growth according to Equation (2). The importance of the between component in driving the poor aggregate productivity growth outcomes suggests that market frictions have pervasive effects, hindering the efficient flow of factor inputs towards the most productive firms. The structural transformation component follows the between component, particularly during years with significant productivity losses. Despite the positive contribution of the within component, which reflects slightly higher firm technical efficiency, it has been unable to offset the negative trends caused by inefficient reallocation at the firm and sector levels. Lastly, the entry and exit components have 15 The measure of aggregate productivity used in this section corresponds only to formal sector activity, as the data sources used only consider formal firms. 16 All disaggregate 4-digit economic activities within the 1-digit industries: “O – Public Administration and Defense”, “P – Education”, and “Q – Human Health and Social Work Activities” are excluded from the aggregate productivity growth decomposition exercise, as these sectors are considered to be non-commercial. The Oil economic activity is comprised of the following disaggregate 4-digit economic activities: “B0610 – Extraction of crude petroleum”, “B0620 – Extraction of natural gas”, “B0910 – Support activities for petroleum and natural gas extraction”, and “C1920 – Manufacture of refined petroleum products”. The remaining industries make up the non-oil economic activity. 13 negligible effects on aggregate TFP growth, indicating that existing frictions impede productivity- enhancing market selection. Figure 7: Aggregate Productivity Growth Decomposition Source: INEC. Authors’ Elaboration. Note: This figure studies the decomposition of yearly aggregate productivity growth in non-oil activity in Ecuador for the 2012 – 2020 period. Aggregate productivity growth is decomposed into five components using the Melitz-Polanec Decomposition: the within component, the between component, the entry component, the exit component, and the structural transformation component. Ecuador's negative contributions of the between component suggest the presence of market distortions affecting the allocation of resources across firms. Notably, significant drops in allocative efficiency occurred in 2013, 2015, and 2018, indicating that the impact of the between component on aggregate productivity remained negative and persistent irrespective of the economic cycle.17 This contrasts with the US, where the between component remained steady and positive even during economic fluctuations, as documented by Haltiwanger et al. (2016). In well-functioning product and factor markets, the allocative efficiency channel serves as a buffer against adverse economic shocks, mitigating their impact on aggregate productivity by facilitating the flow of resources toward more productive firms, as demonstrated in the US context. However, in Ecuador, the allocative efficiency channel exacerbates the propagation of recessions, intensifying downturns by further depressing aggregate TFP. Furthermore, in the aftermath of negative external shocks, such as the oil price reductions in 2014 and 2018 and the COVID-19 pandemic in 2020, the combined effect of the misallocation of factors across firms (i.e., the allocative efficiency channel) and the negative reallocation of factors across sectors (i.e., the structural transformation channel) magnified the economy's inability to naturally shift capital and labor to more productive firms and sectors.18 This interplay underscores the prevalence of factor and product market distortions as well as macro-structural frictions within the Ecuadorian context. The within-firm component of the aggregate TFP growth decomposition in Ecuador exhibited a positive year-to-year trend, but its magnitude remained small, and its impact was offset by the between component in all years. These modest contributions of the within component suggest a gradual increase in aggregate 17 Between 2012 and 2020, 2013 and 2014 had high economic growth rates. In contrast, during this sample period, 2016 and 2020 experience recessions, while in 2015 and 2019 GDP growth rates were negligible. 18 This paper focuses on examining the impact of firm-level dynamics on aggregate productivity outcomes. Therefore, this section primarily assesses patterns of firm misallocation within sectors and does not delve into a comprehensive discussion of the reallocation of factors across sectors (i.e., the structural transformation component). Nonetheless, it is important to note that the behavior of the structural transformation component reveals that, in most years, the reallocation of economic activity across sectors contributed to a reduction in aggregate productivity of non-oil activities (Figure 7). Hence, during the analyzed period, there was a shift of economic activity from more productive industries to less productive ones. It is worth highlighting that the years exhibiting the most significant negative contributions in this component coincided with substantial drops in oil prices compared to the previous year (2014 and 2018). This indicates that Ecuador's susceptibility to negative oil price shocks leads to an inefficient adjustment of economic activity driven by structural conditions inherent to the economy, such as low labor market mobility and the rigidity of the nominal exchange rate (due to Ecuador’s dollarization). Furthermore, the years characterized by positive structural transformation also witnessed a negative reallocation of factors across firms within sectors (i.e., the allocative efficiency channel), which more than offset the positive shifts in economic activity at the sectoral level. 14 technical efficiency for operating firms over the analyzed period. However, the bottom left panel of Figure 3 indicates that firm productivity growth in Ecuador is likely constrained, suggesting that growth in aggregate technical efficiency is primarily driven by spillovers of international technology trends rather than firm capability upgrading. Figure 7 indicates that firm exit in Ecuador was slightly “aggregate productivity enhancing” between 2012 and 2016.19 However, as shown in Figure 4, despite survivors having higher average productivity than exiters, the substantial overlap in their revenue TFP distributions suggests that aggregate productivity growth could have been higher if market selection had been efficient. From 2016 onwards, the contribution of the exit component to aggregate productivity growth became negligible (and sometimes negative) due to the reduction in the productivity gap between survivors and exiters. This was driven by an increase in high-productivity exiters and a subsequent decrease in high-productivity survivors. These trends align with the slowdown experienced by Ecuador since 2014, where the struggling economy led to higher exit probabilities for more productive firms, resulting in their increased exit over time. Furthermore, this persistent and increasing exit of high-productivity firms negatively impacted aggregate productivity through two channels. Firstly, there was a decline in the overall participation of high-productivity firms in the market, leading to a contraction in aggregate TFP through lower technical efficiency. Secondly, the exit of more productive firms resulted in a reduced allocation of resources to higher productivity enterprises, leading to lower allocative efficiency. Moreover, the inefficient exit of more productive firms was costlier as these establishments typically employed more workers and utilized more capital, resulting in larger amounts of capital and labor being destroyed upon exit.20 Between 2012 and 2020, the distribution of entrants became more left-skewed, and the distribution of incumbents became slightly less left-skewed, indicating an increase in high-productivity entrants and a decline in high productivity incumbents (Figure 5). At the aggregate level, these shifts in distributions toward more productive entrants are reflected in mostly positive contributions of the entry component to aggregate productivity growth (Figure 7).21,22 However, the persistent prevalence of lower TFP entrants, relative to incumbents, accounts for the modest (and sometimes negative) contribution of this channel to aggregate productivity growth. This latter trend would be less detrimental to aggregate productivity growth if entrant firms demonstrated significant potential for enhancing their capabilities. Foster et al. (2001) find that US entrants also exhibit lower productivity compared to incumbents but possess high potential for technical efficiency growth. In contrast, the invariant distribution of incumbents in Ecuador ( Figure 5), the stagnation and later decline of firm productivity growth over the life cycle, and the poorer growth outcomes of more productive entrants (Figure 6) align with the presence of barriers that impede firm productivity growth. Furthermore, the entry of these less productive and growth-constrained firms exacerbates allocative inefficiencies, as lower performing firms utilize factor inputs. 19 Levy (2018) defines “aggregate productivity enhancing exit” as the exit of low-productivity firms from the market, which is be expected to occur in markets that function efficiently. “Aggregate productivity reducing exit” occurs when high-productivity firms exit the market, characterizing the presence of market frictions. 20 These trends are only for the sample of formal firms in Ecuador. Informal firms are generally less productive and smaller. Hence, the revenue productivity distributions including formal and informal enterprises would be expected to have a higher concentration of lower productivity firms for both survivors and exiters. However, the reduction in the productivity enhancement magnitude would be expected to persist, since exit trends have increased towards more productive enterprises that are generally formal. 21 Levy (2018) defines entry as “aggregate productivity enhancing” if higher productivity firms enter the market. “Aggregate productivity reducing” entry occurs when lower productivity enterprises enter the market. 22 The distributional trends of entrants and incumbents in Figure 5 only characterize formal firm distributions. One characteristic of markets in developing economies like Ecuador is that there is constant entry of informal subsistence entrepreneurs, who open businesses, not because of high and innovative entrepreneurial abilities, but because of the lack of formal employment opportunities. Hence, for the period analyzed, the “aggregate productivity enhancing” entry of formal firms may have been ameliorated by the entry of low productivity informal entrants, even more so during recessions, as levels of informality tend to increase in economic busts. This hypothesis is also supported by the productivity distribution of entrants in 2020 (Figure 5). Although the distribution is only for formal firms, the large left hump suggests the potential entry of unproductive entrepreneurs because of the negative economic context in the Ecuadorian labor market. 15 5.2 Drivers of Aggregate Productivity Growth 5.2.1 Labor Regulation Aggregate productivity growth in Ecuador can be partially attributed to the distortionary impact of labor market policy. Labor regulations in Ecuador impose high mandated labor costs on firms. Ecuadorian firms’ labor costs encompass not only salaries but also various other expenses, such as mandatory employment allowances 23 , compulsory worker participation in firm dividends 24 , additional obligatory services like childcare for larger firms, and high firing compensation packages25. These mandated labor costs are an excessive burden for firms. For example, severance pay and firing costs in Ecuador are higher relative to most of its South American peers, as illustrated in Figure A 4.26 Additionally, as estimated by Grosh et al. (2014), mandatory employment allowances can increase costs by up to 25 percent for firms. Minimum wage policy also places a burden on firms, with Ecuador's minimum wage-to-GDP per capita ratio higher than most South American countries (Figure A 5).27 Hence, the combination of these distortionary labor regulations results in a highly inflexible labor market and significantly impacts allocative efficiency. Differences in labor costs by firm size, age, and productivity indicate why the contributions of the between component negatively impact aggregate productivity growth. The top-left panel of Figure 8 illustrates that larger firms report higher total labor costs per worker.28 The top right panel of Figure 8 indicates that this relationship persists even when examining firms’ labor costs per worker net of wages, i.e., non-wage labor costs per worker.29 After controlling for firm age and revenue productivity, large firms report total labor costs per worker and non-wage labor costs per worker that are 1.6 and 1.5 times higher than micro firms, respectively. Non-wage labor costs are expenses associated with Ecuadorian regulations that impose additional costs on firms, such as compulsory allowances and workers’ participation in firm dividends. One plausible explanation for these findings is that large firms are more likely to be under scrutiny by the government to comply with the costs arising from these labor regulations. Additionally, certain obligatory services are tied to workforce size, such as the requirement for firms with 50 or more workers to provide child daycare services. The middle panels of Figure 8 further reveal that total and non-wage labor costs per worker increase as firms age. In particular, older firms have total labor costs per worker that are 1.2 times higher than micro enterprises and non-wage labor costs per worker that are 1.1 times higher. This can be linked to distortionary labor frictions that discourage employee churning, resulting in an aging labor force within firms and higher costs associated with longer job tenure, as compulsory allowances directly increase with salary increments over a worker’s job tenure. Also, the right panel of Figure A 4 illustrates that severance payments increase with job tenure, suggesting that older firms with older labor forces would face higher severance costs if they were to lay off workers. In turn, this encourages older firms to retain their labor force despite being less productive. Furthermore, the bottom panels of Figure 8 demonstrate that more productive firms face 23 Ecuadorian formal firms are legally obliged to pay three compulsory allowances for each worker. The “thirteenth salary” allowance is equal to a worker’s monthly salary and paid to all workers during Christmas time (December). The “fourteenth salary” allowance is equal to a minimum wage payment that is paid to all workers before the school year begins (August/September). Lastly, the “unemployment reserve funds” allowance is paid monthly to every worker and equals 8.3 percent of each worker’s salary. 24 According to Article 97 of Ecuador’s Employment Regulation, firms are obliged to share 15 percent of their profits with their labor force. 25 According to Article 188 of Ecuador’s Employment Regulation, a worker that is fired and has been employed less than three years will receive firing compensation equivalent to three months’ salary. A worker that is fired and has been employed more than three years will receive firing compensation equivalent to the number of years worked in the company times their monthly salary (up until 25 monthly salaries). 26 Bolivia and the República Bolivariana de Venezuela are excluded as data is not reported withing the World Bank’s Employing Workers Database. 27 Chile and the República Bolivariana de Venezuela are excluded as data on minimum wages is not available in the ILO’s Global Wage Report 2020 – 2021. 28 For this estimation, firms’ total labor costs consist of wages plus social security contributions, severance pay costs, mandatory compulsory allowances (i.e., “thirteenth” salary, “fourteenth” salaries, unemployment reserve fund), compulsory worker participation in firms’ dividends, and other remunerations (i.e., transportation, commissions, bonuses, childcare). 29 For this estimation, firms’ non-wage labor costs include mandatory compulsory allowances (i.e., “thirteenth” salary, “fourteenth” salaries, unemployment reserve fund), compulsory worker participation in firms’ dividends, and other remunerations (i.e., transportation, commissions, bonuses, childcare). It is important to caveat that it is not possible to include firing costs into the non-wage labor cost measure as the data sources used do not allow for this. 16 higher total and non-wage labor costs per worker. In particular, firms in the top productivity quintile report total labor costs per worker that are 1.9 times higher than the bottom productivity quintile, after controlling for firm size and age. More strikingly, non-wage labor costs per worker are 3.7 times higher in firms in the top productivity quintile relative to the bottom quintile. This indicates that labor market frictions imposed by labor regulations disproportionately impact productive firms, hindering their ability to demand more capital and labor, and, in turn, resulting in higher resource allocation in less productive firms. These negative misallocation patterns are manifested in the negative contributions of the between component. Figure 8: Firms’ Total and Non-Wage Labor Costs by Size, Age, and Productivity Levels Source: INEC. Authors’ Elaboration. Note: The left panels of this figure report the size, age, and revenue productivity premiums of total labor costs per worker. The right panels of this figure report the size, age, and revenue productivity premiums of non-wage labor costs per worker. These premiums are estimated from a regression of total or non-wage labor costs per worker on size groups, age groups, revenue productivity quintiles, 2-digit industry controls, and year controls for the 2012 – 2020 period. Hence, the size group premiums control for age and revenue productivity, the age group premiums control for size and revenue productivity, and the revenue productivity premiums control for size and age. The reference size group is “microenterprises”, the reference age group is “less than 5 years of age”, and the reference revenue productivity group is “revenue TFP Quintile 1”. Total labor costs consist of wages plus social security contributions, severance pay costs, mandatory compulsory allowances (i.e., “thirteenth” and “fourteenth” salaries), compulsory worker participation in firms’ dividends, and other remunerations (i.e., transportation, commissions, bonuses, childcare). Non-wage labor costs include mandatory compulsory allowances (i.e., “thirteenth” and “fourteenth” salaries), compulsory worker participation in firms’ dividends, and other remunerations (i.e., transportation, commissions, bonuses, childcare). The poor contributions of the within, entry, and exit components to aggregate productivity growth are also linked to these distortionary labor costs. Rising labor costs as firms age divert resources away from productive investments and technology adoption. In turn, technical efficiency grows at a slower rate (the within component). High firing costs incentivize unproductive firms to continue operating rather than 17 exiting, in order to avoid layoff-related costs. Moreover, high labor costs that impede the productivity growth of firms across the life cycle have a second-order effect on aggregate productivity growth by encouraging the entry of unproductive firms while discouraging the entry of productive enterprises. Atkeson and Burnstein (2010) and Hsieh and Klenow (2014) explain that constraints on incumbents’ productivity growth lead to higher entry rates for unproductive firms as they face less competition from productive incumbents, thus explaining the large share of low-productivity entrants in Ecuador. 5.2.2 Credit Access Frictions The between-firm component's underperformance in Ecuador can also be attributed to credit constraints arising from distortive interest rate regulation and excessive loan collateral requirements. Ecuador’s interest rate regulation includes differential rates based on firm sales and interest rate ceilings. 30 Figure A 6 reveals that Ecuadorian firms with more than USD 5 million face an average effective loan interest rate of 8.5 percent, almost 5 percentage points higher than the US average. However, this complex regulatory system is very distortive. Effective interest rates are close to the ceiling for all loan categories, implying that interest rates do not accurately reflect credit risks. As a result, financial institutions demand substantial collateral, in addition to high interest rates, hindering firms' access to financial resources and efficient capital allocation (Vijil, 2021). Weak institutional frameworks and poor contractual enforcement, as noted by Iacovone and Patiño Peña (2021), further exacerbate collateral constraints in developing countries like Ecuador. Figure A 7 confirms that firm loans in Ecuador more often require collateral in comparison to peers, with the required value exceeding 200 percent of the loan value, on average. Moreover, Figure A 8 indicates that smaller firms are the most affected by these constraints, as they face higher interest rates (with those earning less than USD 100,000 in sales encountering rates above 20 percent) and more restrictive collateral constraints. Consequently, the SME segment is underserved, with private banks primarily focusing on large enterprises and only a limited number of micro firms receiving financing from microfinance institutions (Partow and Vijil, 2021). The left panel of Figure 9 reveals the average difference in revenue TFP between firms with credit access and those without, categorized by firm size. On average, firms with credit access demonstrate higher productivity compared to credit-constrained firms across all size groups, but this TFP differential diminishes for larger firms. Notably, the revenue TFP differentials of firms with credit access from financial institutions, are positive for all size groups except large firms. Large firms with access to credit are 3 percent less productive than their credit constrained large counterparts, but this difference is not significant. This implies that large enterprises with bank credit are not more productive than their credit-constrained large counterparts, indicating higher misallocation of credit, particularly bank lending, among larger firms. Furthermore, this reinforces that credit constraints become less binding as firms grow larger, allowing some unproductive large firms to access credit while limiting credit access to only highly productive small firms. The left panel of Figure 10 shows that, conditional on credit access, smaller firms exhibit higher levels of TFP relative to larger firms. Large firms with access to credit are 20 percent less productive than micro enterprises with credit access. These findings suggest an inefficient allocation of capital between firm size groups, with unproductive large firms inefficiently hoarding capital and productive small firms facing limited access to additional capital due to higher credit constraints, as suggested by Figure A 8. Hence, existing credit frictions result in an inefficient allocation of banking credit among larger firms and lower access to credit for small productive firms, which in turn negatively impact aggregate productivity growth through the between component. The right panel of Figure 9 demonstrates that, on average, firms with 30 In Ecuador, there are 23 different interest rate categories depending on the type of loan and all have a corresponding interest rate ceiling. Business loan interest rates vary according to the level of firms’ sales. The different business loan interest rate categories are the following. The Corporate interest rate applies to firms with sales above USD 5 million. The Business interest rate corresponds to firms with sales between USD 1 million and USD 5 million. The SMEs interest rate corresponds to firms with sales between USD 100,000 and USD 1 million. The Microcredit – Expanded Accumulation interest rate corresponds to firms with sales between USD 20,000 and USD 100,000. The Microcredit – Simple Accumulation interest rate corresponds to firms with sales between USD 5,000 and USD 20,000. The Microcredit – Retail interest rate corresponds to firms with sales less than USD 5,000. 18 credit access are more productive than those without, across all age groups. However, as seen in the right panel of Figure 10, conditional on credit access, older firms are less productive than younger ones after 5 years of experience. Additionally, firms aged 20 years or older are 3 percent less productive than firms aged between 0 and 4 years. This also highlights allocative inefficiencies that impact the between component, as older firms, benefiting from lower collateral constraints or reduced interest rates due to their size or market experience, inefficiently hoard capital relative to younger, more productive firms. Figure 9: Revenue TFP Differentials of Credit Access by Firm Size and Age Source: INEC. Authors’ Elaboration. Note: This Figure reports the average treatment effect of access to credit by firm size (top left panel) and age (top right panel), capturing the difference in revenue TFP between those firms with access to credit relative to firms without access to credit for the different size and age groups. The results of this Figure are constructed from a regression of revenue TFP on a dummy of access to finance or access to finance from banks, interactions between these dummies and size (left panel) and age (right panel), for the 2012 – 2020 period. Regressions control for year fixed effects. Credit constraints also impact technical efficiency growth and market selection in Ecuador. Limited access to credit also reduces incentives for productive investments, depleting technical efficiency growth (the within component). Figure 10 suggests that younger firms with access to lending tend to be more productive than older firms with credit access. However, due to their smaller size, younger firms are less likely to have access to credit, which hampers their ability to invest in capability upgrading. Like the effect generated by highly distortive labor costs, credit constraints that impede productivity growth of firms, promote the inefficient entry of firms, which depletes the contribution of the entry component to aggregate productivity growth. Figure 10: Revenue TFP Premium by Firm Size and Age for Firms with Credit Access Source: INEC. Authors’ Elaboration. Note: This Figure reports the revenue TFP premiums of firm size groups, conditional on firms having credit access. The control group is the micro firm group. The bottom fight panel reports the revenue TFP premiums of firm age groups conditional on firms having credit access. The control group is the less than 5 years age group. Regressions control for year fixed effects. 19 5.2.3 Trends of Innovation Inputs The slow growth in aggregate technical efficiency in Ecuador during the analyzed period can also be attributed to poor trends in firms’ innovation inputs, which are also closely linked to the previously discussed factor market distortions.31 The left panel of Figure A 9 illustrates that the average years of managerial experience across firms increased marginally for all firm size groups. The right panel of Figure A 9 shows a decline in the share of medium and large firms conducting research expenditures, indicating a reduced willingness to make productive investments over the analyzed period. Labor cost distortions play a role, as larger firms facing higher labor costs allocate resources to meet these expenses instead of financing research activities for innovation. Furthermore, credit constraints for small firms limit their innovation investments, leading to a lower share of research-conducting small firms compared to medium and large firms. Moreover, factor market distortions have also reduced the sorting of higher-skilled workers into more productive firms. Figure A 10 shows that the share of high-skilled workers in high-productivity firms (top quintile of the revenue TFP distribution) decreased over the period analyzed, while it increased for low productivity firms (bottom two quintiles of the revenue TFP distribution). Hence, the complementary human capital necessary for innovation in highly productive establishments has decreased, as it is misallocated in less productive firms. This deficient performance of innovation inputs supports the hypothesis that aggregate technical efficiency growth in Ecuador is driven by international technology trends rather than by firm catch-up to the technological frontier through innovation. 5.2.4. Insolvency Regimes In Ecuador, weak insolvency regimes contribute to poor aggregate TFP outcomes primarily through inadequate market selection. The time to resolve insolvency in Ecuador exceeded five years during the analyzed period, contrasting with three years in the rest of Latin America and one year in the US. 32 As such, weak insolvency regimes have likely discouraged unproductive firms from leaving the market, like zombie firms.33 Figure A 11 reveals that zombie firms in Ecuador are slightly less productive than non-zombie firms, consistent with the findings of Adalet McGowan et al. (2017). Figure A 12 shows that the share of firms unable to service their debt has increased over the analyzed period, particularly after 2017. Furthermore, as seen in Figure 7, the contribution of the exit component to aggregate TFP growth declined after 2016. Reconciling the trends in Figures A 11 and A 12 with those of Figure 7 indicates that firms that should be exiting the market, like zombie firms, are not doing so. As insolvent zombie firms are less productive, their continued participation in the market leads to inefficient hoarding of resources, which could have been allocated to more productive firms that are likely exiting. In turn, this inefficient exit of more productive firms is manifested through the declining contribution of the exit component to aggregate productivity growth. 5.2.5 Evidence of Drivers of Allocative Inefficiencies To provide further insight on the potential drivers of poor productivity trends in Ecuador, this paper studies the relationship between sector-level allocative efficiency and proxies of structural and policy drivers of misallocation. Formally, the following regression is estimated: . = + , + + + , , (3) where . is sector ’s covariance between firms' revenue productivity and their value-added share in year . This covariance measures the extent to which firms with higher productivity capture a larger share 31 As explained by Cirera and Maloney (2017), inputs for the innovation process of firms are: technology, equipment, R&D, intellectual property use, human capital, training, engineering and design, software and databases, and managerial and organizational capital and practices. 32 World Development Indicators. 33 Zombie firms are defined as in Adalet McGowan et al (2017): A firm is considered a zombie firm if it is more than ten years old, and it has an interest coverage ratio less than one for more than three years. The interest coverage ratio is the ratio of operating income (gross income or sales minus operating expenses) to interest payments. 20 of economic activity, and thus, higher levels of capital and labor inputs. Higher values for this covariance reflect higher within-sector allocative efficiency.34 The structural or policy variables of interest are denoted by , . The policy variables considered are job-to-job mobility, non-wage labor costs relative to total labor costs, minimum wage incidence, share of firms with access to credit from a financial institution, zombie firm presence, and product market concentration. 35 Last, and are sector and year fixed-effects, respectively. Columns (1) – (6) of Table 1 report the regression estimates of coefficient in Equation (3) for each policy variable independently. Column (7) refers to the same specification but includes all regressors simultaneously. The results of Table 1 suggest that sectors characterized by higher non-wage labor costs, increased minimum wage incidence, and lower job-to-job mobility display lower allocative efficiency, signaling distortive labor market policies. Also, sectors with more access to credit from financial institutions exhibit higher levels of allocative efficiency, indicating that firms in industries with lower credit constraints, such as less binding collateral requirements, experience more efficient resource allocation. Weak insolvency systems enable unproductive firms to persist in the market, leading to excessive retention of labor and capital resources by these firms. As such, industries with a higher prevalence of zombie firms display lower levels of allocative efficiency. Furthermore, industries characterized by higher levels of product market concentration exhibit lower levels of allocative efficiency. This negative relationship underscores the presence of potential competition constraints and product market frictions that hinder the efficient allocation of resources across firms. Consequently, more productive firms have lower market shares of sales. For example, Ferro and Patiño Peña (2023) find a negative correlation between the presence of State-owned enterprises (SOEs) and allocative efficiency at the sector level.36 Higher product market concentration also reflects barriers to entry, especially for higher-productivity firms (as most entrants are of low productivity), with detrimental effects on aggregate TFP through the market selection channel. Thus, while input market frictions play a significant role in depleting aggregate productivity, it is also important to address product market frictions to facilitate the efficient allocation of factors in Ecuador. 34 Regressions are carried out at the ISIC 2-digit level of sector granularity. Angrist and Pischke (2008) recommend that aggregated sectoral observations, from firm-level data, be weighted by the number of firm observations within each sector-year. Hence, regressions are weighted in this manner. 35 Job-to-job mobility is measured as the ratio of workers in a sector that transitioned into other jobs between t-1 and t relative to the average, between t-1 and t, of the number of workers in the same sector. The number of industry observations for specification (1) is lower given that one period is lost because of the job-to-job mobility definition. Non-wage labor costs relative to total labor costs measures how much of total labor costs are allocated towards employment costs that arise from mandated labor regulations instead of workers’ labor compensation. It is calculated as the sector average of firms’ ratio of non-wage labor costs to total labor costs. Total labor costs consist of wages plus social security contributions, severance pay costs, mandatory compulsory allowances (i.e., “thirteenth” and “fourteenth” salaries), compulsory worker participation in firms’ dividends, and other remunerations (i.e., transportation, commissions, bonuses, and childcare). Non-wage labor costs include mandatory compulsory allowances (i.e., “thirteenth” and “fourteenth” salaries), compulsory worker participation in firms’ dividends, and other remunerations (i.e., transportation, commissions, bonuses, childcare). The minimum wage incidence is measured as the ratio between the mandated minimum wage and the sector median wage. Higher values of this ratio imply that the minimum wage is more binding for firms within the sector. The share of firms with access to credit from a financial institution is measured as the ratio of firms in a sector with a loan from a financial institution relative to the total number of firms in the sector. Zombie firm presence is measured as the share of zombie firms relative to the total number of firms in a sector. The product market concentration variable was calculated using the sectors’ HHI. 36 Ferro and Patiño Peña (2023) find a negative and significant relationship between the SOE share of sector-level employment and the sector covariance between firms' revenue productivity and their value-added share, i.e., a measure of allocative efficiency. 21 Table 1: Drivers of Allocative Efficiency Dependent Variable: Sector covariance between firm revenue productivity and value-added share Structural/Policy Drivers: (1) (2) (3) (4) (5) (6) (7) Job-to-job mobility 5.804 3.892 [0.115]*** [0.111]*** Non-Wage Labor Costs/Total Labor Costs -0.554 -0.42 [0.015]*** [0.015]*** Minimum Wage Incidence -1.289 -0.295 [0.030]*** [0.032]*** Access to Finance from Financial Institution 3.225 3.038 [0.021]*** [0.024]*** Zombie Firm Presence -0.140 -0.270 [0.020]*** [0.02]*** Product Market Concentration -0.444 -0.037 [0.022]*** [0.022]* N 592 666 666 666 666 666 592 Source: INEC. Authors’ Elaboration. Note: This table reports the coefficient, , of Equation (3)’s specification. Regressions are carried out at the ISIC 2-digit level of sector granularity. The following sector indicators of structural and/or policy drivers of misallocation are considered: job-to-job mobility, non-wage labor costs relative to total labor costs, minimum wage incidence, share of firms with access to credit from a financial institution, zombie firm presence, and product market concentration. Job-to-job mobility is measured as the ratio of workers in a sector that transitioned into other jobs between − 1 and relative to the average, between − 1 and , of total the number of workers in the same sector. The number of industry observations for specification (1) is lower given that one period is lost as a result of the job-to-job mobility definition. Non-wage labor costs relative to total labor costs measures how much of total labor costs are allocated towards employment costs that arise from mandated labor regulations instead of workers’ labor compensation. It is calculated as the sector average of firms’ ratio of non- wage labor costs to total labor costs. Total labor costs consist of wages plus social security contributions, severance pay costs, mandatory compulsory allowances (i.e., “thirteenth” and “fourteenth” salaries), compulsory worker participation in firms’ dividends, and other remunerations (i.e., transportation, commissions, bonuses, childcare). Non-wage labor costs include mandatory compulsory allowances (i.e., “thirteenth” and “fourteenth” salaries), compulsory worker participation in firms’ dividends, and other remunerations (i.e., transportation, commissions, bonuses, childcare). The minimum wage incidence is measured as the ratio between the mandated minimum wage and the sector median wage. Higher values of this ratio implies that the minimum wage is more binding for firms within the sector. The share of firms with access to credit from a financial institution is measured as the ratio of firms in a sector with a loan form a financial institution relative to the total number of firms in the sector. Zombie firm presence is measured as the share of zombie firms relative to the total number of firms in a sector. The product market concentration variable calculated using the sectors’ HHI. Each regression controls for industry and year fixed effects. Columns (1) – (6) consider only one regressor at a time, while column (7) includes all regressors within the specification. As Angrist and Pischke (2008), regressions weigh aggregate sectoral observations by the number of firm observations within industry-year. 6 “Missing Large” and “Missing Young” Firms in Ecuador This section discusses supportive evidence that the previously studied market distortions have also led to anemic firm size and firm age distributions in Ecuador. The distribution of firms by size in Ecuador remained stable for the sample period, as Figure 11 illustrates. Approximately 90.9 percent of formal firms in Ecuador are classified as micro firms, while small, medium, and large firms represent around 7.0, 1.7, and 0.4 percent, respectively.37 Figure A 13 (left panel) displays the distribution of firms by size in the US. Although microenterprises still dominate, their contribution is more than 30 percentage points lower than that of Ecuador. Conversely, the share of large firms in the US is more than double that of Ecuador. Consequently, relative to the US, Ecuador exhibits an underrepresentation of large firms and an overrepresentation of microenterprises. 37 Considering the limitations of the data, Figure 11 depicts the distribution of only formal firms. Given that smaller employment levels are typically associated with informal productive units, it is expected that the concentration of firms in smaller size groups would be even higher for the total economy (formal plus informal enterprises). 22 Figure 11: Distribution of Firms by Firm Size Group Source: Author’s Elaboration, INEC. Note: This figure studies the distribution of firms by firm size for the 2012 – 2020 period. Despite comprising less than 1 percent of the total number of firms, large firms held the largest share of formal employment, rising from less than 37 percent in 2012 to 39 percent in 2029 (Figure 12). Microenterprises constituted the second-largest contributor to formal employment, accounting for approximately 30 percent of the labor force. The employment share of medium-sized firms was around 17 percent between 2012 and 2020, while the share of small firms fell by more than 3 percentage points, from 16.9 to 13.8 percent. Figure A 14 displays the distribution of sales by firm size. Large firms accounted for around 58 percent of total sales between 2012 and 2020, surpassing their contribution to employment. Moreover, the medium, small, and micro-sized groups averaged sales shares of 20 percent, 15 percent, and 7 percent, respectively. Comparing the employment distribution by firm size between Ecuador and the US (right panel of Figure A 13) shows that large firms in the US account for over 65 percent of employment, which is 25 percentage points higher than in Ecuador. In contrast, micro firms only contribute 5 percent to employment in the US, which is 25 percentage points lower than in Ecuador. This disparity indicates a notable absence of large firms in Ecuador, not only in terms of number of firms, but also in terms of their share of labor. It also highlights that smaller firms in Ecuador account for an excessive share of employment. Many developing countries exhibit an underrepresentation of medium and large firms compared to their developed counterparts. These distributional differences between emerging economies and developed ones have been linked to market inefficiencies, including those discussed in the preceding sections (Hsieh and Olken, 2014; Krueger, 2013; Banerjee and Duflo, 2005, 2011; among others). In the case of Ecuador, the positive relationship between labor costs and firm size implies that firms have less incentive to expand their workforce, leading to a scarcity of large companies. When considering the combined findings from Figure 11 and Figure 3, it becomes evident that this distributional trend implies a larger presence of unproductive firms operating, ultimately contributing to lower aggregate technical efficiency. Moreover, the lower productivity levels of smaller firms (Figure 3), combined with their larger employment share (Figure 12), suggest a lower allocative efficiency. These static considerations are also complemented by dynamic trends to yield the observed size distribution. For example, increasing labor costs across the life cycle and limited credit access for more productive younger firms constrain firm TFP growth and subsequently lead to inefficient accumulation of factors (Figure 6). Although firms’ capital and employment grow throughout their life cycle, growth rates are suboptimal as they are attributed to market imperfections encouraging inefficient factor hoarding of inputs rather than TFP enhancements. Consequently, the absence of sustained firm TFP growth hinders size expansion and contributes to the underrepresentation of large firms in the economy. 23 Figure 12: Distribution of Employment by Firm Size Group Source: Author’s Elaboration, INEC. Note: This figure studies the distribution of employment by firm size for the 2012 – 2020 period. Between 2012 and 2020, the age composition of formal firms in Ecuador experienced a shift toward greater maturity. That is, the proportion of firms with less than five years of age declined from nearly 50 percent in 2012 to 32 percent in 2019, while the shares of firms aged between 15 and 19 years and those with 20 or more years increased by more than 5 percentage points each (Figure 13). The aging of firms can be beneficial if they grow in productivity across their life cycle. However, since firm aging in Ecuador is not accompanied by productivity growth, this further highlights constraints to technical efficiency growth as well as barriers to efficient market selection. Furthermore, persistent barriers to firm exit, such as high firing costs and weak insolvency regimes, likely contributed to the aging of firms by preventing unproductive older firms from leaving the market. Figure 13: Distribution of Firms by Firm Age Group Source: Author’s Elaboration, INEC. Note: This figure studies the distribution of firms by firm age for the 2012 – 2020 period. Figure 14 shows that employment in Ecuador has consistently been concentrated in older firms. Like the distribution of firms, this concentration has intensified between 2012 and 2020, with the share of employment in firms aged 20 or more increasing by over ten percentage points. Conversely, the share of employment in firms with less than five years of age experienced a similar magnitude of decline throughout the analysis period. These trends persisted during the pandemic, indicating the vulnerability of younger firms to shocks compared to their older counterparts, as older firms were more successful in retaining employment. The concentration of sales in older firms also grew over the same period (Figure A 15). Throughout the analysis period, firms aged 20 or more consistently accounted for a higher share of sales relative to their representation in terms of firm count and employment. This share reached its peak during 24 2020, with older firms contributing to over 50 percent of total sales. On one hand, the resilience of older firms highlighted their ability to navigate the challenges of the economic slowdown as well as the pandemic. However, the shift of economic activity to older firms also reflects frictions that lead to allocative inefficiencies, as older firms are not necessarily more productive, as suggested by Figure 3 and Figure 8. Increasing labor costs across the firm life cycle suggest that older firms likely have higher firing costs associated with an aging workforce, making it more difficult for older, unproductive firms to downsize during economic downturns and encouraging their inefficient hoarding of labor. Figure 14: Distribution of Employment by Firm Age Group Source: Author’s Elaboration, INEC. Note: This figure studies the distribution of employment by firm age for the 2012 – 2020 period. 7 Role of Firm Dynamics in Job Flows This section examines the job creation and destruction patterns of Ecuadorian firms to understand how market imperfections affect labor flows. The top left panel of Figure 15 illustrates that job flows of formal firms in Ecuador followed the country's business cycle. The highest net job creation by formal firms coincided with the years of highest economic growth (2013 and 2014), while the two years marked by recessions, 2016 and 2020, resulted in significant job destruction. Additionally, formal employment declined in 2015 and 2019, coinciding with GDP growth rates close to zero percent. These patterns align with findings in the literature for developing countries, which show that formal employment is pro- cyclical.38 Despite the notable variation in net employment changes, formal firms experienced a net increase of 181,000 workers between 2012 and 2019. However, due to the substantial employment reduction during the pandemic-led recession, the net change in formal employment between 2012 and 2020 was negative, with 11,000 jobs lost. Following Davis et al. (1996), Decker et al. (2014), and Haltiwanger et al. (2016), year-to-year changes in aggregate employment are decomposed according to: ∆ = ∆ + ∆ + ∆ + ∆ . (4) ∆ ∆ The extensive margin of aggregate job changes is made up of vacancies filled by entrants and jobs destroyed by exiter firms. The intensive margin accounts for the net creation of jobs by continuing firms. For a further discussion of this decomposition, please refer to Appendix 1. 38 This finding is documented by various papers such as Leyva and Urrutia (2023), Leyva and Urrutia (2020), Bosch and Maloney (2008), among others. 25 The top right panel of Figure 15 reports that in the four years with negative employment variations, the intensive margin accounted for the negative aggregate employment changes, indicating that these were driven by continuing firms laying off workers. Hence, during economic downturns, job creation by continuing firms was small compared to job destruction. Conversely, the job creation by entrant firms was offset by job destruction by exiting firms in recessions. In years with positive employment changes, the extensive margin had a greater impact on net employment variation in comparison to the intensive margin, underscoring the importance of firm entry in creating formal jobs in economic booms. The bottom panel of Figure 15 illustrates the decomposition of employment flows into the four components of Equation (4). The contribution of firm entry to job creation rose relative to the expansion of labor demand by continuing firms over the period of analysis. In 2012, entrants accounted for 52 percent of new jobs in Ecuador, while this contribution increased to 64 percent by 2020. This increasing contribution of firm entry to aggregate job creation is indicative that operating firms in Ecuador experienced sluggish technical efficiency growth between 2012 and 2020, as depicted in Figure 7 by the marginal contribution of the within component. This resulted in a higher influx of new firms creating jobs, since they faced less competition from high-productivity incumbents (Atkeson and Burstein [2010] and Hsieh and Klenow [2014]). Given that a large portion of entrants exhibit low productivity, this increase in job creation through the extensive margin was detrimental to aggregate TFP by exacerbating allocative inefficiencies. Figure 15: Decomposition of Net Job Flows Source: Authors’ Elaboration, INEC. Note: This figure studies the decomposition of yearly net employment variation in Ecuador for the 2012 – 2020 period. The decompositions follow the approaches developed by Davis et al. (1996), Decker et al. (2014), Haltiwanger et al. (2016) by decomposing job flows into two intensive and two extensive margin components following Equation (4). The intensive margin (layoffs by continuing firms) and extensive margin (jobs destroyed by exiting firms) contributed similarly to the decline in formal jobs, except in recessions. Economic downturns witnessed job destruction primarily through downsizing by operating firms rather than firms exiting the market. This further underscores that distortions, such as high firing costs and weak insolvency regimes, impede the 26 optimal exit of firms during recessions. For instance, in years marked by negative economic shocks (such as 2016 and 2020), unproductive firms in Ecuador opted to continue their operations and reduce only a fraction of their workforce, incurring in lower costs associated with severance pay compared to the costs of firing their entire workforce upon their exit. Consequently, the persistent market presence and employment retention of these unproductive firms contribute to lower aggregate TFP. In addition to their overrepresented presence in employment (Figure 12), micro firms play a crucial role in driving aggregate job flows in the country.39 Large enterprises, although underrepresented, also contribute significantly to employment dynamics. Notably, the intensive margin holds greater importance for aggregate job flows across all firm sizes (Figure 16), particularly during economic downturns, such as the years 2016 and 2020. This suggests that, during recessions, the previously discussed distortions to efficient market selection constrain firm exit regardless of firms’ employment level, as enterprises of all sizes opt to reduce their workforce instead of exiting the market. Such dynamics impede allocative efficiency when unproductive firms are discouraged from exiting, as is the case of many microenterprises, which are generally less productive and still choose to contract their labor demand instead of exiting. Conversely, during economic booms, such as in 2013 and 2014, the extensive margin plays a prominent role in job flows for micro and small enterprises, indicating that smaller firms, despite their lower productivity, create numerous jobs. This reinforces previous findings of inefficient entry by unproductive firms due to limited competition from incumbents with constrained productivity growth. Notably, even during recession years (2016 and 2020), entrant micro firms still generated over 100,000 jobs, suggesting the persistence of this mechanism across all years, with amplified effects during economic booms. Figure 16: Decomposition of Net Job Flows by Firm Size Group Source: Authors’ Elaboration, INEC. Note: This figure studies the decomposition of yearly net employment variation in Ecuador by firm size group for the 2012 – 2020 period. It follows the approaches developed by Davis et al. (1996), Decker et al. (2014), Haltiwanger et al. (2016) to decompose job flows. Figure 17, which analyzes net employment changes by age group, reinforces the significance of young firms (less than 5 years of age) in generating employment opportunities, surpassing the job creation of continuing firms in other age categories. This trend was maintained through economic booms. During recessions, young firms still contribute to net job creation, although to a lesser extent, as the jobs created by entrants are offset by labor demand contractions among young incumbents. In contrast, older continuing firms tend to reduce labor demand during economic downturns, leading to significant job displacements. For instance, in 2016, firms aged 20 years or older displaced 56,000 workers, while in 2020, they eliminated 69,000 jobs. The labor adjustment through the intensive margin by older firms relative to the extensive margin is worrisome, particularly considering their declining productivity levels illustrated in Figure 6. Due 39 If informal firms were considered, it would be expected that the contributions of micro and small firms to job flows would increase as informal firms are likely smaller. 27 to existing barriers to exit, unproductive older firms persist in the market and continue to hoard employment, resulting in inefficiencies in resource allocation. Figure 17: Decomposition of Net Job Flows by Firm Age Group Source: Authors’ Elaboration, INEC. Note: This figure studies the yearly decomposition of net employment variation in Ecuador by firm age group for the 2012 – 2020 period. It follows the approaches developed by Davis et al. (1996), Decker et al. (2014), and Haltiwanger et al. (2016) to decompose job flows. There are entrants of age greater than 0 because firm dynamics are defined based on whether a firm is active or not, and age is defined based on a firm’s year of constitution. A firm is active if it reports employment or sales in a given year. Figure 18 depicts the patterns of job creation across revenue TFP quintiles for net job-creating firms (left panel), and for net job-destroying firms (right panel). It is evident that firms in the top quintile of the productivity distribution generate fewer jobs but also exhibit greater job stability by destroying fewer positions.40 This suggests that higher productivity firms experience lower labor turnover, a pattern likely influenced by existing market frictions. For instance, as highlighted in Figure 8, higher productivity firms face higher labor costs, which reduce their employment churning. This is because higher productivity firms attract workers with greater human capital (Figure A 10), leading to higher wage payments, on average, compared to lower productivity firms. However, these higher wages are linked with increased compulsory allowances and firing costs, which take away incentives for these firms to expand their workforce or terminate poor performers. Figure 18: Net Job Creation and Destruction by TFP Quintiles Source: INEC. Authors’ Elaboration. Note: The left panel of this figure plots the average change in employment for firms that reported positive net job creation by revenue productivity quintile. The right panel of this figure plots the average change in employment for firms that reported negative net job creation (i.e., job destruction) by revenue productivity quintile. 40 If informal firms were considered, it would be expected that the net creation and net destruction of jobs for the lower productivity quintiles would be even higher, as informal firms are likely less productive. 28 8 Role of Firm Dynamics in Wage Inequality This section explores the nexus between firm outcomes and wage inequality in Ecuador. Figure 2 reveals a steady growth in aggregate productivity until 2014, accompanied by a decline in inequality. However, since 2014, productivity declined, while inequality stagnated at first, followed by a rise since 2018. Section 5’s empirical results indicate that the decline in aggregate productivity was driven by allocative efficiency losses caused by various distortions in factor markets as well as other institutional constraints. This raises the question of whether these drivers also impact aggregate inequality trends, highlighting the need to examine the link between firm-level outcomes (i.e., productivity, factor demands) and worker-level outcomes (i.e., wages). After characterizing the importance firms have for wage inequality, this section also examines the extent to which changes in firm performance translate into changes in wages by estimating the revenue productivity-wage pass-through. 8.1 Empirical Methodology: Wage Inequality Decomposition and Revenue Productivity-Wage Pass-Through Extensive literature has documented the importance of firm outcomes in explaining wage differences among workers in developed and emerging economies.41 This paper decomposes aggregate wage inequality into within-firm and between-firm wage inequality. Within-firm wage inequality accounts for wage disparities for different workers within the same firm, while between-firm wage inequality considers the variation in wages of similar workers across different firms. Following Barth et al. (2016) and Criscuolo et al. (2020, 2021), wages are expressed using a standard human capital earnings equation with firm fixed effects: , = , + + , , (5) where ℎ denotes a worker and denotes a firm. Aggregate wage inequality is measured as the variation in log wages, (ln ), and its decomposition is given by: (6) , = , + , − , + , + , +( ) , , , , , where = , = , and is the average of all workers’ predicted wages based on , , observable characteristics, , , within a firm . As shown in Equation (6), within-firm wage inequality arises from two channels: intra-firm workforce composition and returns to workers' characteristics. The intra-firm workforce composition channel refers to heterogeneity in workers' abilities within the firm, while the returns to characteristics channel relates to factors such as education, experience (age), and gender. Similarly, between-firm wage inequality is determined by two other channels: worker sorting across firms and firm wage premia. The sorting channel captures wage variation arising from differences in the skill composition of workers across firms. For example, high-skilled workers with high earnings may likely be employed in firms that offer high wage premia or that hire other high-skilled workers. On the other hand, the firm wage premia channel reflects wage differences resulting from variation in firms' revenue TFP and/or the sharing of rents between workers and firms. Technological frictions in labor markets, such as search and information frictions, as well as market distortions affecting product and labor markets (i.e., firing costs, limited access to credit, barriers to competition, among others), contribute to the generation of 41 Papers such as Faggio et al. (2010), Barth et al. (2016), Berlingieri et al. (2017), Manning (2021), Criscuolo (2020, 2021, 2022), among others, have documented that firms are important for explaining the wage variation of workers. 29 dispersion in revenue productivity and labor market rents. Consequently, the firm wage premia channel represents how factor misallocation across firms can impact workers' wages.42 Moreover, the impact of misallocation (i.e., policy distortions) on wage inequality can be captured through the relationship between firms' revenue TFP and workers' wages, commonly known as the revenue productivity-wage pass-through. This pass-through determines how changes in revenue TFP are translated into changes in wages. Following Criscuolo (2021) revenue productivity-wage pass-through can be as follows: , = , + , + + + , , (7) where ℎ denotes a worker, denotes a firm, denotes a sector, and denotes a year. Parameter captures the pass-through of revenue productivity, , to wages. To identify structural or policy factors that are associated with the pass-through of revenue TFP to wages, this paper follows Criscuolo et al. (2021) by allowing the coefficient on revenue TFP to vary according to structural or policy characteristics: , = , + , + , + , ∙ , + + + , , (8) where , is the structural or policy variable of interest. 8.2 Drivers of Wage Inequality As illustrated in the left panels of Figure 19 and Figure A 16, aggregate wage inequality among formal workers in Ecuador ranged from 0.3 to 0.4 between 2012 and 2021. These levels of wage inequality are consistent with the wage variation estimates reported by Criscuolo (2020) for OECD countries. Moreover, the left panel of Figure 19 highlights that more than half of the aggregate wage inequality in Ecuador can be attributed to differences in pay across firms for workers with similar characteristics (i.e., between-firm wage inequality). The right panels of Figure 19 and Figure A 16 reveal a rapid decline in aggregate wage inequality among formal workers until 2015, followed by a slower rate of reduction until it eventually stagnated and began to increase by the end of the period, mirroring the trend observed in the Gini Index as depicted in Figure 2. Furthermore, the right panel of Figure 19 highlights that over 60 percent of the changes in wage inequality stem from fluctuations in between-firm wage inequality. These findings underscore the importance of firm-level outcomes in driving wage inequality in Ecuador, which aligns with the findings of Barth et al. (2016) and Criscuolo (2020). 42 As highlighted by Van Biesebroeck (2014) and Criscuolo et al. (2021), under frictionless product and factor markets, differences in firm-level physical productivity result in variations in employment, with more productive firms employing a larger workforce compared to less productive firms. Additionally, marginal products of labor for workers with the same skill level equalize across firms, resulting in equal firm revenue TFP and wages for workers with similar characteristics. However, when input and output markets are characterized by technological frictions or policy- based distortions, marginal products of labor no longer equalize across firms, leading to wage disparities among workers with similar characteristics for two reasons. First, market distortions generate labor market rents that can be shared between firms and workers. Differences in workers' bargaining power across firms determine the sharing of these rents, leading to wage variations between firms for similar workers. Second, market distortions generate dispersion in revenue TFP, as marginal products differ across firms. Even with the same worker-firm bargaining power or any wage-setting mechanism, similar workers will receive different wages due to the distorted performance of firms, characterized by variations in marginal products and revenue TFP. This occurs because the labor market does not solely adjust through quantities but also through wages, given that marginal products do not equalize. For instance, firms with higher physical productivity demand more labor relative to those with lower physical productivity. However, since marginal products do not equalize, revenue TFP differs across firms, causing higher physical productivity firms to pay higher wages than lower physical productivity firms for workers with similar characteristics. Consequently, market frictions contribute to variation in firms' marginal products, resulting in revenue TFP dispersion across firms and wage inequality among similar workers. 30 Figure 19: Decomposition of Aggregate Wage Inequality and Changes of Aggregate Wage Inequality Source: INEC. Authors’ Elaboration. Note: The left panel of this figure decomposes aggregate wage inequality into the within- and between-firm components following Equation (6). The right panel of this figure decomposes changes in aggregate wage inequality into changes in the within-firm and changes in the between- firm components. The decomposition of between-firm wage inequality into sorting and firm wage premia, as depicted in Figure 20, reveals that between-firm wage inequality and its changes are mainly explained by firm wage premia.43 As explained in Section 3.3., firm wage premia represent variations in wages that stem from technological or policy-driven (i.e., high labor costs, limited access to credit, weak insolvency regimes, barriers to innovation) frictions that generate revenue productivity dispersion and labor market rents. Hence, in addition to negatively impacting firm performance and resource allocation, distortionary policies account for up to 41 percent of wage inequality in Ecuador. These frictions also generate sorting of workers according to firms’ wages instead of firms’ productivity, which results in the sorting channel contributing 12 percent of aggregate wage inequality. Figure A 18 demonstrates that firms in the top quintile of the firm wage distribution have a much higher share of high-skilled workers (ranging from 38 to 47 percent) compared to firms in the bottom quintile (ranging from 11 to 14 percent).44 In contrast, Figure A 10 shows that the sorting of high-skilled workers to high-revenue TFP firms is less pronounced, highlighting the presence of allocative inefficiencies.45 Figure 20: Decomposition of Between-firm Wage Inequality and Changes of Between-firm Wage Inequality Source: INEC. Authors’ Elaboration. Note: The left panel of this figure decomposes between-firm wage inequality into the firm wage premia and the sorting components following Equation (6). The right panel of this figure decomposes changes in between-firm wage inequality into changes in the firm wage premia and changes in the sorting components. 43 Figure A 17 of the Appendix decomposes firm wage premia variation into a within-industry and between-industry component. As the within- industry component accounts for most of the variation of firm wage premia, it can be inferred that this variation is not driven by technological differences across sectors. 44 This substantial sorting of workers based on firms' wage premia is like the findings of OECD countries (Criscuolo et al., 2020). 45 Lochner and Schulz (2022) also find that productivity sorting of workers is also less pronounced than wage sorting of workers for Germany, explaining that this pattern is indicative of labor misallocation. 31 Figure 21 presents the decomposition of wage differences within specific firm size groups. These findings highlight that wage inequality increases with firm size. However, the channels driving this inequality differ across firm sizes. For workers in large firms, within-firm wage inequality accounts for up to 60 percent of their overall wage variation, followed by firm wage premia (approximately 25 percent) and sorting (around 15 percent). Notably, the differences in workers' abilities within firms contribute significantly to the wage inequality observed in larger establishments, as these firms are more likely to employ individuals with diverse skill levels. Consequently, while large firms may face more distortive frictions, disparities in large firms’ skill requirements primarily explain the wage differences within this size group. In contrast, workers in micro firms experience lower levels of wage inequality. In microenterprises, the within-firm component is considerably smaller, likely due to the employment of more homogeneous workers given their limited scale of operations. Additionally, the explanatory power of technological features and market distortions in generating rents is higher for wage inequality among workers in smaller firms. This suggests that misallocation has a more pervasive impact on earnings for individuals employed in smaller enterprises. When considering the overrepresentation of micro firms in the Ecuadorian economy, the cumulative effect of market inefficiencies on wage inequality becomes even more pronounced.46 Figure 21: Decomposition of Wage Inequality by Firm Size Source: Authors’ Elaboration, INEC. Note: This figure decomposes aggregate wage inequality into the within-firm, the firm wage premia, and the sorting components by firm size group following Equation (6). Figure 22 illustrates that the pass-through of revenue productivity to wages among formal workers in Ecuador ranged between 0.05 and 0.11. Hence, a 10 percent increase in revenue productivity corresponds to a 0.5 to 1.1 percent rise in wages, as estimated by Equation (7). In comparison to findings by Criscuolo et al. (2021), the pass-through estimates for formal workers in Ecuador are at the lower end, as OECD countries reported, on average, a pass-through of 0.15. As explained by Criscuolo et al. (2021), elevated 46 Additionally, this impact of misallocation on wage inequality would be even larger if informal firms were considered, as they are likely smaller and less productive. 32 pass-through levels can indicate the presence of resource misallocation. Figure 7 suggests significant levels of misallocation, which persistently affect aggregate productivity levels and growth. Figure 19 and Figure 20 reveal that firm wage premia, which capture the impact of resource misallocation on wage variation, account for substantial portions of aggregate wage inequality. How can these misallocation trends observed in Ecuador be reconciled with the low pass-through levels of revenue productivity to formal workers' wages? The revenue productivity-wage pass-through also captures rent sharing between firms and workers. Given the limited availability of formal sector jobs, workers experience diminished bargaining power, resulting in a reduced pass-through of changes in revenue productivity to wage increases. Figure 22: Revenue Productivity-Wage Pass-through Source: INEC. Authors’ Elaboration. Note: This figure plots the elasticity of wages to revenue TFP from a regression specified by Equation (7). In particular, it plots the estimated parameter from this specification. Analyzing the pass-through of revenue productivity to wages by gender shows that male workers demonstrated higher pass-through compared to female workers, between 2012 and 2017. However, after 2018, female workers experienced larger wage increments than their male counterparts, following shocks to revenue productivity. This catch-up in pass-through rates suggests that labor market opportunities may have expanded for women, subsequently enhancing their bargaining power. Additionally, similar to findings in other countries, high-skilled workers in Ecuador display a greater pass-through of revenue productivity to wages relative to low-skilled workers. This implies that individuals with higher abilities likely have more job opportunities in the formal labor market and consequently enjoy stronger bargaining power. As a result, these workers capture a larger share of the rents generated from increases in firms' revenue TFP. Figure 23: Revenue Productivity-Wage Pass-through by Gender and Skill-level Gender Skill Level Source: INEC. Authors’ Elaboration. Note: The left panel of the figure plots the elasticity of wages to revenue TFP from a regression specified by Equation (7) for each gender category. The right panel of the figure plots the elasticity of wages to revenue TFP from a regression specified by Equation (7) for each skill category. In particular, these panels plot the estimated parameter from Equation (7). 33 Figure 24 illustrates the impact of structural and policy factors on wage inequality through the firm performance channel (i.e., revenue productivity). Industries with higher non-wage labor costs demonstrate a higher pass-through rate. This suggests that labor frictions, such as firing costs and compulsory allowances, contribute to allocative inefficiencies that lead to wage disparities among workers with similar characteristics. Thus, these labor market distortions not only contribute to higher levels of misallocation, as shown in Table 1, but also generate wage dispersion through the misallocation channel. Figure 24 also demonstrates that industries with greater job mobility show a slightly lower pass-through rate, though not statistically significant. This contrasts with Criscuolo et al. (2021), who find that industries with higher job mobility have a much lower pass-through rate because sectors with higher job mobility reflect fewer labor market distortions, not only reducing within-sector labor misallocation but also reducing distortion-driven wage variation. Although Table 1 indicates higher allocative efficiency in industries with greater job mobility in Ecuador, this does not translate into lower pass-through levels for these industries. This is explained by the fact that workers in sectors with higher job mobility have more employment opportunities, increasing their bargaining power and resulting in a larger share of revenue productivity rents being shared with them. As a consequence, the effect of lower misallocation on pass-through levels is offset by the higher bargaining power of workers in sectors with greater job mobility. Figure 24: Revenue Productivity-Wage Pass-through by Structural/Policy Drivers Source: INEC. Authors’ Elaboration. Note: This figure plots the elasticity of wages to revenue TFP from a regression specified by Equation (8). In particular, it plots the estimated effect “ + ” from this specification. The regressors used in this exercise to study the four structural/policy drivers of misallocation and wage inequality are dummies that indicate whether the sector value of the continuous structural/policy variable is above (high) or below (low) the median industry value of the variable. Each regression controls for 1-digit industry and year fixed effects. Following Angrist and Pischke (2008), regressions weigh aggregate sectoral observations by the number of firm observations within an industry-year. Consistent with the findings of Criscuolo et al. (2021, 2022), industries subject to a more binding minimum wage demonstrate a reduced pass-through of revenue productivity to wages, indicating that minimum wages act as a mechanism to mitigate wage inequality.47 Under a scenario where firms possess high bargaining power, such as in the Ecuadorian formal sector, low-productivity firms can compete by offering lower wages without the risk of losing their workers to other firms. A more binding minimum wage policy hinders low-productivity firms from competing with lower wages, thereby reducing the transmission of revenue productivity differences into wage premia differences. Despite reducing wage inequality through this mechanism, minimum wage policy also generates misallocation, as seen in Table 1. Lastly, industries with lower product market competition exhibit higher pass-through levels. Table 1 highlights higher levels of misallocation in industries with limited competition, indicating that frictions that impede competition in 47 This has been documented in several papers in the literature, such as Engbom and Mozer (2022), Criscuolo et al. (2021, 2022), Autor et al. (2016), Lee (1999), among others. 34 product markets negatively affect the allocative efficiency of factor inputs. The higher pass-through in sectors with lower competition implies that increased barriers to competition result in higher wage inequality due to more allocative inefficiencies. 9 Conclusion Between 2012 and 2020, Ecuador's economic performance stagnated as growth driven by factor accumulation was offset by a significant decline in aggregate total factor productivity (TFP). This decline in aggregate TFP was driven by regulatory and institutional distortions that caused a persistent misallocation of factor inputs across firms. Furthermore, these frictions constrained firm-level technical efficiency growth and impeded healthy firm entry and exit dynamics. The decline in aggregate TFP was accompanied by a change in the trend of income inequality, whose decreasing tendency before 2014 plateaued thereafter and increased since 2018. These empirical patterns suggest that distortive policies with adverse effects on aggregate TFP also impact labor outcomes. Factor misallocation across firms in Ecuador can be attributed to distortions in labor and capital input markets, as distortionary labor regulations (such as high firing costs and ad-hoc compulsory allowances) as well as barriers to credit (such as high collateral constraints and high interest rates) hinder the fluidity of the labor market, constrain firm access to capital, and allocate factors towards less productive enterprises. These highly distortive factor market frictions also constrain firm productivity growth, as firms must divert resources from innovation and capability upgrading to finance costly factor inputs. As a result, innovation related activities, like R&D expenditures, have reduced and firm-level TFP growth is mostly stagnated across the life cycle. Furthermore, these factor market distortions along with institutional inefficiencies (such as weak insolvency regimes) yield inefficient market selection that is characterized by a large entry and low exit of unproductive firms. The policy-driven distortions that have led to Ecuador's underwhelming aggregate productivity outcomes have also generated distributional patterns of firms and job flows that further indicate the country's poor economic performance. Ecuador exhibits an underrepresentation of large firms in terms of numbers and economic activity (i.e., employment), while micro firms are overrepresented, also highlighting allocative inefficiencies as micro firms are less productive than large firms, on average. Additionally, these market frictions, coupled with the economic slowdown, have resulted in a shift of economic activity from younger to older firms. Aggregate job flows also portray anemic trends, with small, unproductive entrants creating most new jobs, while existing jobs are primarily destroyed by old, unproductive firms that stay in the market instead of exiting. Furthermore, these distortions that generate resource misallocation also account for up to 41 percent of aggregate wage inequality. These findings suggest that to enhance workers’ outcomes, it is also essential to address distortions that affect firm dynamics. References Ackerberg, D. A., Caves, K., & Frazer, G. (2015). Identification properties of recent production function estimators. Econometrica, 83(6), 2411-2451. Adalet McGowan, M., Andrews, D., & Millot, V. (2017). The walking dead?: Zombie firms and productivity performance in OECD countries. Aguilar, V. y Sarmiento, J. (2012). Estimación de la economía oculta del Ecuador: Aplicación de los métodos de consumo de energía, monetario y múltiples causas-múltiples efectos, para el período 1980- 2006. Servicio de Rentas Internas, 38-81. https://repositorio.flacsoandes.edu.ec/handle/10469/3867 Angrist, Joshua D., and Jörn-Steffen Pischke. "Parallel worlds: fixed effects, differences-in-differences, and panel data." Mostly harmless econometrics. Princeton University Press, 2008. 221-248. Atkeson, A., & Burstein, A. T. (2010). Innovation, firm dynamics, and international trade. Journal of political economy, 118(3), 433-484. 35 Autor, David H., Alan Manning, and Christopher L. Smith. "The contribution of the minimum wage to US wage inequality over three decades: a reassessment." American Economic Journal: Applied Economics 8.1 (2016): 58-99. Avellan, L., & Ferro, E (2017). Productividad Sectorial en Ecuador: 2009 – 2015. Mimeo. Banerjee, A. V., & Duflo, E. (2005). Growth theory through the lens of development economics. Handbook of Economic Growth, 1, 473-552. Banerjee, A, & Duflo, E. (2011). Poor economics (Vol. 619). Public Affairs. Bartelsman, E., and Mark Doms. 2000. “Understanding Productivity: Lessons from Longitudinal Microdata.” Journal of Economic Literature, 38 (3): 569-594. Bartelsman, E., Haltiwanger, J., & Scarpetta, S. (2013). Cross-country differences in productivity: The role of allocation and selection. American Economic Review, 103(1), 305-34. Barth, E., Bryson, A., Davis, J. C., & Freeman, R. (2016). It’s where you work: Increases in the dispersion of earnings across establishments and individuals in the United States. Journal of Labor Economics, 34(S2), S67-S97. Becker, R. A., Haltiwanger, J., Jarmin, R. S., Klimek, S. D., & Wilson, D. J. (2007). 12. Micro and Macro Data Integration: The Case of Capital (pp. 541-610). University of Chicago Press. Berlingieri, Giuseppe, Patrick Blanchenay, and Chiara Criscuolo. "The great divergence (s)." (2017). Bosch, Mariano, and William Maloney. Cyclical movements in unemployment and informality in developing countries. World Bank Publications, 2008. Busso, M., Madrigal, L., & Pagés, C. (2013). Productivity and resource misallocation in Latin America. The BE Journal of Macroeconomics, 13(1), 903-932. Camino‐Mogro, S., & Bermudez‐Barrezueta, N. (2021). Productivity determinants in the construction sector in emerging country: New evidence from Ecuadorian firms. Review of Development Economics, 25(4), 2391-2413. Cirera, X., & Maloney, W. F. (2017). The innovation paradox: Developing-country capabilities and the unrealized promise of technological catch-up. World Bank Publications. Criscuolo, C., Hijzen, A., Schwellnus, C., Chen, W. H., Fabling, R., Fialho, P., … & Zwysen, W. (2020). Workforce composition, productivity and pay: the role of firms in wage inequality. Criscuolo, C., Hijzen, A., Koelle, M., Schwellnus, C., Barth, E., Chen, W. H., ... & Zwysen, W. (2021). The firm-level link between productivity dispersion and wage inequality: A symptom of low job mobility?. Criscuolo, Chiara, Antton Haramboure, Alexander Hijzen, Michael Koelle, and Cyrille Schwellnus. "The role of firm performance and wage‑setting practices in wage inequality." (2022). Cusolito, A. P., & Maloney, W. F. (2018). Productivity revisited: Shifting paradigms in analysis and policy. World Bank Publications. Davis, S. J., & Haltiwanger, J. (1990). Gross job creation and destruction: Microeconomic evidence and macroeconomic implications. NBER macroeconomics annual, 5, 123-168. Davis, S. J., & Haltiwanger, J. (1992). Gross job creation, gross job destruction, and employment reallocation. The Quarterly Journal of Economics, 107(3), 819-863. Davis, S. J., Haltiwanger, J., & Schuh, S. (1996). Job Creation and Job destruction. Cambridge, MA: MIT Press. 36 Decker, R., Haltiwanger, J., Jarmin, R., & Miranda, J. (2014). The role of entrepreneurship in US job creation and economic dynamism. Journal of Economic Perspectives, 28(3), 3-24. Dunne, T., Roberts, M. J., & Samuelson, L. (1989). The growth and failure of US manufacturing plants. The Quarterly Journal of Economics, 104(4), 671-698. Eslava, M., & Haltiwanger, J. (2016). Who creates jobs: the role of age and size in Colombia vs. the US. Engbom, N., and Moser, C. "Earnings inequality and the minimum wage: Evidence from Brazil." American Economic Review 112.12 (2022): 3803-47. Faggio, G., Salvanes, K.G, and Van Reenen, J. "The evolution of inequality in productivity and wages: panel data evidence." Industrial and Corporate Change 19.6 (2010): 1919-1951. Foster, L., Haltiwanger, J. C., & Krizan, C. J. (2001). Aggregate productivity growth: lessons from microeconomic evidence. In New developments in productivity analysis (pp. 303-372). University of Chicago Press. Ferro, E, & Patiño Peña, F. (2023). Private Sector Performance Under the Pressure of SOE Competition – The Case of Ecuador. Mimeo. Gopinath, G., Kalemli-Özcan, Ş., Karabarbounis, L., & Villegas-Sanchez, C. (2017). Capital allocation and productivity in South Europe. The Quarterly Journal of Economics, 132(4), 1915-1967. Grosh,M., Kouame, A., Oviedo, A.M., Siegenthaler, P., Vostroknutova, E.2014. Informality and the labor market. Washington, DC : World Bank Group. Haltiwanger, J., Jarmin, R. S., & Miranda, J. (2013). Who creates jobs? Small versus large versus young. Review of Economics and Statistics, 95(2), 347-361. Haltiwanger, J., Jarmin, R. S., Kulick, R. B., & Miranda, J. (2016). High growth young firms: Contribution to job, output and productivity growth, US Census Bureau Center for Economic Studies Paper No. CES-WP-16-49. Hsieh, C. T., & Klenow, P. J. (2009). Misallocation and manufacturing TFP in China and India. The Quarterly Journal of Economics, 124(4), 1403-1448. Hsieh, C. T., & Klenow, P. J. (2014). The life cycle of plants in India and Mexico. The Quarterly Journal of Economics, 129(3), 1035-1084. Hsieh, C. T., & Olken, B. A. (2014). The missing “missing middle". Journal of Economic Perspectives, 28(3), 89-108. Hopenhayn, H. A. (1992). Entry, exit, and firm dynamics in long run equilibrium. Econometrica: Journal of the Econometric Society, 1127-1150. Iacovone, L., Muñoz Moreno, R., Olaberria, E., & Pereira Lopez, M. D. L. P. (2022). Productivity Growth in Mexico: Understanding Main Dynamics and Key Drivers. Iacovone, L. & Patiño Peña, F. (2021). Economywide Productivity Drivers and Job Dynamics. In Iacovone, L., Muñoz Moreno, R., Olaberria, E., Pereira López, M. D. L. P. Productivity Growth in Mexico: Understanding Main Dynamics and Key Drivers (pp 49 – 71). The World Bank Group, 2021. International Labour Organization-ILO (2020). Global Wage Report 2020-2021: Wages and minimum wages in the time of COVID-19. Jovanovic, B. (1982). Selection and the Evolution of Industry. Econometrica, 50(3), 649-670. Krueger, A. O. "The missing middle. Economic Reform in India: Challenges." Prospects, and Lessons (2013): 299. 37 Lee, D. S. "Wage inequality in the United States during the 1980s: Rising dispersion or falling minimum wage?." The Quarterly Journal of Economics 114.3 (1999): 977-1023. Levinsohn, J., & Petrin, A. (2003). Estimating production functions using inputs to control for unobservables. The review of economic studies, 70(2), 317-341. Levy Algazi, Santiago. "Under-rewarded efforts: The elusive quest for prosperity in Mexico." IDB Publications (Books) (2018). Leyva, Gustavo, and Carlos Urrutia. "Informality, labor regulation, and the business cycle." Journal of International Economics 126 (2020): 103340. Leyva, Gustavo, and Carlos Urrutia. "Informal labor markets in times of pandemic." Review of Economic Dynamics 47 (2023): 158-185. Lochner, B., & Schulz, B. (2022). “Firm productivity, wages, and sorting.” Journal of Labor Economics. López, A., Álava, T. y Marriot, X. (2003). La economía sumergida en el Ecuador: tamaño, causas y consecuencias. Escuela Politécnica del Litoral, 5-77. Manning, Alan. "Monopsony in labor markets: A review." ILR Review 74.1 (2021): 3-26. Mejía Chavez, M., Mejía Chavez, L., Mejía Morales, M., & Lara Haro, D. (2023). ¿Es Ecuador una economía informal? Aproximación al tamaño del sector informal ecuatoriano desde la perspectiva del análisis multivariado. Melitz, M. J., & Polanec, S. (2015). Dynamic Olley‐Pakes productivity decomposition with entry and exit. The Rand Journal of Economics, 46(2), 362-375. Midrigan, V., & Xu, D. Y. (2014). Finance and misallocation: Evidence from plant-level data. American economic review, 104(2), 422-458. Olley, G., & Pakes, A. (1996). The Dynamics of Productivity in the Telecommunications Equipment Industry. Econometrica, 64(6), 1263-1297. Partow, Z, & Vijil, M. Country Private Sector Diagnostic: Creating Markets in Ecuador - Fostering a Dynamic and Resilient Private Sector. Country Private Sector Diagnostic (CPSD) Washington, D.C.: World Bank Group. Restuccia, D., & Rogerson, R. (2008). Policy distortions and aggregate productivity with heterogeneous establishments. Review of Economic dynamics, 11(4), 707-720. Restuccia, D., & Rogerson, R. (2013). Misallocation and productivity. Review of Economic Dynamics, 16(1), 1-10. Restuccia, D., & Rogerson, R. (2017). The causes and costs of misallocation. Journal of Economic Perspectives, 31(3), 151-74. Ruesga, S., Pérez, L. y Delgado, J. (2020). Estimación del tamaño de la economía sumergida en Ecuador a través del Modelo MIMIC. Revista Espacios, 41(12), 16-29. Van Biesebroeck, Johannes. How tight is the link between wages and productivity?: a survey of the literature. ILO, 2014. Vijil, M (2021). Determinantes del Crecimiento de la Demanda Laboral y de la Productividad del Sector Privado. World Bank. Wooldridge, J. M. (2009). On estimating firm-level production functions using proxy variables to control for unobservables. Economics letters, 104(3), 112-114. 38 Appendix Appendix 1: Empirical Methodologies Aggregate Productivity Growth Decomposition To decompose aggregate productivity and its growth, this paper follows the Olley and Pakes (1996) and Melitz and Polanec (2015) approaches, which are summarized here. Sector ′ productivity, Ω , , is defined as the weighted average of firm revenue TFP: , (9) , = , , , where s , is the share of firm ’s value-added in total sector value-added, ω , is the revenue TFP of firm , and N , is the total number of firms in sector and in year . As in Olley and Pakes, sector productivity, Ω , , can be decomposed into two terms: Ω, = , + , . (10) The first term in Equation (10) captures the level of efficiency with which firms in sector transform factor inputs into output, i.e., the technical efficiency of sector . It is measured as the unweighted mean of firms’ revenue TFP within the sector, , = ∑ , , . The second component of Equation (10) measures the , extent to which more productive firms capture a larger share of the sector's economic activity, i.e., the allocative efficiency of sector , expressed as the covariance between firms’ revenue TFP and the share of their value-added in the overall value added of sector , . = ∑ , , − , , − ̅ , . Higher values of this term indicate greater allocative efficiency within the sector. Last, the economy’s aggregate productivity is defined as the weighted average of sector-level productivities: Ω = , Ω, , (11) where λ , is the share of sector ′ value-added in the total economy’s value-added. Building on the approach developed by Melitz and Polanec, aggregate productivity growth, ∆Ω = Ω − Ω , can be decomposed into five components: the within, between, entry, exit, and structural transformation components. For this decomposition, firms within a sector are classified into three groups. The first group of firms in sector corresponds to survivors, which are firms that operate in both years − 1 and . The set of survivor firms is denoted as , = , ∩ , , where , = 1, … . . , , is the set of all firms in sector and in year . The second group of firms in sector is comprised of enterprises, which did not operate in year − 1, but operated in year , the entrant firms. This set of entrants is defined as , = ∈ , and ∉ , . Last, the third group of firms in sector are exiters, which operated in year − 1, but did not operate in year . The set of exiters is given by , = ∉ , and ∈ , . Using these sets of firms, sector ’s productivity growth, ∆Ω , = Ω , − Ω , , can be decomposed as: ∆Ω, = ∆ , + ∆ , + , + , . (12) where the first element of Equation (12) captures changes in the technical efficiency of survivor firms in sector , measured as the difference in the simple average of survivor firms’ productivity in the sector: ∆ , = , − , . 39 where , = ∑∈ , , and , = ∑∈ , , . The second element of Equation , , (12) measures shifts of value-added market shares, within sector , between firms that operated in both periods by differencing the covariance between firms’ productivity and the share of firms’ value-added share in sector ’s value-added at times and − 1: ∆ , = , − , , − ̅ , − , − , , − ̅ , ∈ , ∈ , where ̅ , = ∑∈ , , and ̅ , = ∑∈ , , . The third element of Equation (12) , , captures changes in sector aggregate productivity arising from the entrance of new firms. It is calculated as the market shares of sector ’s entrants times the difference between the weighted sum of firm-level productivities of sector ’s entrants, Ω , = ∑ ∈ , s , ω , , and the weighted sum of firm-level productivities of sector 's survivors, Ω , = ∑ ∈ , s , ω , , in period : , = , Ω, −Ω, . ∈ , The last element of Equation (12) quantifies the contribution of exiters to sector aggregate productivity growth and is measured as the market share of sector ’s exiter firms times the difference between the weighted sum of firm-level productivities of sector ’s survivors, Ω , = ∑ ∈ , s , ω , , and the weighted sum of firm-level productivities of sector ’s exiters, Ω , = ∑ ∈ , s , ω , , in period − 1: , = , Ω, −Ω , . ∈ , This decomposition of sector productivity growth is then used to define the decomposition of economy- wide aggregate productivity growth, ∆Ω . Using Equation (11), aggregate productivity growth can be expressed as: ∆Ω = ,Ω, − ,−1 Ω,−1 = , ∆Ω , + ∆ , Ω , . Replacing ∆Ω , with the expression of Equation (12) into the equation above, aggregate productivity growth is characterized as in Equation (2): ∆Ω = , ∆ + , ∆ + , , + , , + ∆, Ω, . (2) , , The within component, ∑ λ , ∆ , , captures aggregate productivity changes resulting from changes in the technical efficiency of survivor firms. The between component, ∑ λ , ∆ , , accounts for changes in aggregate productivity due to resource reallocation among surviving firms. The entry component, ∑ λ , , , reflects aggregate productivity changes driven by the entry of new firms in the market in period . Last, the exit component, ∑ λ , , , corresponds to aggregate productivity changes resulting from the exit of firms between period − 1 and . These first four terms of Equation (2) capture changes in aggregate productivity driven by firm dynamics that occur within sectors, as they are weighted averages of sector-level measures. The last term of Equation (2) is the structural transformation component, 40 ∑ ∆λ , Ω , , and reflects changes in aggregate productivity driven by the reallocation of economic activity across sectors. Job Flows Decomposition To decompose aggregate employment flows, this paper uses standard methodologies developed by Davis et al. (1996), Decker et al. (2014), and Haltiwanger et al. (2016). The decomposition of aggregate changes in employment between year and year − 1, is given by Equation (4): ∆ = ∆ + ∆ + ∆ + ∆ . (4) ∆ ∆ Equation (4) consists of four components, two of which represent the extensive margin of aggregate job flows, ∆ , and two which make up the intensive margin, ∆ . The extensive margin components reflect aggregate employment changes resulting from job creation by new firms (∆ ) or job destruction by exiting firms (∆ ). The intensive margin components account for changes driven by continuing firms expanding their labor demand ( ∆ ) or contracting it (∆ ). To measure the different components of Equation (4) between two consecutive periods, year − 1 and year , firms, denoted by , are categorized into three groups. The first group of firms is comprised of enterprises, which did not operate in year − 1, but operated in year , the entrant firms. This set of entrants is defined as = { ∈ and ∉ }, where = {1, … . . , } is the set of all firms in year . The second group of firms are exiters, which operated in year − 1, but did not operate in year . The set of exiters is given by = { ∉ and ∈ }. The third group of firms corresponds to continuing firms which are firms that operated in both years − 1 and . Formally, the set of firms that operated in both periods is defined as = ∩ . Given these sets, the first component of Equation (4), which corresponds to the employment created by entrant firms between years − 1 and , is calculated as: ∆ = , , ∈ where , is firm ’s employment in year . The second element of Equation (4) measures the employment destroyed by exiting firms between years − 1 and : ∆ =− , . ∈ The third component of Equation (4) aggregates the employment change of firms that were present in years and − 1 and had a positive variation in their number of workers between these two years, i.e., the expanding continuing firms: ∆ = , − , ∙1 , , . ∈ Last, the fourth component of Equation (4) aggregates the employment variation of firms that were present in years and − 1 and reported a negative change in their employment between these two years, i.e., the contracting continuing firms: ∆ = , − , ∙1 , , . ∈ 41 Appendix 2: Figures Figure A 1: Box Plots of Revenue Productivity: Survivor vs. Exiter Firms (2012 – 2020) Source: Author’s Elaboration, INEC. Note: This figure studies the distributions of revenue productivity for the 2012 – 2020 period. In this figure, we plot the distributions of survivors and exiters for 2013, 2016, and 2019 to analyze changes across time for these distributions. Given that exiters are defined with respect to period t+1, we are unable to report the distribution for 2020, as we cannot yet observe the firms that exited between 2020 and 2021. Figure A 2: Box Plots of Revenue Productivity: Incumbents vs. Entrants (2012 – 2020) Source: Authors’ Elaboration, INEC. Note: This figure studies the distributions of revenue productivity for the 2012 – 2020 period. In this figure, we plot the distributions of incumbents and entrants for 2013, 2016, 2019, and 2020 to analyze changes across time for these distributions. 42 Figure A 3: Life Cycle Trends of Firms in Ecuador: Surviving Sample Source: INEC. Authors’ Elaboration. Note: This figure studies the life cycle patterns of firms in Ecuador for the 2012 – 2020 period. It only considers those observations that were present in the sample for the entire period of analysis, i.e., the surviving sample. To plot the graphs in this Figure, observations are pooled across all years, and then the P10, median, mean, and P90 are identified for each variable (TFP, labor productivity, sales, capital, employment, and wage bill) for five age groups in our sample. The five age groups are: 1) less than five years of age, 2) between five and nine years of age, 3) between 10 and 14 years of age, 4) between 15 and 19 years of age, and 5) 20 or more years of age. 43 Figure A 4: Firing Costs: Ecuador vs. South American Peers Source: Author’s Elaboration, World Bank’s Employing Workers Database and Deloitte. Note: The left panel of this figure reports the severance pay for redundancy dismissal in weeks of salary. It is the simple average for workers of 1, 5, and 10 years of tenure. Bolivia and Venezuela are excluded as data is not reported withing the World Bank’s Employing Workers Database. The right panel of this figure illustrates the firing cost schedules for Ecuador and three neighboring countries (Chile, Colombia, and Peru) for a worker earning the equivalent of the minimum wage in Ecuador in 2019, which was USD 394. Firing cost regulations vary among these countries based on the type of worker contract. In the case of Ecuador and Chile, the same firing cost schedule applies to all types of worker contracts. However, for Colombia and Peru, the schedules depicted in the figure only pertain to workers with open-ended contracts. Workers with fixed-term contracts typically have firing cost structures that depend on the remaining time left in their contracts. Figure A 5: Ratio of Minimum Wage Relative to GDP: Ecuador vs. South American Peers Source: Author’s Elaboration, WDI and ILO’s Global Wage Report 2020-2021. Note: This figure reports the ratio of countries’ minimum wage relative to GDP per capita for 2019. Chile and Venezuela are excluded, as data on minimum wages is not available in the ILO’s Global Wage Report 2020 – 2021. Figure A 6: Business Loan Interest Rates in Ecuador Source: Central Bank of Ecuador and WDI. Authors’ Elaboration. Note: This figure reports the business loan interest rates for Ecuadorian firms. Rates vary according to the level of firms’ sales. The corporate interest rate applies to firms with sales above USD 5 million. The business interest rate corresponds to firms with sales between USD 1 and USD 5 million. The SMEs interest rate corresponds to firms with sales between USD 100,000 and USD 1 million. The Microcredit – Expanded Accumulation interest rate corresponds to firms with sales between USD 20,000 and USD 100,000. The Microcredit – Simple Accumulation interest rate corresponds to firms with sales between USD 5,000 and USD 20,000. The Microcredit – Retail interest rate corresponds to firms with sales less than USD 5,000. The US Bank Prime Loan interest rate serves as a comparator between business interest rates in Ecuador relative to the short-term loan interest rate for businesses in the US. 44 Figure A 7: Collateral Constraints: Ecuador vs. South American Peers (2017) Source: WBES. Authors’ Elaboration. Note: This figure reports the portion of loans requiring collateral and the value of collateral needed for a loan (as a percentage of the loan) for Ecuador and its South American peers. Brazil, Chile, and Venezuela are excluded as their latest WB Enterprise Surveys correspond to 2010, while the surveys for the countries considered in the figure correspond to 2017. Figure A 8: Access to Credit by Firm Size Source: WBES. Authors’ Elaboration. Figure A 9: Innovation Drivers by Firm Size Source: WBES. Authors’ Elaboration. Note: The left panel of this figure considers the average years of managerial experience across firms by firm size group. The right panel of this figure corresponds to the share of firms that carried out research expenditures by firm size groups. The firm size groups considered are: small: firms with 5 to 19 workers, medium: firms with 20 to 99 workers, and large: firms with more than 100 workers. 45 Figure A 10: Productivity Sorting of Workers Across Firms Source: Authors’ Elaboration, INEC. Note: This figure reports firms’ average share of high-skilled workers by revenue TFP quintile. Figure A 11: Distributions of Zombie firms vs. Non-zombie firms All Firms Firms with Access to Finance Source: INEC. Authors’ Elaboration. Figure A 12: Share of Zombie Firms All Firms Firms with Access to Finance Source: INEC. Authors’ Elaboration. Note: This figure shows the share of zombie firms in relation to the total number of formal firms (left panel) and firms with access to credit (right panel). A firm is considered a zombie firm if it is more than ten years old and has an interest coverage ratio less than one for more than three years. The interest coverage ratio is the ratio of operating income (gross income or sales minus operating expenses) to interest payments. 46 Figure A 13: Distribution of Firms and Employment by Firm Size in the USA (2012 – 2020) Source: Business Employment Dynamics, US Bureau of Labor Statistics. Author’s Elaboration. Note: This figure studies the distribution of firms and employment by firm size in the United States. Figure A 14: Distribution of Sales by Firm Size (2012 – 2020) Source: Author’s Elaboration, INEC. Note: This figure studies the distribution of employment by firm size for the 2012 – 2020 period. Figure A 15: Distribution of Sales by Firm Age (2012 – 2020) Source: Author’s Elaboration, INEC. Note: This figure studies the distribution of sales by firm age for the 2012 – 2020 period. 47 Figure A 16: Evolution of Wage Inequality in Ecuador Source: INEC. Authors’ Elaboration. Figure A 17: Decomposition of Between-firm Wage Premia by Industry Source: Authors’ Elaboration, INEC. Figure A 18: Wage Sorting of Workers Across Firms Source: Authors’ Elaboration, INEC. Note: This figure reports firms’ average share of high-skilled workers by quintiles of firm wage. 48