Policy Research Working Paper 10820 What Happens When the State Is Bossing around Markets? An Analysis of the Performance Differentials between Businesses of the State (BOS) and Private-Owned Enterprises (POEs) Dennis Sanchez-Navarro Finance, Competitiveness and Innovation Global Practice June 2024 Policy Research Working Paper 10820 Abstract This paper studies the performance differentials between degree of separation, and type of sector. Businesses of the privately-owned enterprises and businesses of the state. state appear to be more financially leveraged vis-à-vis their Businesses of the State (BOS) are firms with 10 percent private counterparts, suggesting potential soft budget con- or more direct or indirect state participation. By analyz- straints. Wider differentials in profitability and return on ing firm-level data across 16 European countries between investments are evidenced when the state operates in fully 2011 and 2020, the paper finds evidence that state own- competitive sectors that are viable for private participation, ership matters for operational and financial performance underscoring the opportunity costs of state ownership in and sheds light on how and when it matters. The analysis those sectors. Furthermore, the findings show that mixed disentangles the multiple forms of state participation and its ownership with the private sector can drive better results effects on firms’ performance by exploring the heterogeneity when compared to fully owned businesses of the state. Sim- across sector type, levels of state participation, and degree ilarly, a higher degree of separation from the state seems to of separation (direct versus indirect shareholding). It also improve performance, highlighting the role of corporate analyzes the early response of businesses of the state to the governance and ownership reforms to foster indepen- COVID-19 shock. The results suggest that businesses of the dence. Finally, businesses of the state demonstrated greater state underperform in terms of labor productivity, profit- resilience in preserving jobs in the short term during the ability, and return on investments, although the effects are COVID-19 pandemic in 2020. heterogeneous depending on the level of state participation, This paper is a product of 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 author may be contacted at dsancheznavarro@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 What Happens When the State Is Bossing around Markets? An Analysis of the Performance Differentials between Businesses of the State (BOS) and Private-Owned Enterprises (POEs). Dennis Sanchez-Navarro1 World Bank, Washington, D.C., USA JEL: L32, L38, L21, L25, D22. Keywords: SOEs, public enterprises, state ownership, firm performance, competition, private investment. 1 Email: dsancheznavarro@worldbank.org. The author wants to thank Juan Felipe Rodrigo, Carla Scarlato, Regina Onglao, Francis Ratsimbazafy for their research assistance as well as comments and suggestions received from Martha Martinez Licetti, Denisse Pierola, Xavier Cirera, Mary Hallward-Driemeier, Mariem Malouche, Birgit Hansl, Fausto Patino-Pena in the World Bank and professor Ufuk Akcigit (University of Chicago). The author also thanks for the comments received as part of the academic workshops conducted as part of the elaboration of the WB Business of the State (BOS) flagship. The author also wants to thank to the peer reviewers Dorothe Singer and Yuheng Ding for their valuable comments and suggestions, which are reflected in this final version of the paper. This paper is linked to the programmatic ASA State Footprint in markets: database and indicators for SOE reforms (P179791). INTRODUCTION Despite decades of privatization and reforms to rationalize the role of the state in the economy, the state remains as a large and significant market player in the global economy. Firms with 10% or more state participation account for a significant share of the economic activity worldwide with an average of up to 17% of the GDP in developing economies (World Bank, 2023). State-owned enterprises mobilize 20% of the global investment and drive 11% of the global FDI inflows (WBG- IEG, 2018). Although state ownership in markets is not new, evidence suggests that economic shocks such as the financial crisis of 2007-2008 and more recently the COVID-19 pandemic can trigger the rise of state ownership and reorientation of state property policies (Szanyi, 2016) (OECD, 2020). This paper explores whether state ownership matters for firm operational and financial performance and to what degree the level of private participation (e.g., minority vs. full state ownership), the type of sector of operation (e.g., competitive activities vis-à-vis natural monopolies), and proximity to the government (e.g., direct vs. indirect shareholding) can explain potential performance gaps, aiming to shed light on policy implications for future reforms. For this purpose, this paper compares the performance between privately-owned enterprises (POEs) and the Businesses of the State (BOSs) on two main dimensions. First, it measures operational performance through labor productivity. Second, it evaluates financial performance through indicators of profitability, liquidity, and return on investments. Businesses of State (BOS) are defined as legally separated entities, engaged in market production in which either national or subnational governments hold 10% or more participation through direct and indirect ownership (Dall'Olio, et al., 2022). By using this broader definition, we are able to capture a wider set of entities in which the state can have a certain degree of influence and go beyond the conventional State-Owned Enterprises (SOEs) approach, which is often associated to majority and directly owned firms. This allows us to explore how state ownership matters for operational and financial performance even through minority and indirect participation. While state ownership has been widely studied to understand its effects on firm performance, surprisingly most studies follow either a country-specific or a sectoral approach and focus on directly and majority-owned firms. Nonetheless, these studies tend to overlook the ability of the state to influence firms’ performance through other channels including indirect shareholding and minority participation (World Bank, 2023). Moreover, not all markets in which the state operates are the same. The state can serve competitive sectors that are viable for private provision and exhibit low entry barriers such as the manufacturing of food or textiles, in partially contestable sectors featured by market failures (e.g., externalities) such as air transportation, or natural monopolies such as energy transmission. Yet, there is a significant gap in the literature regarding the role of state ownership in competitive sectors and how heterogeneous the effect is when operating in areas that are often served by the private sector.2 2 Although the state has different channels to intervene in the markets beyond shareholding, including regulation and support packages, this paper studies only the ownership channel in line with the analytical approach of Le (2021). 2 The contribution of this paper to the existing literature is fivefold. First, it offers a cross-country approach and employs a harmonized definition that ensures comparability using the new World Bank’s Businesses of the State (BOS) database (Dall'Olio, et al., 2022). Second, it studies the differences in firm performance by the type of sector and potential opportunity costs of state ownership in sectors where the economic rationale is less clear by implementing the World Bank BOS sector taxonomy (Dall'Olio, et al., 2022b). Thus, we differentiate the role of state ownership in competitive sectors such as the manufacturing of textiles or accommodation services vis-à-vis natural monopolies (e.g., electricity transmission) and partially contestable sectors (e.g., extraction of gas). Third, this study shows how performance differential varies by the level of state participation and how mixed and private ownership can explain better financial and operational outcomes vis-à-vis cases where the state is the sole owner. Fourth, this paper unveils performance differentials between directly and indirectly owned BOS and explores to what extent a higher degree of separation or autonomy from the state can drive better operational and financial results. Finally, we explore how POEs and BOSs respond to economic shocks in the short-term (e.g., COVID-19). The remainder of the paper is organized as follows. Section 1 discusses the analytical framework of principal-agent theory to understand the role of state ownership in firms’ operational and financial performance. Section 2 presents the review of the existing literature. Section 3 presents the data sources followed by some descriptive statistics in section 4. Section 5 describes the empirical methodology. Our empirical results are presented and examined in Section 6, followed by Section 7, which discusses policy implications and concludes. 1. PRINCIPAL-AGENT THEORY TO UNDERSTAND THE ROLE STATE OWNERSHIP IN FIRMS’ PERFORMANCE The principal-agent problem proposed by Jensen & Meckling (1976) underscores potential issues that can arise when one or more principals engage an agent to perform a service on their behalf. The delegation of decision-making authority from the principal (owner of the enterprise) to the agent (enterprise's management) creates an agency problem due to differing interests between the two parties. Discrepancies in incentives can emerge as the agent is motivated by utility maximization, which may not always align with the best interests of the principal (Jensen & Meckling, 1976). This divergence in interests can result in poor performance and suboptimal decision-making in enterprises. While both BOSs and POEs can experience these issues, the literature suggests that firms with state participation are more vulnerable to be captured by certain interest groups, notably political parties (Boycko, Shleifer, & Vishny, 1996); (Campos & Esfahani, 1996). Some issues might stem from the appointment of board members, who can relate to political parties or decision-making that serves the interests of certain politically connected firms. Furthermore, appointments linked to political patronage can cause financial costs related to severance payments, loss of experienced personnel, discontinuation of projects, and sudden changes in management strategies, and investment plans (Totleben & Kardziejonek, 2019). These factors can result in lower financial results, reduced competitiveness, and uncertainty for the firm and its partners. 3 To address this issue, the principal can limit deviations from their interests by providing suitable incentives for the agent and incurring monitoring costs that aim to curb rent-seeking behavior. This includes potential solutions through corporate governance mechanisms that can reduce agency costs and ownership problems (McColgan, 2001; Hermuningsih et al., 2020). Beyond corporate governance, some authors suggest it is important to bring in the private sector through mixed ownership that can help to mitigate principal-agent problems by strengthening supervisory power and weakening the executive’s control power (Shang, Yuan, Li, & Fan, 2022). Through mixed ownership, non-state shareholders can restrain and supervise the behavior of state shareholders reducing the discretion of executives, which can ultimately impact firm performance. Recent assessments suggest that private shareholding can help reduce the power of the chair and improve internal governance, particularly in monopoly sectors as competitive sectors are more exposed to market forces (Guan, et al, 2021; Lazzarini & Mussacchio, 2018). 2. LITERATURE REVIEW When assessing firm operational and financial performance, one key assumption is that firms have profit-maximizing objectives that guide their strategic decisions. However, this might not be the case when governments act as market players to achieve non-economic objectives (Andrews & Dowling, 1998). For instance, state ownership is often justified as a policy strategy to control natural resources, provide employment, and develop specific economic sectors or regions (Grout & Stevens, 2003 ). Since the late 1970s, both theoretical and empirical studies have assessed the role of state ownership in firm performance. Early studies discussed the effect of state ownership on performance using variables such as the costs or price of goods and services to compare SOEs and POEs (Davies, 1977) (Bennett & Johnson, 1979).3 Although there seems to be consensus and increasing evidence on the significant effect of the state ownership over firm’s performance, there is no consensus on the direction of such effects. One strand of research suggests that state participation may improve firm performance as it broadens business opportunities and increases financial leverage. Some authors argue that state participation has a signaling effect indicating that the firm is backed up by the government, reducing information asymmetries, and providing comparative advantages. These can translate into access to finance (Nguyen, Do, & Le, 2021), favorable interest rates (Dewenter & Malatesta, 2001), access to essential infrastructure, direct subsidies, or even preferential regulatory treatment (Le T. , 2021). Some authors propose state participation can increase profitability in sectors in which state dominance is high and competition is low (Liljeblom, Maury, & Hörhammer, 2020) or in case the state holds a golden share (Kočenda & Svejnar, 2003). However, as highlighted by (Le, & Buck, 2011), it is not clear whether the positive relationship between state shareholding and firm performance is driven by efficiency or power given the benefits of government support and political connections (Yu, 2013) (Le & Buck, 2009). 3 For instance, Bennett and Johnson (1979) found that private garbage collection enterprises were more efficient than their public counterparts by comparing the prices of private and public garbage collection enterprises in jurisdictions where private and public actors coexisted. (Davies, 1977) analysis found that the Australian private transcontinental airline had higher productivity -measured by mail, passenger, and revenues per employee- than the Australian state-owned airline. 4 Another strand of the literature sustains that firms with state participation underperform vis-à-vis private peers. It argues that state ownership poses important challenges for a firm performance, particularly when measured in terms of profitability as it is often paired with social mandates such as employment protection and provision of public goods, which are not necessarily aligned with profit maximization behavior (Sun, Zhang, & Li, 2005). Evidence suggests firms with state ownership are more financially leveraged, more labor-intensive, and less profitable (Dewenter & Malatesta, 2001). Furthermore, several studies find that state participation can impact performance as the state can influence strategies resulting in less risk-averse behavior, which ultimately impacts profitability, returns to capital (Dollar & Wei, 2007), R&D investments, and internationalization plans (Tihanyi, et al., 2019). Nonetheless, most empirical studies are country-specific. A vast set of studies are found on China (Dollar & Wei, 2007), (Chun, 2009), (Lu, Siu, Au, & Leung, 2009), (Amighini, Rabellotti, & Sanfilippo, 2013), (Shahab, Ntim, & Ullah, 2019), (Chen, Xie, & Van Essen, 2021). Additional studies that find a negative relationship between state ownership and firm performance are found on the Russian Federation (Abramov et al. 2017) (Liljeblom, Maury, & Hörhammer, 2020) and Malaysia (Isa & Lee, 2016), highlighting decreasing returns on assets in Greece (Halkos, 2002), and lower productivity in Türkiye (Zaim & Taskin, 1997), among others. Other studies have approached a similar question through a sector-specific focus analyzing the role of state ownership in the financial sector (Cornett, Guo, Khaksari, & Tehranian, 2010), state- owned commercial banks (Panizza, 2021), natural monopolies or regulated oligopolies (Vining & Boardman, 1992), and network sectors (Bogart & Chaudhary, 2015). Fewer studies provide a cross-country and cross-sector analysis, especially in competitive sectors, which are particularly relevant for private-sector development. Some studies on competitive sectors include Funkhouser & MacAvoy (1979), Boardman & Vining (1989), and Picot & Kaulmann (1989). Some recent studies with firm-level data across countries include Dewenter & Malatesta (2001), Kabaciński, Kubiak, & Szarzec (2020), Phi, et al., (2019), IMF (2019) and Le (2021). 3. DATA SOURCES The data employed for the analysis corresponds to a combination of two firm-level datasets: (i) the World Bank Businesses of the State (BOS) database (Dall'Olio, et al., 2022), and (ii) the Bureau van Dijk (BvD) ORBIS financial module. 3.1 World Bank Businesses of the State (BOS) database First, we use the World Bank Global Businesses of the State (BOS) database that provides firm- level information on companies in which the state holds 10 percent or more direct and indirect participation. Although there is no academic consensus on the definition of state-owned enterprises (SOEs), they are often associated with directly and fully or majority-owned firms by the state.4 In this paper, we cover a much broader set of entities to capture other channels of state ownership including minority and indirect ownership following the WB BOS definition (Dall'Olio, et al., 4The SOE definition can vary across countries depending on the different legal forms (e.g., corporatized vs. non-corporatized, level of participation (%)), and local definition of control. 5 2022). By doing so, we can study how state ownership and its different shapes and forms can potentially influence firm decisions even without having full control and measure how this can affect firm performance. The WB BOS database provides information on entities with state participation across more than 90 countries. It collects firm-level data including characteristics such as age, sector of operation, and size, as well as financial indicators of revenues and profits. It also provides detailed ownership information to differentiate between directly and indirectly owned companies, the level of state participation, and the degree of proximity with the state (i.e., layers of the ownership tree). Following Abramov, Radygin, & Chernova (2017), we differentiate between firms directly owned by the state vis-à-vis those in which the state owns shares through other organizations or chains of firms (i.e., indirectly owned). For the latter, the BOS database informs how many firms or layers connect a firm with the state.5 Finally, the WB BOS database provides information on the level of state participation to differentiate by minority shareholding (10-24.9%), blocking minority (25- 49.9%), majority shareholding (50-99.9%), and full state participation (Dall'Olio, et al., 2022).6 3.2 ORBIS financial module - Bureau van Dijk (BvD) The second source of information is the raw financial module of ORBIS compiled by the Bureau van Dijk Electronic Publishing (BvD). The raw data was processed and cleaned implementing the protocol in the literature following Kalemli-Ozcan et al. (2015) and Cusolito (2020). This dataset provides a rich set of financial indicators for both privately and government owned entities. However, it does not have an indicator to differentiate whether a firm is a BOSs or a private firm.7 To identify which firms in the ORBIS financial module can be defined as Businesses of the State (BOS), we merge the WB BOS database with the ORBIS financial module. BOSs are defined as firms with 10% or more state direct or indirect participation, whereas POEs are defined as those with no state participation or state shareholding below 10%. This paper focuses on European countries in which there is consensus in the literature on having a higher coverage in the ORBIS financial data despite the caveats of completeness of information (Kalemli-Ozcan et al., 2015), (Bajgar et al., 2020), (Cusolito, et al., 2022). We cover a subset of 16 European countries over the period 2011-2020 for which we have simultaneously i) ORBIS firm-level data, ii) the WB BOS database for the specific country, and iii) representativeness of at least 60% of the national employment in the financial module in ORBIS.8 We compare the performance of BOSs and POEs through a multivariate approach following Kabaciński, Kubiak, & Szarzec (2020) and Chan, Chen, & Wong (2018) analyzing operational performance measured by labor productivity and three aspects of financial performance: profitability, financial liquidity, and return on investments. We measure labor productivity as operating revenues per worker. Financial performance is assessed through financial liquidity using 5 For instance, C is considered as a BOS through 3 layers of ownership in the case where the Ministry of Finance owns (by 10% or more) a company A, which owns company B (by 10% or more), which also owns company C. 6 The WB BOS database does not provide information on golden shares or veto powers, so the level of participation and thresholds proposed are used as a proxy of control and influence of the state over the firms. 7 The ORBIS dataset does not provide a variable to directly identify those with state participation Although a proxy variable in ORBIS denoted as the Global Ultimate Owner can serve to identify some companies with state shareholding, it provides an underestimation of the real footprint of the state in the markets and suffer from omission errors that could misclassify SOEs as private firms (Dall'Olio, et al., 2022). 8 Yet, there is also variation of coverage within these countries as documented by (Bajgar, Berlingieri, Calligaris, Criscuolo, & Timmis, 2020). 6 the current ratio and liquidity ratio. These indicators provide an estimate of the firm capability to meet its short-term and long-term obligations, respectively. Additionally, we assess profitability in terms of profit margins and return on investments using the return on assets (ROA) and returns on equity (ROE). 4. DESCRIPTIVE STATISTICS Table 1 summarizes the sample distribution over time. Our sample covers approximately 2 million firms across 16 European countries between 2011 and 20209 operating across sectors such as agriculture, manufacturing, construction, to wholesale and trade services. Nonetheless, the sample excludes public administration, health, education and activities of households, and membership organizations following the WB BOS approach.10 Our sample includes more than 27,000 BOS firms, which on average account for approximately 1.4% of the total number of firms in the sample. These firms with state participation contribute to 14% of the total revenues and 15% of the total employment (Table 1). Further details to document variations across countries are described in Annex 1.11 Table 1. Distribution of the sample by firm type and year Firm 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Average type Number of firms POEs 98.3% 98.4% 98.5% 98.6% 98.7% 98.7% 98.6% 98.7% 98.8% 98.8% 98.6% BOSs 1.7% 1.6% 1.5% 1.4% 1.3% 1.3% 1.4% 1.3% 1.2% 1.2% 1.4% Revenues POEs 84.9% 84.3% 84.8% 85.7% 86.3% 86.6% 86.4% 86.2% 87.3% 87.4% 86.0% BOSs 15.1% 15.7% 15.2% 14.3% 13.7% 13.4% 13.6% 13.8% 12.7% 12.6% 14.0% Employment POEs 83.3% 84.0% 84.3% 85.0% 86.5% 87.5% 84.2% 84.7% 86.1% 85.7% 85.1% BOSs 16.7% 16.0% 15.7% 15.0% 13.5% 12.5% 15.8% 15.3% 13.9% 14.3% 15.0% Source: Author’s calculations using WB BOS database and ORBIS. Table 2 describes some summary statistics by country and shows some indications of performance gaps between BOSs and POEs. BOSs seem to be older and larger than POEs when measured by number of years in the market and number of workers across all countries in the sample. Yet, in most countries BOSs seem to be less profitable than POEs and obtain lower returns of investments -measured by assets and equity. BOSs also seem to achieve lower revenues per worker compared to POEs, except for four countries -Estonia, Spain, Italy, and Portugal. In terms of financial leverage, in about 90 percent of the countries, BOSs have a lower ability to cover short-term and long-term liabilities, measured by the current ratio and solvency ratio, respectively. 9 The total sample includes over 19.9 million observations. 10 The sectors include those covered by the WB BOS database (Dall'Olio, et al., 2022b). 11 The countries covered are Bosnia and Herzegovina, Bulgaria, Estonia, Spain, Croatia, Italy, Lithuania, Latvia, North Macedonia, Poland, Portugal, Romania, the Russian Federation, Serbia, Slovenia, and Ukraine. 7 Table 2. Summary statistics: BOSs vis-à-vis POEs (median representative firm) in sample by country 2011-2020 Country Number of Age Employment Profit margin Labor Current Solvency ROA ROE (ISO) firms (years) Productivity * ratio Ratio (%) POE BOS POE BOS POE BOS POE BOS POE BOS POE BOS POE BOS POE BOS POE BOS BA 6,647 271 14 17 21 50 4.3 0.9 10.6 9.7 1.3 1.3 47.5 55.7 3.3 0.2 10.2 0.5 BG 63,363 344 8 16 18 55 4.5 1.4 9.8 9.3 1.8 1.7 55.4 71.3 3.7 0.7 13.4 1.3 EE 63,343 124 5 16 18 25 11.5 5.2 10.8 10.9 2.2 1.6 92.2 73.8 1.3 2.0 3.1 3.7 ES 288,776 1,324 13 17 18 43 2.9 2.0 11.6 11.6 1.4 1.5 43.3 49.8 0.8 0.6 3.0 1.6 HR 32,309 499 9 14 18 38 4.2 0.9 10.9 10.5 1.3 0.9 34.1 39.9 0.4 0.4 5.7 1.6 IT 401,118 3,448 10 13 17 44 3.8 2.7 11.9 12.1 1.4 1.3 25.6 29.2 0.5 0.5 3.2 2.4 LT 14,549 95 12 21 20 82 3.3 3.7 10.5 10.4 1.7 1.8 50.7 77.2 5.0 1.4 12.2 2.3 LV 19,352 271 8 21 19 46 3.5 0.6 10.4 10.1 1.2 1.3 44.5 43.5 0.7 0.2 8.6 0.5 MK 10,018 126 11 18 19 65 4.5 2.1 10.2 9.5 1.4 1.6 62.1 63.8 2.2 0.5 6.2 0.9 PL 128,157 2,703 8 16 41 94 3.6 2.2 11.3 10.9 1.5 1.3 51.8 65.3 3.4 0.8 11.6 1.5 PT 64,708 406 12 15 18 51 3.3 3.7 11.1 11.5 1.6 1.3 36.1 44.0 1.4 0.8 5.6 3.1 RO 213,884 762 7 10 19 58 11.2 2.0 10.1 9.5 1.2 1.9 41.3 49.6 0.7 1.4 17.8 4.8 RS 33,788 594 10 18 19 78 3.4 2.0 10.5 9.6 1.1 1.0 35.7 54.6 2.1 0.4 12.3 1.3 RU 453,193 13,369 7 14 25 49 4.4 1.7 9.5 9.4 1.4 1.3 39.3 56.2 3.5 0.6 19.7 2.3 SI 50,483 364 7 22 18 62 9.8 1.5 11.5 11.4 2.0 1.4 65.1 48.7 4.9 1.2 14.6 3.0 UA 125,312 2,637 11 15 22 41 2.8 0.4 9.4 8.5 1.2 1.2 54.5 75.2 2.4 0.2 10.4 0.5 Note: *Labor productivity is measured as the logarithm of operating revenues per worker. Source: Author’s calculations using WB BOS database and ORBIS. Some authors argue that firms owned by the state are not always operating under a profit- maximizing behavior as they may pursue public mandates (Matuszak & Kabaciński, 2021). Hence, some performance differentials are expected (Lawson, 1994) (Bozec, Breton, & Coté, 2002). Previous studies have tried to account for commercial and non-commercial mandates assigned to SOEs. However, there are two caveats on former attempts to do so. First, most studies are country-specific hindering external validity and comparability of the findings. Second, the differentiation of the type of sectors to determine whether state-owned firms have commercial and non-commercial functions has been based on profitability, sector-specific regulations, and president announcements (Bozec, Breton, & Coté, 2002). These approaches might raise potential endogeneity issues since the distinction between commercial and non-commercial activities would be linked to underlying performance issues rather than the economic nature of the sector where the state operates. For example, a loss-making BOS could be resulting from inadequate management or insufficient market incentives even when performing fully commercial functions such as manufacturing of textiles. Similarly, using policy regulations and presidential interventions to differentiate between commercial and non-commercial functions could be linked to other interrelated policy objectives that are not necesarilly related to the technology or economic characteristics of a specific sector.12 12 Some companies could be not-profit maximizing as per the legal framework, but still operate under management contracts that set profit targets. In these cases, it is not clear whether not-profit companies could be categorized as non-commercial given their low performance. Hence, some endogeneity issues might emerge. As indicated by (Megginson & Netter, 2001), assessment on performance might itself be linked to the system that includes both political and performance goals. 8 The WB sector taxonomy serves to overcome these limitations.13 It builds on the industrial organization literature to classify the economic activities (Table 3) into competitive, natural monopolies, and partially contestable sectors based on the market failures, technology and intrinsec economic characteristics (e.g., externalities, public goods, etc.). Table 3. Sector taxonomy to differentiate by the degree of contestability of sectors and rationale for state ownership. Sector type Definition Competitive Sectors Economic activities with little to no entry barriers that are commercially viable for multiple firms to operate. Inherent market features, such as cost structure, technology, or demand characteristics, make entry into these sectors largely unproblematic. Some examples are the manufacturing of food and beverages, textiles and apparel, wholesale, and retail trade. In these sectors, the economic rationale for state ownership is less clear as those sectors can be viable and efficiently provided by the private sector. Partially contestable sectors Economic activities are characterized by some form of market power, externalities, or other market failures that could be addressed -although not uniquely- by state ownership. Some examples include the production of electricity, extraction of natural gas, and air passenger transportation services. Natural monopoly sectors Economic activities that exhibit high entry barriers, scale economies, or sub-additivity cost structures (i.e., when a single firm can produce a product at a lower cost than the combined cost of two or more firms producing identical products), thus making them more efficient for just one producer to operate. Some examples include water utilities, energy transmission, and postal activities with universal obligation. Source : (Dall'Olio, et al., 2022b) We implement the WB sector taxonomy to classify the economic activities using the NACE 4- digit code for the 16 European countries studied. In total, 95% of the firms in our sample operate in competitive sectors. Table 4 shows the distribution by sector type between BOSs and POEs by sector type. On average, 67% BOSs are found in competitive sectors for which the economic rationale for state ownership is less clear as can be often served by the private sector. The remaining 17% of BOSs are found in partially contestable sectors such as passenger transportation, and 16% in natural monopolies such as energy transmission. Table 4. Share of POEs and BOSs by sector type and year 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 POEs Competitive 95.8% 95.7% 95.6% 95.6% 95.5% 95.7% 95.6% 95.7% 95.7% 95.7% Natural Monopoly 0.8% 0.8% 0.8% 0.9% 1.0% 0.9% 0.8% 0.8% 0.7% 0.8% Partially Contestable 3.5% 3.5% 3.6% 3.6% 3.5% 3.4% 3.6% 3.6% 3.5% 3.6% BOSs Competitive 68.1% 68.0% 67.9% 68.0% 68.4% 68.1% 67.5% 67.4% 67.6% 67.3% Natural Monopoly 15.3% 15.4% 15.4% 15.6% 15.7% 16.1% 16.7% 16.7% 17.0% 17.1% Partially Contestable 16.6% 16.6% 16.6% 16.4% 15.9% 15.8% 15.9% 15.9% 15.4% 15.6% Source: Author’s calculations for 16 ECA countries 13 This taxonomy offers a harmonized approach to differentiate by the type of sector where SOEs operate independently of the country or political and economic system. For more details see (Dall'Olio, et al. 2022). We used the main sector of operation of the firm using the NACE 4-digit classification and combine it with the sector taxonomy that follows the same sector disaggregation and classification. 9 We argue potential performance gaps between BOSs and POEs depend on the degree of contestability in the sector. We expect BOS performance to be closer to that of private peers in competitive sectors as those firms might be subject to higher entry and competition from private providers and are less likely to fulfill public service obligations in such activities.14 On the contrary, we expect higher differentials in natural monopoly activities that are usually linked to public service obligations (PSOs) such as water, sewage, electricity transmission, and postal services. In partially contestable sectors, important externalities might also require BOSs to fulfill a public mandate such as ensuring connectivity (cargo and passenger) services. In other words, we argue the likelihood of BOSs having a PSO that justifies a potential performance differential is lower for those operating in competitive sectors. We explore this in more detail in the next section. Finally, we differentiate by the level of state participation and degree of separation between the BOS firm and the government. In our sample, 58% of the BOSs are fully owned by the state (100%), but we also observe majority-owned entities that account for 16% of the BOS, as well as minority state shareholdings (Table 5). Minority shareholdings account for about a quarter of the BOS firms in our sample segmented by minority participation (13%) and blocking minority (14%). In our sample, we also observe firms that are directly owned by the government vis-à-vis those owned through another firm or chain of firms (i.e., indirectly owned) and to what degree they can be linked to the state.15 Table 5. Degree of state ownership – WB BOS database Degree of State Definition participation Minority Firms in which the state holds 10-24.9% of the stakes Blocking minority Firms in which the state holds 25-49.9% of the stakes Majority ownership Firms in which the state holds 50-99.9% of the stakes Full ownership Firms in which the state holds 100% of the stakes Direct Firms directly owned by a Public Authority (national or subnational), which holds at least 10% participation. Indirect Firms participated by the government through another firm or chain of firms owned by the State where it holds at least 10% participation. Source: Author’s elaboration based on World Bank (BOS) database and (Dall'Olio, et al., 2022). 5. EMPIRICAL METHODOLOGY This paper explores two main econometric exercises. First, it estimates the conditional correlation of the state participation and the performance differentials between BOSs and POEs. For this purpose, we analyze the role of state ownership on firm performance through the following specification: (1) = + 1 + 2 + + + + Where, denotes the performance indicator for the firm using several performance measurements to have a comprehensive overview of the soundness of BOSs vis-à-vis POEs. First, we assess the operational performance in terms of productivity, followed by the financial performance measured by indicators of profitability, liquidity, and returns on investment. Thus, 14 Alternative government interventions in the form of regulation can be suitable in these sectors to achieve public objectives. 15 This is denoted as the layer of ownership. For instance, a firm linked to the state through another BOS has 2 layers or degree of connection to the state. 10 the outcomes of interest are: i) labor productivity (operating revenues per workers), ii) profit margins, iii) liquidity based on the current ratio (short-term) and solvency ratio (long-term), and iv) return on investments measured by the returns on assets (ROA) and return on equity (ROE). The explanatory variable -- denotes the measurement of the state participation in the firm and is the main parameter of our analysis. The state participation is tested in two ways to measure both the extensive and intensive margins. First, it takes the form of a dichotomic variable that takes the value of 1 when the firm has state participation -either direct or indirect- of at least 10 percent, and 0 otherwise. It serves to measure the state participation as the extensive margin. Second, we include a measurement of the intensive margin by breaking down the state participation in the form of a categorical variable to differentiate between minority-owned companies [10-25%), blocking minority [25%, 50%), majority participation [50%, 100), and fully owned firms by the state [100%]. When interpreting the results, the reference (excluded) category refers to privately owned enterprises (POEs) defined as firms with no state participation or in which the state has a non-controlling participation of less than 9.9%. As explanatory variables, the vector denotes firm-level controls including age, sector, size (measured by total fixed assets and number of workers), and listing status.16 Additionally, , , , refer to the fixed effects by country to remove any country-specific unobserved characteristics such as governance arrangements, sectoral and time-bound shocks. Additionally, we control for potential country, sector, and time-varying shocks. Finally, we include country-sector-year fixed effects, which allow to demean the results, so the coefficients can be interpreted as the distance to the country-sectoral mean and therefore be comparable across countries.17 Furthermore, we explore the heterogeneous response between BOSs and POEs to economic shocks, particularly to the short-term shock of the COVID-19 pandemic. For this purpose, we evaluate both levels and growth of labor force and fixed assets as a proxy of capital investments using the specification below following Brolhato, Cirera, & Martins-Neto (2023) and Ferro & Patiño-Peña (2023): (2) = + 1 + 1 ∗ 2020 + 2 + + + + Thus, we evaluate whether state ownership can play a role as a stabilizer during periods of crisis. As before, the BOSs variable takes value 1 if the state holds 10% or more state participation, and we also delve deeper into whether there are differences in the response to the shock by the level of state participation. 16 Both size and age variables are included as lagged values in t-1. 17 Since the BOS variable from the WB BOS database is time-invariant in the sample period, we cannot control for firm-level fixed effects in the assessment. Hence, these results do not claim causality between state ownership and firm performance. Nonetheless, we perform different robustness checks to test different specifications that can help minimize the effect of unobserved firm-level characteristics. The results hold when using shorter timeframes 5-years and 3-years, and when using the subset of firms fully owned by the state. These exercises are conducted to test the robustness of the results in the event some changes in ownership have occurred in the sample period. Estimations are conducted using OLS. 11 6. RESULTS Our results shed light on the role of state participation in firms’ performance. Although this analysis does not claim causality, the findings provide new evidence on the relationship between state ownership and operational and financial outcomes shedding light on the multi-faceted role of the state in markets. In this section, we explore the role of the state in multiple economic sectors, the level of state participation, degree of separation of the state, among others. The results are structured as follows. First, we explore the extensive margin to assess to what extent state participation matters for firms’ performance in the four dimensions proposed -productivity, profitability, liquidity, and return on investments. Subsequently, we explore the intensive margin to study how these gaps might vary by the degree of state ownership. Finally, we analyze to what extent proximity to the state measured by the degree of separation to the public authority (i.e., direct, or indirect participation) is linked to performance gaps. 6.1 Extensive margin: Does state participation matter for operational and financial performance? The short answer is yes. Evidence suggests there is a performance gap between BOSs and POEs. However, the differences depend on the outcome analyzed, the type of sector in which BOSs operate, and whether there is mixed ownership with the private sector. Operational performance is measured by the labor productivity, whereas financial performance is assessed through the profit margins, current ratio, solvency ratio, ROA and ROE. 6.1.1 Labor productivity In terms of operational performance, results in Table 6 suggest that BOSs achieve, on average, lower labor productivity outcomes compared to private peers. Results in columns (6-9) indicate that BOS are, on average, 32% less productive than the median POEs within the same industry.18 These results are robust and statistically significant when controlling by firm’s size, age, listing status, country, sector of operation, and fixed effects to remove potential country, sector, time varying and time-specific shocks. Table 6. Results for labor productivity Log (Labor productivity) (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS -0.449*** -0.409*** -0.401*** -0.401*** -0.399*** -0.389*** -0.386*** -0.386*** -0.386*** (0.00291) (0.00299) (0.00298) (0.00302) (0.00298) (0.00304) (0.00304) (0.00679) (0.0184) Size (Log Capital) 0.433*** 0.452*** 0.454*** 0.442*** 0.437*** 0.435*** 0.432*** 0.432*** 0.432*** (0.000308) (0.000319) (0.000320) (0.000322) (0.000320) (0.000326) (0.000327) (0.000603) (0.00863) Age -0.0149*** -0.0148*** -0.0151*** -0.0150*** -0.0157*** -0.0156*** -0.0156*** -0.0156*** (4.49e-05) (4.49e-05) (4.43e-05) (4.41e-05) (4.45e-05) (4.45e-05) (9.03e-05) (0.000739) Listed -0.685*** -0.658*** -0.647*** -0.633*** -0.633*** -0.633*** -0.633*** (0.00861) (0.00861) (0.00860) (0.00878) (0.00881) (0.0210) (0.0285) Constant 4.554*** 4.496*** 4.479*** 4.636*** 4.702*** 4.739*** 4.769*** 4.769*** 4.769*** (0.00412) (0.00415) (0.00415) (0.00418) (0.00415) (0.00422) (0.00424) (0.00767) (0.105) 18 Since only the dependent variable is log transformed, the coefficient should be interpreted as ((1 ) − 1) ∗ 100 = (exp(−0.386) − 1) ∗ 100 = −32.02%. Therefore, in companies where SOE=1, the labor productivity is 32% lower compared to POEs. 12 Observations 6,612,011 6,459,551 6,459,551 6,459,551 6,459,534 6,459,468 6,452,778 6,452,778 6,452,778 R-squared 0.685 0.689 0.689 0.697 0.702 0.715 0.718 0.718 0.718 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: Labor Productivity is estimated as the log of total revenues per worker. The reference (excluded category) refers to Privately-owned Enterprises (POEs), so results should be interpreted as compared to private firms within the same sector, size, and age, and sector and controlling by the respective fixed effects. Capital refers to lagged values of the log of total assets of the firm (year t-1), age refers to the lagged number of years in the market in t-1. Listed status is a dichotomous variable that takes value 1 if the company is listed in the stock market, 0 otherwise. Fixed effects are included to absorb for potential shocks including time-bound shocks (year-FE), sectoral shocks (sector-FE), time-varying sectoral shocks (sector-year FE), country specific shocks (country-FE), country time varying shocks (country-year FE), and country, sector specific and time varying shocks (country-sector-year, FE). Clustered standard errors are used to control for potential autocorrelation within firms (column 8 to capture potential serial correlation) and sector (column 9). 6.1.2 Profitability: Profit margin When comparing financial performance, results in Table 7 highlight that BOSs achieve, on average, relatively lower profit margins when compared to private counterparts. After controlling for the firm’ size, age, and listing status, results in column (6-9) indicate that BOS profits are, on average, 6.3 percentage points lower compared to private peers. Listing status, however, does not seem to have a statistically significant effect on performance differentials. These findings are aligned with former empirical analyses that confirm government-owned firms tend to be significantly less profitable than privately owned firms (Kathryn, Dewenter, & Malatesta, 2001) (Phi, Taghizadeh-Hesary, Tu, Yoshino, & Kim, 2019) (Panizza, 2021). Table 7. Results for profit margins Profit margins (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS -6.018*** -6.238*** -6.232*** -6.213*** -6.094*** -6.346*** -6.318*** -6.318*** -6.318*** (0.0548) (0.0561) (0.0562) (0.0561) (0.0560) (0.0570) (0.0572) (0.110) (0.323) Size -0.539*** -0.611*** -0.609*** -0.620*** -0.605*** -0.597*** -0.599*** -0.599*** -0.599*** (0.0071) (0.0075) (0.0075) (0.0075) (0.0075) (0.0076) (0.0076) (0.013) (0.058) Age 0.0281*** 0.0282*** 0.0287*** 0.0270*** 0.0157*** 0.0160*** 0.0160*** 0.0160*** (0.00077) (0.00077) (0.00077) (0.00077) (0.00078) (0.00078) (0.0013) (0.0061) Listing status -0.610*** -0.600*** -0.637*** -0.0725 -0.0301 -0.0301 -0.0301 (0.183) (0.183) (0.182) (0.186) (0.185) (0.356) (0.285) Constant 7.513*** 7.389*** 7.380*** 7.411*** 7.380*** 7.510*** 7.512*** 7.512*** 7.512*** (0.0259) (0.0269) (0.0270) (0.0270) (0.0269) (0.0271) (0.0271) (0.0443) (0.169) Observations 6,500,636 6,354,633 6,354,633 6,354,633 6,354,614 6,354,541 6,347,824 6,347,824 6,347,824 R-squared 0.068 0.064 0.064 0.066 0.074 0.096 0.105 0.105 0.105 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes 13 Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: Profit margin is estimated as the profit before tax over operating revenue (%). The reference (excluded category) refers to Privately-owned Enterprises (POEs), so results should be interpreted as compared to private firms within the same sector, size, and age, and sector and controlling by the respective fixed effects. Capital refers to lagged values of the log of total assets of the firm (year t-1), age refers to the lagged number of years in the market in t-1. Listed status is a dichotomous variable that takes value 1 if the company is listed in the stock market, 0 otherwise. Fixed effects are included to absorb for potential shocks such as temporary shocks (year-FE), sectoral shocks (sector-FE), time-varying sectoral shocks (sector-year FE), country specific shocks (country-FE), country time varying events (country-year FE), and country and sector specific and time varying shocks (country-sector-year, FE). Clustered standard errors are used to control for potential autocorrelation within firms (column 8 to capture potential serial correlation) and sector (column 9). 6.1.3 Liquidity: Current ratio and solvency ratio Next, we analyze the ability to cover short-term and long-term financial obligations. First, by using the current ratio we assess how much the financial obligations due within a year are covered by the current assets. Results in Table 8 show negative and significant differences in terms of liquidity between BOSs and POEs. These results are robust when controlling by firm size, age, fixed effects, and clustered-standard errors and suggest BOSs are more leveraged or assume higher levels of debt compared to private peers. The estimates (columns 7-9) indicate that BOSs have on average 0.35 fewer dollars in assets to cover each dollar in short-term liabilities (due within a year). Listed BOS partially offset the differentials compared to POEs, potentially linked to stronger market discipline and budget constraints (Wang & Colin, 2004). These findings could be explained by potential soft-budget constraints such that BOSs can receive implicit or explicit loan guarantees that allow them to borrow at favorable rates (Dewenter & Malatesta, 2001), receive fiscal means in the form of subsidies or reduction of tax obligations (Kornai, Maskin, & Roland, 2003). Table 8. Results for current ratio Current ratio (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS -0.249*** -0.358*** -0.359*** -0.346*** -0.332*** -0.360*** -0.353*** -0.353*** -0.353*** (0.019) (0.020) (0.020) (0.020) (0.020) (0.021) (0.021) (0.038) (0.058) Size -0.468*** -0.572*** -0.572*** -0.571*** -0.568*** -0.569*** -0.571*** -0.571*** -0.571*** (0.003) (0.003) (0.003) (0.003) (0.003) (0.003) (0.003) (0.005) (0.014) Age 0.0575*** 0.0575*** 0.0579*** 0.0575*** 0.0549*** 0.0550*** 0.0550*** 0.0550*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) (0.002) Listing status 0.116*** 0.111** 0.107** 0.190*** 0.189*** 0.189* 0.189*** (0.044) (0.045) (0.045) (0.046) (0.047) (0.097) (0.049) Constant 5.176*** 4.767*** 4.769*** 4.757*** 4.754*** 4.793*** 4.797*** 4.797*** 4.797*** (0.011) (0.011) (0.011) (0.011) (0.011) (0.011) (0.011) (0.018) (0.048) Observations 6,553,883 6,407,428 6,407,428 6,407,428 6,407,414 6,407,345 6,400,630 6,400,630 6,400,630 R-squared 0.047 0.051 0.051 0.051 0.053 0.063 0.067 0.067 0.067 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes 14 Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: Current ratio is estimated as current assets over current liabilities (due within a year). The reference (excluded category) refers to Privately- owned Enterprises (POEs), so results should be interpreted as compared to private firms within the same sector, size, and age, and sector and controlling by the respective fixed effects. Firm’s size is measured as the lagged values of log workers in year t-1, age refers to the lagged number of years in the market in t-1. Listed status is a dichotomous variable that takes value 1 if the company is listed in the stock market, 0 otherwise. Fixed effects are included to absorb for potential shocks such as temporary shocks (year-FE), sectoral shocks (sector-FE), time-varying sectoral shocks (sector-year FE), country specific shocks (country-FE), country time varying events (country-year FE), and country and sector specific and time varying shocks (country-sector-year, FE). Clustered standard errors are used to control for potential autocorrelation within firms (column 8 to capture potential serial correlation) and sectors (column 9). Furthermore, we examine the ability of firms to cover their total financial obligations and the overall balance between debt and equity using the solvency ratio. The solvency ratio is provided by the ORBIS financial module and is computed as the ratio between shareholder’s funds and the total assets. Results in Table 9 suggest BOSs have a higher solvency ratio compared to private counterparts. On average, BOS’ solvency ratio is 5.4 percentage points higher than POEs. In other words, BOSs rely relatively more on debt than on shareholders’ funds for their operation or are relatively more leveraged than private peers as evidenced before. Following Phi et al. (2019), this result could be interpreted as a potential signal of soft budget constraints in the sense that BOSs may be more dependent on debt to cover their financial needs and that it can result from indirect or direct state support, debt guarantees, or access to finance at a lower cost. Despite being exposed to higher levels of debt and lower profitability, soft budget constraints can make firms with state participation less likely to face bankruptcy (Megginson & Netter, 2001).19 Table 9. Solvency ratio Solvency ratio (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS 7.045*** 5.606*** 5.542*** 5.582*** 5.624*** 5.455*** 5.481*** 5.481*** 5.481*** (0.092) (0.095) (0.095) (0.095) (0.095) (0.097) (0.098) (0.224) (0.359) Size -0.602*** -2.004*** -2.030*** -2.059*** -2.027*** -1.926*** -1.922*** -1.922*** -1.922*** (0.015) (0.015) (0.015) (0.015) (0.015) (0.015) (0.015) (0.032) (0.088) Age 0.780*** 0.779*** 0.781*** 0.778*** 0.741*** 0.741*** 0.741*** 0.741*** (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.003) (0.009) Listing status 5.675*** 5.567*** 5.501*** 6.367*** 6.295*** 6.295*** 6.295*** (0.240) (0.241) (0.241) (0.250) (0.252) (0.614) (0.335) Constant 42.67*** 36.87*** 36.95*** 37.02*** 36.95*** 37.11*** 37.09*** 37.09*** 37.09*** (0.0514) (0.0528) (0.0529) (0.0530) (0.0530) (0.0536) (0.0538) (0.108) (0.295) Observations 6,565,319 6,418,327 6,418,327 6,418,327 6,418,312 6,418,241 6,411,529 6,411,529 6,411,529 R-squared 0.07 0.106 0.106 0.107 0.109 0.132 0.137 0.137 0.137 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 19This can impact the firm dynamism, creation of jobs, entry and exit of firms, which are beyond the scope of this paper, but that can have important implications for private sector development. 15 Note: Solvency ratio is measured as the shareholders’ funds over total assets. This variable is provided directly in the financial statements from ORBIS (BvD). The reference (excluded category) refers to Privately-owned Enterprises (POEs), so results should be interpreted as compared to private firms within the same sector, size, and age, and sector and controlling by the respective fixed effects. Firm size is measured as lagged values of the log of total workers (year t-1), age refers to the lagged number of years in the market in t-1. Listed status is a dichotomous variable that takes value 1 if the company is listed in the stock market, 0 otherwise. Fixed effects are included to absorb for potential shocks such as temporary shocks (year-FE), sectoral shocks (sector-FE), time-varying sectoral shocks (sector-year FE), country specific shocks (country-FE), country time varying events (country-year FE), and country and sector specific and time varying shocks (country-sector-year, FE). Clustered standard errors are used to control for potential autocorrelation within firms (column 9 to capture potential serial correlation) and sectors (column 10). 6.1.4 Return on investments: ROA and ROE Finally, we assess the return on investments measured by the return on assets (ROA) and returns on equity (ROE). ROE provides an assessment of the firm capability to generate value for their investors for each dollar in equity, while the ROA offers a comparison between the firm profitability and its assets. Results in Table 10 and Table 11 are indicative of a negative relationship between state ownership and the returns of the firm. On average, for every dollar in assets in a BOS firm, the return is about 4.9 percentage points lower than the median private peer in the same industry. BOSs’ returns on equity are approximately 15 percentage points lower compared to private peers. Results hold even after controlling by fixed effects and clustered-standard errors. These results showing a negative relationship between state participation and return on assets are consistent with former analysis by Phi et al., (2019) and Lin & Rowe (2006). Table 10. Results for Return on Assets (ROA) Return on Assets (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS -4.992*** -4.939*** -4.905*** -4.901*** -4.882*** -5.001*** -4.986*** -4.986*** -4.986*** (0.042) (0.044) (0.044) (0.044) (0.044) (0.046) (0.046) (0.074) (0.123) Size -0.651*** -0.546*** -0.532*** -0.556*** -0.540*** -0.505*** -0.499*** -0.499*** -0.499*** (0.007) (0.008) (0.008) (0.008) (0.008) (0.008) (0.008) (0.012) (0.032) Age -0.065*** -0.065*** -0.065*** -0.065*** -0.076*** -0.076*** -0.076*** -0.076*** (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.004) Listing status -3.034*** -2.950*** -2.979*** -2.447*** -2.413*** -2.413*** -2.413*** (0.100) (0.100) (0.100) (0.107) (0.107) (0.183) (0.176) Constant 8.576*** 9.158*** 9.113*** 9.191*** 9.139*** 9.176*** 9.151*** 9.151*** 9.151*** (0.027) (0.028) (0.028) (0.028) (0.028) (0.029) (0.029) (0.042) (0.126) Observations 6,576,325 6,428,898 6,428,898 6,428,898 6,428,883 6,428,815 6,422,117 6,422,117 6,422,117 R-squared 0.040 0.041 0.041 0.044 0.049 0.061 0.069 0.069 0.069 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: Return on Assets (ROA) is computed as the ratio between net income and total assets *100 (%). This variable is provided directly in the financial statements from ORBIS (BvD). The reference (excluded category) refers to Privately-owned Enterprises (POEs), so results should be interpreted as compared to private firms within the same sector, size, and age, and sector and controlling by the respective fixed effects. Size is measured as the log of total workers in year t-1, age refers to the lagged number of years in the market in t-1. Listed status is a dichotomous variable that takes value 1 if the company is listed in the stock market, 0 otherwise. Fixed effects are included to absorb for potential shocks such as temporary shocks (year-FE), sectoral shocks (sector-FE), time-varying sectoral shocks (sector-year FE), country specific shocks (country-FE), 16 country time varying events (country-year FE), and country, sector, and time varying shocks (country-sector-year, FE). Clustered standard errors are used to control for potential autocorrelation within firms (column 8 to capture potential serial correlation) and sectors (column 9). Table 11. Results for Return on Equity Return on Equity (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS -16.17*** -15.34*** -15.24*** -15.29*** -15.27*** -15.48*** -15.45*** -15.45*** -15.45*** (0.190) (0.196) (0.196) (0.196) (0.196) (0.205) (0.206) (0.267) (0.345) Size -2.251*** -1.242*** -1.201*** -1.264*** -1.252*** -1.127*** -1.116*** -1.116*** -1.116*** (0.034) (0.036) (0.036) (0.036) (0.036) (0.037) (0.037) (0.048) (0.076) Age -0.577*** -0.576*** -0.575*** -0.574*** -0.599*** -0.597*** -0.597*** -0.597*** (0.004) (0.004) (0.004) (0.004) (0.004) (0.004) (0.005) (0.018) Listing status -8.680*** -8.409*** -8.439*** -7.690*** -7.535*** -7.535*** -7.535*** (0.426) (0.427) (0.427) (0.464) (0.461) (0.621) (0.552) Constant 26.97*** 31.72*** 31.59*** 31.79*** 31.72*** 31.66*** 31.60*** 31.60*** 31.60*** (0.121) (0.128) (0.129) (0.129) (0.129) (0.131) (0.132) (0.167) (0.285) Observations 6,035,007 5,896,478 5,896,478 5,896,478 5,896,459 5,896,382 5,889,443 5,889,443 5,889,443 R-squared 0.026 0.029 0.029 0.030 0.033 0.037 0.044 0.044 0.044 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: Return on Equity (ROE) is computed as the ratio between net income and shareholders’ fund *100 (%). This variable is pr ovided directly in the financial statements from ORBIS (BvD). The reference (excluded category) refers to Privately-owned Enterprises (POEs), so results should be interpreted as compared to private firms within the same sector, size, and age, and sector and controlling by the respective fixed effects. Size is measured as the log of total workers in year t-1, age refers to the lagged number of years in the market in t-1. Listed status is a dichotomous variable that takes value 1 if the company is listed in the stock market, 0 otherwise. Fixed effects are included to absorb for potential shocks such as temporary shocks (year-FE), sectoral shocks (sector-FE), time-varying sectoral shocks (sector-year FE), country specific shocks (country-FE), country time varying events (country-year FE), and country and sector specific and time varying shocks (country-sector-year, FE). Clustered standard errors are used to control for potential autocorrelation within firms (column 8 to capture potential serial correlation) and sectors (column 9). 6.2 Intensive margin: Does the level of state participation matter for operational and financial performance? The state has multiple channels to participate in the corporate sector. It is not necessarily restricted to the majority or full shareholding. While BOSs are often majority or fully owned by the state, the state can also be a minority shareholder with mixed ownership with the private sector. In our sample, 58% of the BOSs are fully owned by the state (100%), and 16% are majority-owned (50%-99%). Additionally, 27% of the BOSs in our sample have minority shareholding (10%-49%). This section explores to what extent the level of state participation as well as involvement of the private sector can influence firm performance. When there is private sector participation, we expect to observe potential gains in performance as private counterparts can contribute with financial, technical, and managerial resources as well as share risks and rewards with the government (UNDP, 2000). 17 For presentation purposes, the results are grouped in the Table 12 for the different outcomes of interest - labor productivity, profitability, liquidity, and return on investments.20 These findings seem to confirm a gap between BOSs and POEs regardless the level of state shareholding. In other words, when the state holds 10% or more shareholding in the firm, the operational and financial results are subpar to those of private peers. Nonetheless, it also reveals private ownership can also play an important role in narrowing the gap between POEs and BOSs and improving performance. Results in Table 12 show performance differentials enlarge as state participation increases -particularly when the government holds majority or full control. On the contrary, BOS firms with mixed ownership jointly with the private sector and with minority state ownership seem to perform relatively better. Table 12.Estimations by level of state participation (1) (2) (3) (4) (5) (6) Labor productivity Current Solvency VARIABLES Profit margin ROA ROE (log) ratio ratio BOS - 10-24.9% -0.187*** -2.810*** -0.203*** 1.460*** -3.190*** -10.66*** (0.019) (0.299) (0.0552) (0.356) (0.157) (0.677) BOS -25-49.9% -0.229*** -3.316*** -0.170*** 1.773*** -3.123*** -10.25*** (0.020) (0.304) (0.0530) (0.392) (0.163) (0.610) BOS -50%-99.9% -0.390*** -4.068*** 0.0547 3.983*** -4.194*** -12.25*** (0.015) (0.233) (0.0699) (0.340) (0.141) (0.534) BOS -100% -0.432*** -8.578*** -0.614*** 7.563*** -6.101*** -18.88*** (0.025) (0.436) (0.0706) (0.504) (0.171) (0.469) Size 0.432*** -0.615*** -0.572*** -1.906*** -0.508*** -1.142*** (0.009) (0.0581) (0.0135) (0.0884) (0.0317) (0.0756) Age -0.0156*** 0.0171*** 0.0551*** 0.740*** -0.0757*** -0.596*** (0.001) (0.00620) (0.00193) (0.00901) (0.00380) (0.0175) Listing status -0.643*** -0.440* 0.144*** 6.675*** -2.610*** -8.133*** (0.028) (0.264) (0.0490) (0.333) (0.168) (0.535) Constant 4.775*** 7.554*** 4.800*** 37.04*** 9.174*** 31.67*** (0.106) (0.168) (0.0482) (0.295) (0.125) (0.284) Observations 6,452,778 6,347,824 6,400,630 6,411,529 6,422,117 5,889,443 R-squared 0.718 0.105 0.067 0.137 0.069 0.044 Country FE Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Country-year FE Yes Yes Yes Yes Yes Yes Sector-Year FE Yes Yes Yes Yes Yes Yes Country-Sector FE Yes Yes Yes Yes Yes Yes Country-Sector-Year FE Yes Yes Yes Yes Yes Yes Clustered errors Yes Yes Yes Yes Yes Yes Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: The reference (excluded category) refers to Privately-owned Enterprises (POEs), so results should be interpreted as compared to private firms within the same sector, size, and age, and controlling by the respective fixed effects. Labor productivity is measured as the total revenues per worker in logarithm. Profit margin is the ratio between profits before tax over total revenues*100 (%). Current ratio is estimated as current assets over current liabilities (due within a year). Solvency ratio is measured as the shareholders’ funds over total assets. Return on Equity (ROE) is computed as the ratio between net income and shareholders’ fund *100 (%). Return on Assets is computed as the ratio between net income and total assets 20 Results are robust and hold for the other specifications that include progressively the control variables and fixed effects as shown in the former section. 18 *100 (%). This variable is provided directly in the financial statements from ORBIS (BvD). Estimates include all sample. Size is measured as the log of total workers in year t-1 for the outcomes in columns (2-6), whereas size is measures as log of total assets when estimating labor productivity (Column 1) since the outcome is normalized by employment. Age is the lagged number of years in the market in t-1. Listed status is a dichotomous variable that takes value 1 if the company is listed in the stock market, 0 otherwise. Fixed effects are included to absorb for potential shocks such as temporary shocks (year-FE), sectoral shocks (sector-FE), time-varying sectoral shocks (sector-year FE), country specific shocks (country-FE), country time varying events (country-year FE), and country and sector specific and time varying shocks (country-sector-year, FE). Clustered standard errors are used to control for potential autocorrelation at the industry-level. Compared to firms solely owned by the state, firms with private participation operate more closely to their private counterparts. As state participation decreases, the gap between BOS and POEs seems to shrink. Results in column 1 (Table 12) suggest fully owned BOSs are approximately 35% less productive than POEs, whereas the gap reduces up to 17% when the state is a minority shareholder (i.e., participation between 10%-24.9%).21 Similar results are found in terms of profitability and return on investments. Profit margins (column 2) in fully-owned BOS are, on average, 8.5 percentage points lower than those reported by private counterparts within the same industry. As the state participation reduces to less than 25% (minority shareholding), the gap is only about 2.8 percentage points. Returns on equity and assets show a similar pattern. Fully owned BOSs are more distant from POEs than firms with mixed ownership. When the state is a minority shareholder, the ROA for BOSs is about 3.1 percentage points lower than POEs, while for majority and fully owned BOS this difference increases up to 4.2 and 6.1 percentage points, respectively. The lower the state participation, the smaller the performance gaps in terms of liquidity. For each dollar due within a year, fully owned BOSs have on average 0.6 dollars less to cover short-term obligations in current assets (column 3). On the contrary, the gap decreases to 0.2 when the state holds less than 25% participation. Overall, firms with 10% or more state participation have lower reserves on assets to fulfill financial commitments in the long term measured by the solvency ratio (column 4). Yet, when there is mixed ownership and the private sector is the main owner, the differential is about five times lower. These results can obey to different risk-taking decisions when the state holds a dominant role as owner. As the state controls or owns more shareholding in a firm, it can outweigh other shareholders and leverage the connection to potential external resources. Hence, the level of state participation can influence risk-taking behavior and investment decisions, which may not necessarily be driven by profitability objectives (Uddin, 2016; Nguyen et al, 2020). Firms with higher state shareholding could be more exposed to the influence of the government, hindering effectiveness, and profitability by squeezing out commercial goals (Boko & Qin, 2011). Results highlight that the private sector can play an important role driving improved performance. As suggested by (Shang, Yuan, Li, & Fan, 2022), mixed ownership and non-state shareholders seem to reduce potential discretion of the state to influence management decisions of the firm and help mitigating principal-agency problems. Results seem indicative that private shareholding can help to balance the decision on whether to take a more prudent or market-oriented approach driven by considerations about performance, survival, and growth rather than social and political objectives (Nguyen et al, 2020). 6.3 Does the degree of separation from the state matter? The state can act as a direct shareholder or operate at arm’s length through other entities. By using the WB BOS database, we can explore whether there are potential differences when the state is the direct shareholder compared to cases in which it owns a BOS through other(s) firm(s) (i.e., indirect owner). Former authors such as Okhmatovskiy (2010) and Liljeblom, Maury, & Hörhammer (2020) argue effects of state ownership on performance may be more evident among directly owned companies, while the effect 21 The interpretation of the coefficients requires the − transformation such that the average relationship is estimated as ((1 ) − 1) ∗ 100. 19 of indirect ownership is more ambiguous.22 We test this potential relationship to address this important empirical question.23 We argue the government’s ability to influence strategic and managerial decisions of BOSs is higher when it acts as a direct owner, whereas it may diminish as the state has participation through other entities or chain of entities. For instance, when acting as a direct shareholder, the government can have voting rights and exercise the decision powers of the board, which then can be reflected in different risk-taking decisions and financial outcomes of the firm. Therefore, we explore how the degree of proximity to the public authority as a shareholder can explain differences in performance among BOSs. For this purpose, we create an indicator of the degree of proximity to the state based on the ownership structures provided by the WB BOS database. We differentiate between BOSs owned directly by the government (i.e., public authority) and those in which it holds participation through another firm or set of firms. For the latter, we identify how many firms -denoted as layers- participate in the ownership path and differentiate among BOSs with 2, 3, and 4+ layers of ownership (Table 13).24 Table 13. Degree of proximity to the state Category Definition Direct ownership Firms owned by a central or subnational public authority with 10% or more participation. For instance, Ministry of Finance owns 25% of firm A. Indirect (2 layers) The state is an indirect owner through another firm/entity. In this case, B is an indirectly owned BOS as the firm A (with 25% direct participation of the MoF) owns 10% or more of company B. Indirect (3 layers) The state is an indirect owner through ownership links that relate three firms. In this case, C is an indirectly owned BOS if the firm B owns 10% or more of company C, which is owned (by 10% or more) by company A, which is owned by the state. So, MoF owns A, which owns B, B owns C. Therefore, C owned by the government through at least 3 layers of ownership. Indirect (4+ layers) Firms indirectly owned by the state such that four or more firms are identified in the ownership path that connects the firm D with the Public Authority. For example, Public Authority owns A, which owns B, which owns C, and it owns D. D is indirectly owned by the government as it is connected to a Public Authority through a chain of four (or more) participated firms. Source: Author’s elaboration using World Bank’s BOS database Results in Table 14 are indicative of indirectly owned BOSs being associated with better performance results when compared to directly owned BOSs. As the distance from the public authority increases, BOSs appear to achieve better performance outcomes. Compared against directly owned BOSs (reference category), column (1) indicates that indirectly owned BOSs are 28% more productive.25 When the degree of separation from the state increases, which we consider as a proxy of autonomy,26 BOSs seem to perform relatively better achieving up to 62% more revenues per worker (labor productivity) when there are 4 or more layers between the BOS firm and the public authority who exercises the ownership rights. Similarly, the results suggest that further separation from the public authority is associated with better profitability up to 4.2 percentage points (column 2) higher and up to 9 percentage points higher returns of investments 22 (Okhmatovskiy, 2010) suggests that indirect state control may still give access to state resources but reduce the negative governance effects arising from direct shareholding. 23 For this assessment, we restrict the analysis to compare among BOS as we do not have the ownership structure of private counterparts in the WB BOS database. 24 To illustrate the difference, A is directly owned by the state if the direct shareholder corresponds to a Public Authority (denoted as such in ORBIS). On the contrary, B is indirectly owned, if it is participated by company A, which is owned by a Public Authority. The entity type is built from the ORBIS entity type vintage files. 25 The interpretation of the coefficients requires the − transformation such that the average relationship is estimated as ((1 ) − 1) ∗ 100 = (exp (0.247)-1) *100= 28.01%. 26 It means more layers of ownership separate the entity from the public authority. 20 (columns 6) vis-à-vis directly owned BOSs. Interestingly in terms of liquidity, we find further separation from the state seems to reduce the financial leverage of BOSs and reduce the gap vis-à-vis POEs. This could be explained by the fact that as firms depart from the public authority, the probability of receiving direct government support could diminish, and therefore their decisions on financial leverage are more aligned with those of POEs. These findings could be indicative than higher separation or autonomy from the firm could serve to aligned better the incentives from the firm to achieve improved performance outcomes, which can curb political interference of the government in the day-to-day decisions of the firm and mitigate principal-agent issues. Table 14. Estimations by degree of proximity of state participation (1) (2) (3) (4) (5) (6) Labor productivity Solvency Profit margin Current ratio ROA ROE VARIABLES (log) ratio Indirect - 2 layers 0.247*** 1.501*** -0.300*** -6.793*** 0.709*** 4.934*** (0.0117) (0.189) (0.0677) (0.303) (0.181) (0.758) Indirect - 3 layers 0.414*** 2.270*** -0.928*** -8.877*** 1.247*** 7.376*** (0.0211) (0.330) (0.107) (0.451) (0.228) (0.868) Indirect - 4+ layers 0.487*** 4.211*** -0.683*** -10.70*** 2.064*** 9.513*** (0.0241) (0.297) (0.135) (0.527) (0.236) (1.040) Size 0.342*** 0.322*** -0.472*** -0.854*** 0.125*** -0.419** (0.00568) (0.101) (0.0201) (0.117) (0.0472) (0.168) Age -0.0101*** 0.0894*** 0.0101*** 0.416*** 0.0265*** -0.118*** (0.000711) (0.0117) (0.00344) (0.0171) (0.00645) (0.0244) Listing status -0.311*** 1.849*** -0.111 1.670** -0.184 0.135 (0.0261) (0.521) (0.0932) (0.693) (0.253) (1.380) Constant 5.026*** -3.509*** 5.066*** 48.35*** -0.665*** 3.789*** (0.0818) (0.538) (0.112) (0.508) (0.250) (0.811) Observations 179,124 169,457 176,038 175,205 177,164 159,442 R-squared 0.700 0.230 0.133 0.195 0.106 0.086 Country FE Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Country-year FE Yes Yes Yes Yes Yes Yes Sector-Year FE Yes Yes Yes Yes Yes Yes Country-Sector FE Yes Yes Yes Yes Yes Yes Country-Sector-Year FE Yes Yes Yes Yes Yes Yes Clustered errors Yes Yes Yes Yes Yes Yes Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: This regression is performed only among BOSs as we do not have the ownership structure of the POEs in the data. Thus, the reference (excluded category) refers to directly owned (BOSs), so results should be interpreted as compared to a firm directly owned by the state (10% or more) in the same sector, size, and age, and controlling by the respective fixed effects. Labor productivity is measured as the total revenues per worker in logarithm. Profit margin is the ratio between profits before tax over total revenues*100 (%). Current ratio is estimated as current assets over current liabilities (due within a year). Solvency ratio is measured as the shareholders’ funds over total assets. Return on Equity (ROE) is computed as the ratio between net income and shareholders’ fund *100 (%). Return on Assets is computed as the ratio between net income and total assets *100 (%). This variable is provided directly in the financial statements from ORBIS (BvD). Estimates include only firms with state participation -BOS only. Size is measured as the log of total workers in year t-1 for the outcomes in columns (2-6), whereas size is measures as log of total assets when estimating labor productivity (column 1) since the outcome is normalized by employment. Age is the lagged number of years in the market in t-1. Listed status is a dichotomous variable that takes value 1 if the company is listed in the stock market, 0 otherwise. Fixed effects are included to absorb for potential shocks such as temporary shocks (year-FE), sectoral shocks (sector-FE), time-varying sectoral shocks (sector- 21 year FE), country specific shocks (country-FE), country time varying events (country-year FE), and country and sector specific and time varying shocks (country-sector-year, FE). Clustered standard errors are used to control for potential autocorrelation at the industry-level. 6.4 Does the performance gaps vary across sectors? It is commonly expected to observe performance differentials between POEs and BOSs as the latter are often coupled with a public mandate and pursue non-commercial functions that are not necessarily aligned with a profit maximization behavior (Pratuckchai & Patanapongse, 2012). Sometimes, these public service mandates are neither accounted for nor properly compensated and can impact firm performance. Although this might justify the deviation of the financial performance of BOSs vis-à-vis POEs, we explore whether this differs across the type of markets in which the state operates. We argue that the likelihood of BOSs fulfilling a public mandate is lower in competitive sectors that can be viable for private investors such as the manufacturing of textiles, accommodation, or retail activities. Unless other market distortions unlevel the playing field, when operating in competitive sectors, BOSs should be expected to behave more closely to their private peers. In other words, signs of lower performance in competitive sectors can potentially have implications of misallocation of resources and opportunity costs of state ownership as those sectors could be more efficiently served by the private sector. In competitive sectors, private entry is viable, technology is relatively homogeneous, and restrictions for operation are lower (Dall'Olio, et al., 2022b). On the contrary, as the degree of contestability of a sector decreases and market failures are more common such as externalities, public goods, or natural monopolies, then it is more likely that BOSs fulfill some public service obligations (e.g., postal services). In those cases, the underperformance of BOS could be expected by the potential of social mandates and the need to provide public services. In this section, we assess to what extent performance differentials vary across sectors where BOSs operate by using the WB sector taxonomy (Dall'Olio, et al., 2022b) and explore the potential differences between competitive, partially contestable, and natural monopolies.27 First, we compare between competitive and non-competitive sectors by combining the sectors denoted as natural monopolies and partially contestable in a single category. Next, we break down the analysis by exploring the differences across the three types of sectors -competitive, partially contestable, and natural monopolies. Results in Table 15Error! Reference source not found. suggest a performance gap between BOSs and POEs across all types of sectors. Nonetheless, it also features some heterogeneity depending on the markets in which the state operates. BOSs in competitive sectors seem to achieve lower profit margins (column 5), lower return on assets and equity (columns 14 and 17), and be more financially able to cover short-term obligations (column 8) vis-à-vis non-competitive sectors. Results in Table 16 show these differences are statistically significant. We delve deeper into these differences within non-competitive sectors for natural monopolies and partially contestable sectors in Error! Reference source not found.columns (3, 6, 9, 12, 15, 18). Findings confirm there is a differential across all sector types, but it seems larger in terms of profitability and returns on investments when BOSs operate in competitive sectors. Again, we confirm these differences are statistically significant (Table 16). These results are consistent with previous studies that find firms with state participation in competitive sectors tend to be less efficient and have lower profitability compared to private firms (Megginson, Nash, & Randenborgh, 1994), (Boardman & Vining, 1989), (Boardman & Vining, 1992), (Shirley & P., 2000). Findings also suggest that BOSs depart the most from POEs in terms of labor 27 Definitions are presented in Table 3. For more information on how the taxonomy was built and sector characteristics please refer to (Dall'Olio, et al., 2022b). 22 productivity and solvency when operating in natural monopoly sectors, which has been documented with POEs outperforming in sewage and garbage collection (Savas, 1977; Bennett & Johnson, 1979).Error! Reference source not found. 23 Table 15. Estimations of performance differentials between BOSs and POEs by sector-type (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Labor Labor Labor Profit Profit Profit Current Current Current Solvency Solvency Solvency VARIABLES productivity productivity productivity margins margins margins ratio ratio ratio ratio ratio ratio BOS -0.448*** -6.297*** -0.436*** 5.38*** (0.021) (0.319) (0.063) (0.365) BOS*Competitive sectors -0.405*** -6.953*** -0.471*** 4.304*** (0.024) (0.390) (0.077) (0.388) BOS*Non-Competitive s. -0.569*** -4.454*** -0.340*** 8.448*** (0.034) (0.681) (0.071) (0.625) BOS*Competitive sector -0.405*** -6.948*** -0.470*** 4.318*** (0.024) (0.389) (0.077) (0.388) BOS*Natural monopoly -0.622*** -2.325*** -0.157*** 14.68*** (0.041) (0.644) (0.073) (0.461) BOS*P. Contestable sector -0.534*** -5.981*** -0.467*** 4.092*** (0.0528) (0.981) (0.099) (0.526) Size 0.443*** 0.444*** 0.444*** -0.609*** -0.615*** -0.614*** -0.580*** -0.571*** -0.581*** -2.068*** -2.07*** -2.07*** (0.008) (0.008) (0.008) (0.053) (0.054) (0.054) (0.013) (0.013) (0.013) (0.085) (0.085) (0.085) Age -0.0154*** -0.0154*** -0.0154*** 0.018*** 0.018*** 0.018*** 0.057*** 0.057*** 0.057*** 0.770*** 0.771*** 0.771*** (0.0006) (0.0006) (0.0006) (0.005) (0.005) (0.005) (0.002) (0.002) (0.002) (0.009) (0.009) (0.009) Listing status -0.688*** -0.686*** -0.685*** 0.168 0.127 0.106 0.226*** 0.224*** 0.223*** 6.43*** 6.373*** 6.315*** (0.028) (0.028) (0.028) (0.359) (0.333) (0.333) (0.050) (0.050) (0.050) (0.329) (0.317) (0.317) Constant 4.622*** 4.621*** 4.621*** 7.508*** 7.524*** 7.516*** 4.804*** 4.805*** 4.805*** 37.19*** 37.22*** 37.2*** (0.109) (0.109) (0.109) (0.159) (0.160) (0.160) (0.048) (0.045) (0.045) (0.279) (0.279) (0.279) Observations 6,459,435 6,459,435 6,459,435 6,354,524 6,354,524 6,354,524 6,407,322 6,407,322 6,407,322 6,418,217 6,418,217 6,418,217 R-squared 0.701 0.701 0.701 0.081 0.082 0.082 0.056 0.056 0.067 0.137 0.117 0.118 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector-year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-Sector-year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Clustered errors Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 24 (Continued) (13) (14) (15) (16) (17) (18) VARIABLES ROA ROA ROA ROE ROE ROE BOS -5.04*** -15.61*** (0.135) (0.394) BOS*Competitive sectors -5.297*** -16.32*** (0.157) (0.500) BOS*Non-Competitive s. -4.328*** -13.63*** (0.358) (0.924) BOS*Competitive sector -5.295*** -16.32*** (0.156) (0.499) BOS*Natural monopoly -3.518*** -12.98*** (0.323) (0.790) BOS*P. Contestable sector -4.889*** -14.09*** (0.504) (1.399) Size -0.526*** -0.529*** -0.529*** -1.120*** -1.128*** -1.127*** (0.032) (0.032) (0.032) (0.075) (0.075) (0.075) Age -0.073*** -0.073*** -0.073*** -0.602*** -0.602*** -0.602*** (0.003) (0.003) (0.003) (0.017) (0.017) (0.017) Listing status -2.413*** -2.511*** -2.518*** -7.43*** -7.470*** -7.476*** (0.176) (0.200) (0.199) (0.552) (0.549) (0.549) Constant 9.20*** 9.211*** 9.208*** 31.67*** 31.70*** 31.69*** (0.125) (0.125) (0.125) (0.284) (0.284) (0.284) Observations 6,422,117 6,428,788 6,428,788 5,896,364 5,896,364 5,896,364 R-squared 0.055 0.056 0.056 0.034 0.035 0.035 Country FE Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Country-year FE Yes Yes Yes Yes Yes Yes Sector-Year FE Yes Yes Yes Yes Yes Yes Country-Sector FE Yes Yes Yes Yes Yes Yes Country-Sector-Year FE Yes Yes Yes Yes Yes Yes Clustered errors Yes Yes Yes Yes Yes Yes Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: Labor productivity is measured as the total revenues per worker in logarithm. Profit margin is the ratio between profits before tax over total revenues*100 (%). Current ratio is estimated as current assets over current liabilities (due within a year). Solvency ratio is measured as the shareholders’ funds over total assets. Return on Equity (ROE) is computed as the ratio between net income and shareholders’ fund *100 (%). Return on Assets is computed as the ratio between net income and total assets *100 (%). The reference (excluded category) refers to Privately-owned Enterprises (POEs), so results should be interpreted as compared to private firms within the same sector, size, and age, and controlling by the respective fixed effects. Size is measured as the log of total assets when assessing the labor productivity as outcome, otherwise is measured by the log of total workers in year t-1. Age is the lagged number of years in the market in t-1. Listed status is a dichotomous variable that takes value 1 if the company is listed in the stock market, 0 otherwise. Fixed effects are included to absorb for potential shocks such as temporary shocks (year-FE), sectoral shocks (sector-FE at 2-digits), time-varying sectoral shocks (sector-year FE), country specific shocks (country-FE), country time varying events (country-year FE), and country and sector specific and time varying shocks (country-sector-year, FE). Clustered standard errors are used to control for potential autocorrelation at the industry-level. In this case, the sectoral fixed effects are considered at the 2-digits level to capture sectoral heterogeneities that are at a higher level than the sector taxonomy used at 4-digits level. For instance, this allows to capture differences within the electricity sector while acknowledging the different type of markets (e.g., production as competitive vs. transmission as natural monopoly 25 Table 16. Testing statistical significance and equivalence of coefficients Competitive vs. partially Competitive vs. Non-competitive contestable and natural Indicator p-value p-value (F-value) monopoly sectors (F-value) Labor productivity 15.99*** 0.0010 12.34*** 0.0000 Profit margin 8.82*** 0.0031 15.87*** 0.0000 Current ratio 1.77 0.1844 5.58*** 0.0039 Solvency ratio 34.78*** 0.0000 69.29*** 0.0000 ROA 5.6** 0.0182 10.61*** 0.0000 ROE 5.79** 0.0163 5.49*** 0.0043 Note: These results test the null hypothesis of equality of the coefficients of BOS in competitive sectors vis-à-vis non-competitive sectors (column 2) and joint hypothesis of BOS in competitive sectors vs. partially contestable sectors, and natural monopolies. *** p<0.01, ** p<0.05, * p<0.1*** Finally, we explore whether some differences in performance can vary simultaneously by the level of state participation and the type of market in which BOSs operate. We observe that when differentiating by competitive, natural monopoly, and partially contestable sectors, BOSs results are lower in terms of labor productivity at all levels of state participation. Yet, the largest differentials are evidenced in competitive and partially contestable sectors when BOSs are fully owned by the state.28 Regarding profit margins, the results in Table 17 indicate that BOSs in natural monopoly sectors and partially contestable sectors in which the state holds a minority (10-24.9%) do not differ statistically from the POEs. On the contrary, BOS profits in competitive sectors are lower than comparator peers at all levels of participation. For all sectors, the differential in profitability between BOSs and POEs is statistically significant and larger when the state is the full owner. The coefficients are statistically different and significant at 1%, 5% and 10% significance levels as shown in Annex 4. In terms of liquidity in the short-term (current ratio), there are no significant differences between BOSs and POEs in natural monopoly sectors when there is minority state participation below 50% (Table 17, columns 9-11). For partially contestable sectors, this is also true when the state is a minority shareholder below 25%. When the BOSs are fully owned by the state, BOSs appear more financially leveraged across all types of sectors. Furthermore, when measuring long-term leverage using the solvency ratio, findings suggest BOSs are more dependent on external debt vis-à-vis POEs for covering long-term obligations across all type of sectors, but particularly when BOSs are fully owned (columns 10-12). Finally, disparity between BOSs and POEs persist across all levels of state participation and sector-types for the indicators of return on investments -both ROA and ROE. As state participation increases within competitive, natural monopoly or partially contestable sectors, the differentials expand vis-à-vis POEs. The largest divergence between POEs and BOSs in terms of ROA and ROE is found for BOSs fully owned by the state across all 3 types of sectors (column 14-18). 28For brevity, the test of the statistical differences among the levels of state participation is included in Annex 4. Nonetheless, the results and p- values show the coefficients for the level of state participation are statistically different and significant. 26 Table 17. Intensive margins and type of sectors where BOSs operate (1) (2) (3) (4) (5) (6) (7) (8) (9) Labor Labor Labor Profit Profit Profit Current Current Current productivity productivity productivity margin margin margin ratio ratio ratio VARIABLES Comp Nat. Mon. Part. Contestable. Comp Nat. Mon. Part. Contestable. Comp Nat. Mon. Part. Contestable. BOS - 10-24.9% -0.178*** -0.144*** -0.172*** -3.612*** 0.120 -0.473 -0.273*** -0.00213 0.0597 (0.0224) (0.0314) (0.0280) (0.341) (0.441) (0.431) (0.0656) (0.107) (0.113) BOS -25-49.9% -0.221*** -0.152*** -0.213*** -3.610*** -1.678*** -3.096*** -0.185*** 0.0377 -0.190* (0.0236) (0.0228) (0.0289) (0.375) (0.378) (0.546) (0.0663) (0.0976) (0.113) BOS -49.9%-99.9% -0.346*** -0.378*** -0.467*** -4.450*** -2.843*** -3.281*** 0.115 0.282** -0.252** (0.0164) (0.0247) (0.0305) (0.277) (0.367) (0.617) (0.0897) (0.120) (0.0973) BOS -100% -0.415*** -0.332*** -0.537*** -8.869*** -6.756*** -8.619*** -0.637*** -0.485*** -0.534*** (0.0326) (0.0242) (0.0235) (0.541) (0.428) (1.066) (0.0916) (0.0812) (0.121) Size 0.435*** 0.376*** 0.400*** -0.628*** -0.195 -0.626*** -0.571*** -0.585*** -0.575*** (0.00910) (0.0111) (0.0116) (0.0626) (0.169) (0.0736) (0.0143) (0.0304) (0.0302) Age -0.0157*** -0.0183*** -0.0122*** 0.0190*** 0.106*** -0.0722*** 0.0559*** 0.0457*** 0.0361*** (0.000780) (0.00106) (0.000746) (0.00643) (0.0133) (0.0103) (0.00202) (0.00407) (0.00316) Listing status -0.684*** -0.373*** -0.346*** -1.090*** 1.409** 3.857*** 0.153*** -0.00744 0.259** (0.0294) (0.0347) (0.0461) (0.248) (0.693) (0.707) (0.0533) (0.147) (0.126) Constant 4.753*** 5.036*** 4.909*** 7.519*** 6.113*** 9.511*** 4.788*** 4.781*** 5.038*** (0.111) (0.140) (0.155) (0.179) (0.680) (0.343) (0.0509) (0.0887) (0.0972) Observations 6,110,399 107,406 234,973 6,014,287 105,235 228,302 6,062,067 106,629 231,934 R-squared 0.719 0.702 0.692 0.102 0.117 0.169 0.068 0.042 0.051 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector-Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-Sector-Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Clustered errors Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 27 (Continued) (10) (11) (12) (13) (14) (15) (16) (17) (18) Solvency ratio Solvency ratio Solvency ratio ROA ROA ROA ROE ROE ROE Part. Part. VARIABLES Comp Nat. Mon. Part. Contestable. Comp Nat. Mon. Contestable. Comp Nat. Mon. Contestable. BOS*[10-25%) 1.310*** 2.524*** 2.197*** -3.549*** -2.050*** -1.892*** -11.96*** -6.669*** -5.737*** (0.416) (0.541) (0.765) (0.179) (0.373) (0.377) (0.795) (1.843) (1.506) BOS* [25-50%) 1.097** 7.193*** 1.391* -3.411*** -1.854*** -2.573*** -11.46*** -4.540*** -8.021*** (0.456) (0.671) (0.787) (0.183) (0.329) (0.392) (0.675) (1.120) (1.869) BOS * [50%-99] 3.934*** 6.230*** 3.729*** -4.361*** -3.516*** -3.878*** -12.93*** -9.609*** -10.50*** (0.409) (0.698) (0.691) (0.175) (0.306) (0.354) (0.668) (1.078) (1.388) BOS*[100%] 6.778*** 13.15*** 6.592*** -6.136*** -5.320*** -6.406*** -19.37*** -15.55*** -18.81*** (0.628) (0.492) (0.671) (0.187) (0.285) (0.573) (0.542) (0.879) (1.504) Size -1.928*** -1.219*** -1.790*** -0.506*** -0.523*** -0.548*** -1.128*** -1.563*** -1.269*** (0.0944) (0.239) (0.181) (0.0339) (0.100) (0.0592) (0.0807) (0.275) (0.188) Age 0.745*** 0.684*** 0.633*** -0.0752*** -0.0449*** -0.104*** -0.596*** -0.502*** -0.633*** (0.00929) (0.0299) (0.0188) (0.00394) (0.0116) (0.00883) (0.0182) (0.0472) (0.0278) Listing status 6.862*** 3.491*** 6.764*** -3.072*** 0.0631 -0.168 -9.011*** -1.568 -4.150*** (0.366) (0.968) (1.018) (0.148) (0.527) (0.458) (0.564) (1.546) (1.514) Constant 36.91*** 39.70*** 39.49*** 9.153*** 8.871*** 9.797*** 31.59*** 32.64*** 32.76*** (0.314) (0.941) (0.618) (0.132) (0.469) (0.255) (0.298) (1.423) (0.808) Observations 6,073,902 106,297 231,330 6,083,695 106,331 232,091 5,583,006 95,858 210,579 R-squared 0.138 0.116 0.106 0.069 0.068 0.073 0.044 0.043 0.044 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector-Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-Sector-Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Clustered errors Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 28 6.5 Robustness checks We also test whether these findings are robust to the consideration of shorter periods to try to minimize the effect of omitted changes in ownership along the sample period. We also explore to what extent these results are robust when restricting the sample only to fully owned BOSs. For this purpose, we conduct two exercises. First, we estimate the model proposed in (1) for shorter time windows of 5-years and 3-years. The results of the first exercise are presented in the Annex for the different specifications, but for brevity are summarized in Table 18 for the 5-year period (2016-2020) and Table 19 (3-year period, 2018-2020). Findings show the effects are robust when analyzing shorter timeframes of 5 and 3 years. The magnitude, direction, and statistical significance of the effects remain confirming a negative relationship between state ownership and labor productivity, profit margins, short-term liquidity, and returns of investments. When reducing the timeframe, the results still indicate that BOSs are on average 32-38% less productive, obtain between 6.3 and 7.8 percentage points lower profit margins, have 0.34-0.35 fewer dollars in assets to cover each dollar in short-term debts, and provide lower returns of the order of 4.9-5.6 dollars less for each dollar invested in assets. As before, BOSs also appear to rely more on external financing vis-à-vis equity for covering long-term debt. Table 18. Robustness check considering 5-year period (2016-2020) (1) (2) (3) (4) (5) (6) Labor productivity Profit margin Current ratio Solvency ratio ROA ROE BOS -0.476*** -7.486*** -0.347*** 4.710*** -5.667*** -17.05*** (0.034) (0.517) (0.109) (0.468) (0.147) (0.52) Labor 0.385*** -0.652*** -0.605*** -2.175*** -0.535*** -1.034*** (0.012) (0.078) (0.022) (0.112) (0.0378) (0.087) Age -0.0133*** 0.0363*** 0.0558*** 0.731*** -0.0636*** -0.549*** (0.001) (0.010) (0.003) (0.011) (0.0039) (0.019) Listing status -0.552*** 0.669 0.142 5.895*** -2.746*** -8.024*** (0.041) (0.427) (0.096) (0.516) (0.293) (0.888) Constant 5.622*** 7.652*** 4.701*** 37.76*** 9.362*** 31.07*** (0.15) (0.191) (0.072) (0.383) (0.151) (0.33) Observations 2,892,917 2,848,595 1,475,667 2,865,387 2,870,563 2,649,938 R-squared 0.694 0.115 0.072 0.146 0.075 0.047 Country FE Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Country-year FE Yes Yes Yes Yes Yes Yes Sector-Year FE Yes Yes Yes Yes Yes Yes Country-Sector FE Yes Yes Yes Yes Yes Yes Country-Sector-Year FE Yes Yes Yes Yes Yes Yes Clustered errors Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 29 Table 19. Robustness check considering 3-year window (2018-2020) Labor Profit margin Current ratio Solvency ratio ROA ROE productivity BOS -0.480*** -7.859*** -0.347*** 4.333*** -5.898*** -17.47*** (0.0517) (0.768) (0.109) (0.648) (0.211) (0.782) Labor 0.385*** -0.671*** -0.605*** -2.224*** -0.559*** -1.016*** (0.0178) (0.102) (0.022) (0.149) (0.0479) (0.125) Age -0.0131*** 0.0379** 0.0558*** 0.732*** -0.0562*** -0.513*** (0.0016) (0.0154) (0.00330) (0.0159) (0.00524) (0.0286) Listing status -0.545*** 0.611 0.142 5.772*** -2.687*** -8.116*** (0.0574) (0.641) (0.0963) (0.745) (0.379) (1.183) Constant 5.618*** 7.441*** 4.701*** 37.96*** 9.173*** 29.72*** (0.222) (0.240) (0.0725) (0.501) (0.194) (0.486) Observations 1,476,309 1,461,512 1,475,667 1,468,951 1,470,483 1,357,960 R-squared 0.687 0.126 0.072 0.146 0.085 0.052 Country FE Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Country-year FE Yes Yes Yes Yes Yes Yes Sector-Year FE Yes Yes Yes Yes Yes Yes Country-Sector FE Yes Yes Yes Yes Yes Yes Country-Sector-Year FE Yes Yes Yes Yes Yes Yes Clustered errors Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: The reference (excluded category) refers to Privately-owned Enterprises (POEs), so results should be interpreted as compared to private firms within the same sector-type, size, and age, and controlling by the respective fixed effects. Labor productivity is measured as the total revenues per worker in logarithm. Profit margin is the ratio between profits before tax over total revenues*100 (%). Current ratio is estimated as current assets over current liabilities (due within a year). Solvency ratio is measured as the shareholders’ funds over total assets. Return on Equity (ROE) is computed as the ratio between net income and shareholders’ fund *100 (%). Return on Assets is computed as the ratio between net income and total assets *100 (%). Size is measured as the log of total assets when assessing the labor productivity as outcome, otherwise is measured by the log of total workers in year t-1. Age is the lagged number of years in the market in t-1. Listed status is a dichotomous variable that takes value 1 if the company is listed in the stock market, 0 otherwise. Fixed effects are included to absorb for potential shocks such as temporary shocks (year- FE), sectoral shocks (sector-FE), time-varying sectoral shocks (sector-year FE), country specific shocks (country-FE), country time varying events (country-year FE), and country and sector specific and time varying shocks (country-sector-year, FE). Clustered standard errors are used to control for potential autocorrelation at the industry-level. Second, we test whether the results hold when restricting the sample only to those firms fully owned by the state. Once more, robustness checks show that the results evidenced before are qualitatively unchanged in Table 20 restricting the sample to fully owned BOSs. As before, we find a negative and statistically significant correlation between state participation -defined here as firms with the state as the sole owner- and labor productivity, profit margins, current ratio, ROA, and ROE, whereas BOSs seem to be more leveraged than POEs. 30 Table 20. Robustness checks: estimates fully owned BOSs only (1) (2) (3) (4) (5) (6) Log (labor Profit margin Current ratio Solvency ratio ROA ROE VARIABLES productivity) BOS-fully owned (only) -0.434*** -8.581*** -0.627*** 7.600*** -6.082*** -18.81*** (0.0257) (0.442) (0.0707) (0.503) (0.176) (0.479) Size 0.433*** -0.638*** -0.572*** -1.909*** -0.523*** -1.165*** (0.00869) (0.0589) (0.0137) (0.0896) (0.0323) (0.0769) Age -0.0156*** 0.0168*** 0.0557*** 0.744*** -0.0767*** -0.601*** (0.000748) (0.00619) (0.00193) (0.00900) (0.00383) (0.0176) Listing status -0.639*** -1.006*** 0.109** 6.842*** -3.242*** -9.967*** (0.0316) (0.252) (0.0536) (0.338) (0.152) (0.559) Constant 4.759*** 7.617*** 4.793*** 37.01*** 9.242*** 31.82*** (0.106) (0.169) (0.0486) (0.298) (0.127) (0.286) Observations 6,376,996 6,275,906 6,325,638 6,336,963 6,346,550 5,821,167 R-squared 0.719 0.104 0.067 0.137 0.069 0.044 Country FE Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Country-year FE Yes Yes Yes Yes Yes Yes Sector-Year FE Yes Yes Yes Yes Yes Yes Country-Sector FE Yes Yes Yes Yes Yes Yes Country-Sector-Year FE Yes Yes Yes Yes Yes Yes Clustered errors Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: The reference (excluded category) refers to Privately-owned Enterprises (POEs), so results should be interpreted as compared to private firms within the same sector-type, size, and age, and controlling by the respective fixed effects. Labor productivity is measured as the total revenues per worker in logarithm. Profit margin is the ratio between profits before tax over total revenues*100 (%). Current ratio is estimated as current assets over current liabilities (due within a year). Solvency ratio is measured as the shareholders’ funds over total assets. Return on Equity (ROE) is computed as the ratio between net income and shareholders’ fund *100 (%). Return on Assets is computed as the ratio between net income and total assets *100 (%). Size is measured as the log of total assets when assessing the labor productivity as outcome, otherwise is measured by the log of total workers in year t-1. Age is the lagged number of years in the market in t-1. Listed status is a dichotomous variable that takes value 1 if the company is listed in the stock market, 0 otherwise. Fixed effects are included to absorb for potential shocks such as temporary shocks (year- FE), sectoral shocks (sector-FE), time-varying sectoral shocks (sector-year FE), country specific shocks (country-FE), country time varying events (country-year FE), and country and sector specific and time varying shocks (country-sector-year, FE). Clustered standard errors are used to control for potential autocorrelation at the industry-level. 6.6 Do BOSs react differently to POEs when facing economic shocks? An assessment of the early COVID-19 shock Although there might exist differences in both operational and financial performance, BOSs can play roles in the economy by contributing to smooth economic shocks, providing counter-cyclical investment, and maintaining and protecting employment (Bai, Li, Tao, & Wang, 2000; Telegdy, 2016). The period of our panel allows us to explore whether this is the case in a short-term response to the COVID-19 shock for the year 2020. We analyze the response of POEs vis-à-vis BOSs for investment and employment to the COVID- 19 shock, and whether there is some divergence in the adjustment paths to an unexpected shock. The specification of this estimation follows the model in (2) presented in section 5. 31 Overall, findings in Table 21 indicate that BOSs operate with a larger number of workers and grow at a higher pace compared to POEs (columns 2 and 8). During 2020, the interaction terms serve to explore whether the shock impacted differently BOSs and non-BOSs. The findings in column 3 show no statistical significance on the level of employment reported by BOS vis-à-vis private peers. Yet, results in column 6 reveal that firms with 25% up to 99% state participation grew faster in terms of labor force than private counterparts during the shock in 2020. Regarding investment measured by fixed assets, the results suggest there was a relative contraction of the level of investment among fully owned BOS (column 9), and no significant change in the rate of investment among BOS at any level of state participation (column 12). In sum, in 2020, during the early COVID-19 pandemic, BOSs with state participation between 25% -99% experienced a higher growth rate in employment. These BOSs grew about 1.5 and 0.92 percentage points faster than private peers during 2020. On the contrary, there is no significant evidence suggesting BOSs played a counter-cyclical role in terms of investment growth during 2020. It is important to note that these results capture only a short-term response and that a more comprehensive analysis will be needed to understand the inter-temporal impact and adjustments of BOSs in job creation and investment and potential effects on reallocation of jobs in the long-term, and potential impact on misallocation. For instance, evidence of former events of privatization shows that while the private sector can have a sharp job reduction phase, it could be temporary and promote efficiency gains that could be later translate into higher investment and eventually new jobs (Birdsall & Nellis, 2003). 32 Table 21. Estimations for employment and investment (levels and growth) during early COVID-19 shock. (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) VARIABLES Employment Employment growth Investment Investment growth BOS 0.0453*** 0.0448*** 2.939*** 2.865*** 0.740*** 0.765*** 2.253*** 2.329*** (0.004) (0.005) (0.290) (0.322) (0.031) (0.032) (0.193) (0.198) BOS*2020 0.00353 0.620 -0.217** -0.636 (0.006) (0.499) (0.097) (0.784) BOS 10-24.9% 0.0376*** 0.998*** 0.857*** 0.974*** (0.005) (0.184) (0.027) (0.326) BOS 25-49.9% 0.0587*** 1.811*** 0.959*** 2.806*** (0.008) (0.335) (0.034) (0.338) BOS 50%-99.9% 0.0538*** 1.999*** 1.017*** 1.852*** (0.006) (0.242) (0.043) (0.328) BOS 100% 0.0411*** 3.713*** 0.671*** 2.918*** (0.005) (0.451) (0.044) (0.262) BOS 10-24.9% *2020 0.00171 0.605 -0.055 -0.134 (0.011) (0.556) (0.087) (0.960) BOS 25-49.9%*2020 -0.00309 1.543** -0.0938 -1.799 (0.011) (0.673) (0.107) (1.119) BOS 50%-99.9%*2020 0.00484 0.928* -0.198 -0.646 (0.009) (0.539) (0.131) (1.101) BOS 100%*2020 0.00468 0.342 -0.282** -0.666 (0.007) (0.681) (0.136) (0.941) Size 0.936*** 0.936*** 0.936*** -3.456*** -3.456*** -3.449*** 1.141*** 1.141*** 1.140*** 2.401*** 2.401*** 2.403*** (0.003) (0.003) (0.003) (0.143) (0.143) (0.143) (0.008) (0.008) (0.008) (0.066) (0.066) (0.066) Age -0.00208*** -0.00208*** -0.00208*** -0.110*** -0.110*** -0.111*** 0.0251*** 0.0251*** 0.0252*** -0.203*** -0.203*** -0.203*** (0.000) (0.000) (0.000) (0.006) (0.006) (0.006) (0.001) (0.001) (0.001) (0.006) (0.006) (0.006) Listing status 0.0546*** 0.0546*** 0.0539*** 2.168*** 2.169*** 2.322*** 1.339*** 1.339*** 1.316*** -0.638** -0.639** -0.554* (0.006) (0.006) (0.006) (0.307) (0.307) (0.314) (0.043) (0.043) (0.043) (0.308) (0.308) (0.307) Constant 0.269*** 0.269*** 0.269*** 14.57*** 14.57*** 14.55*** 7.552*** 7.552*** 7.554*** -7.917*** -7.917*** -7.922*** (0.012) (0.012) (0.012) (0.468) (0.468) (0.469) (0.027) (0.027) (0.027) (0.259) (0.259) (0.259) Observations 6,503,554 6,503,554 6,503,554 6,159,926 6,159,926 6,159,926 5,638,506 5,638,506 5,638,506 5,324,141 5,324,141 5,324,141 R-squared 0.900 0.900 0.900 0.142 0.142 0.142 0.535 0.535 0.535 0.065 0.065 0.065 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: Employment is measured as the log of total workers, employment growth as the percentage annual change in labor force. Investment in levels is measured as the log of fixed assets, and investment growth as the annual (percentage) change of fixed assets. The estimations include country, sector, year fixed effects, as well as clustered standard errors to correct for potential autocorrelation within the industry. The reference (excluded) category refers to POEs, so the results should be interpreted as compared to the benchmark to private peers after controlling by size, sector, age, and other fixed effects. 33 7. CONCLUSIONS AND POLICY IMPLICATIONS The role of the state in markets is still at play and it can have implications for firm performance. This paper analyzes the performance differentials in terms of operational efficiency measured by labor productivity and financial performance through profitability, liquidity (current ratio and solvency ratio), and return on investments (ROA and ROE) between privately owned enterprises (POEs) and Businesses of the State (BOSs). The latter are defined as firms with state direct and indirect state participation of at least 10% following Dall'Olio et al. (2022). This paper uses two main data sources: firm-level data from the ORBIS financial module and the World Bank’s Businesses of the State (BOS) database. This paper provides new evidence on the relationship between state ownership and firm performance studying 16 countries in the European and Central Asia region. Findings suggest important opportunity costs of state ownership as firms with state participation seem to achieve lower labor productivity outcomes and subpar financial performance in terms of profitability and return on investments. Yet, BOSs seem to be more financially leveraged and rely more on external debt compared to private peers suggesting potential soft budget constraints. This paper suggests important trade-offs between state ownership and firm performance and highlights some policy implications that are relevant when designing and embarking on reforms. First, the findings highlight that the private sector can play an important role in driving operational and financial performance. Firms with private shareholding seem to obtain better outcomes in terms of labor productivity as well as profit margins, and higher returns on assets and equity, and are less likely to support operations through external debt. Compared to firms in which the state is the sole owner, the results of this paper are indicative that private ownership can instill market- based incentives to improve performance.29 It has important policy implications. Reforms can explore alternatives to crowd-in private participation beyond full divestiture or privatization through public-private partnerships or joint ventures (Alfalter & Sanchez-Navarro, 2023). The findings emphasize that private ownership can help mitigate principal-agent problems by strengthening supervisory power and weakening the government’s control of day-to-day management. Second, the paper finds evidence of the potential opportunity costs of state ownership that policy makers should consider when determining in which sectors to operate. While state ownership can address market failures and provide public services, it also seems to curb operational and financial performance for firms operating in competitive markets where the economic rationale for state ownership is less clear. Compared to POEs, BOSs obtain lower results in terms of labor productivity, profit margins, liquidity, and return on investments across all sector types but achieve lower profit margins and returns on investments when the state invests in competitive sectors. BOSs in natural monopolies such as electricity transmission and utilities tend to exhibit lower levels of labor productivity and higher levels of debt against their assets vis-à-vis POEs, which could be linked to potential public service obligations (PSOs). These findings highlight the importance of differentiating by markets in which BOSs operate and underscore the importance of understanding ex-ante what the economic rationale for state ownership and whether the benefits (e.g., provision of PSOs) offset the potential costs in terms of reduced profits and returns on 29 It can occur through the transference of skills, management, and ‘know-how’, which may translate into better financial outcomes. 34 investments in competitive sectors vis-à-vis others. Policy tools such as the principles of subsidiarity can serve for this purpose.30 Third, this study highlights that a higher degree of separation from the government can serve to mitigate principal-agent issues and improve operational and financial performance. Even though we do not measure corporate governance arrangements such as appointment of board members, independence, and other key aspects, our proxy of autonomy based on the distance from the government (i.e., layers of ownership) serves as an entry point and offers some indicative evidence. Compared to other BOSs, indirectly owned BOSs are more productive and profitable, and achieve better returns for each dollar invested in assets or equity. Moreover, as the distance from the public authority (i.e., layers of ownership) increases, BOSs seem to narrow the gap vis-à-vis private peers. This could be interpreted to mean that higher autonomy and separation of functions of the government from the day-to-day operations can lead to performance and operational gains. In that sense, policy measures that foster BOSs’ autonomy and a higher degree of separation of the state as owner such as corporate governance and ownership reforms could potentially improve accountability and transparency and translate into significant performance gains. Finally, we find that BOSs can play a counter-cyclical role in periods of crisis by protecting employment and job creation in the short term. However, we do not find significant evidence that BOSs maintain relatively higher levels of investment compared to private counterparts in response to economic shocks. Further analysis is required to explore the medium- and long-term adjustments and potential trade-offs in such counter-cyclical effects and long-term reallocation of resources and efficiency gains. Further studies are also needed to understand other policy implications of state ownership in terms of firm dynamism, misallocation, innovation, and productivity to understand the full extent of the implications of the role of the state in the economy. 30 Under the subsidiarity principle, the state has a subsidiarity duty to perform only in sectors that cannot be provided by the private sector. For more details on how to implement it, please refer to Alfalter & Sanchez-Navarro (2023) and the World Bank Markets and Competition policy Assessment Toolkit (World Bank Group, Forthcoming). 35 REFERENCES Abramov, A., Radygin, A., & Chernova, M. (2017 ). State-owned enterprises in the Russian market: Ownership structure and their role in the economy. Russian Journal of Economics Vol 3 (1) , 1-23. Abramov, A., Radygin, A., Entov, R., & Chernova, M. (2017). State ownership and efficiency characteristics. Russian Journal of Economics 3(2) , 129-157. Alfalter, S., & Sanchez-Navarro, D. (2023). Businesses Of The State (BOS) and Private Sector Development - A Policy Toolkit for Practitioners (English). World Bank Group. Amighini, A. A., Rabellotti, R., & Sanfilippo, M. (2013). Do Chinese state-owned and private enterprises differ in their internatinalization strategies? China Economic Review, 312- 325. Andrews, W. A., & Dowling, M. J. (1998). Explaining performance changes in newly privatized firms 35(5), . Journal of Management Studies, 601-617. Bai, C., Li, D., Tao, Z., & Wang, Y. (2000). A multitask theory of State Enterprise Reform. Journal of Comparative Economics, 716-738. Bajgar, M., Berlingieri, G., Calligaris, S., Criscuolo, C., & Timmis, J. (2020). Coverage and representativeness of ORBIS data. OECD Science, Technology, and Industry Working Papers. Bennett, & Johnson. (1979). Public versus Private Provision of Collective Goods and Services: Garbage Collection Revisited. Public Choice, 55-63. Birdsall, N., & Nellis, J. (2003). Winners and losers: Assessing the distributional impact of privatization. World Development, 1617-1633. Boardman, & Vining. (1989). Ownership and Performance in Competitive Environments: A Comparison of the Performance of Private, Mixed, and State-Owned Enterprises. The Journal of Law & Economics, 1-33 . Boardman, & Vining. (1992). Ownership versus Competition: Efficiency in Public Enterprise. Public Choice, 205-239. Boardman, A. E., Vining, A. R., & Weimer, D. L. (2016). The long-run effects of privatization on productivity: Evidence from Canada. . Journal of Policy Modeling, 38(6), , 1001- 1017. Bogart, D., & Chaudhary, L. (2015). Off the rails: Is state ownership bad for productivity? Journal of Comparative Economics, 997-1013. Boko, F., & Qin, Y. (2011). Failure of State-Owned Enterprises in Benin. International Conference on Management and Service Science, 1-4. 36 Boubakri , N., Chen , R., El Ghoul, S., Guedhami, O., & Nash, O. (2020). State ownership and stock liquidity: Evidence from privatization. Journal of Corporate Finance . Boycko, M., Shleifer, A., & Vishny, R. W. (1996). A Theory of Privatisation. The Economic Journal , 395. Retrieved from https://doi.org/10.2307/2235248 Bozec, R., Breton, G., & Coté, L. (2002). The performance of State-owned enterprises revisited. Financial Accountability & Management 18 (4). Brolhato, S., Cirera, X., & Martins-Neto, A. (2023). Businesses of the State in Brazil. Employment and Impact on Business Dynamism. World Bank manuscript. Campos, J. E., & Esfahani, H. S. (1996). Why and when do governments initiate public enterprise reform? World Bank Economic Review, 10(3), , 451-485. doi:https://doi.org/10.1093/wber/10.3.451 Chan, K. K., Chen, L., & Wong, N. (2018). New Zealand State-owned enterprises: is state- ownership detrimental to firm performance? . New Zealand Economic Papers, 52(2), , 170-184. Chan, K., Chen, L., & Wong, N. (2022). New Zealand State-Owned Enterprises: Is State- Ownership Detrimental to Firm Performance? New Zealand Economic Papers 52 (2), 170-184. Chen, X., Xie, E., & Van Essen, M. (2021). Performance feedback and firm's R&D frequency: a comparison between state-owned and private-owned enterprises in China. Asian Business and Management 20 (2), 221-258. Chun, R. (2009). Ethical values and environmentalism in China: comparing employees from state-owned and private firms. Journal of Business Ethics 84 (3), 341-348. Cook, & Kirkpatrick. (1988). Privatisation in Less Developed Countries An Overview. Cornett, M., Guo, L., Khaksari, S., & Tehranian, H. (2010). The impact of state ownership on perfromance differences in privately-owned versis state-owned banks: an international comparison. Journal of Financial intermediation, 74-94. Cusolito, A. (2020). WBG Orbis Financial module. User's Guide and Global Trends. WBG Manuscript. Cusolito, A., Fattal-Jaef, R., Patiño Peña, F., & Singh, A. (2022). The Financial Premium and Real Cost of Bureaucrats in Businesses. World Bank Manuscript. Dall'Olio, A., Goodwin, T., Martinez Licetti, M., Alonso, A., Drodz, M., Orlowski, J., . . . Sanchez-Navarro, D. (2022b). Are all SOEs equal? World Bank working papers. Dall'Olio, A., Goodwin, T., Orlowski, J., Patino-Pena, F., Sanchez-Navarro, D., & Ralambotsiferana , F. (2022). Using ORBIS to Build a Global Database of Firms with State Participation. WB Working Paper Series WB 10261. Davies, D. G. (1977). Property rights and economic efficiency--the Australian airlines revisited. The Journal of Law and Economics, 20(1), 223-226. 37 De Wenter, K., & Malatesta, P. (2001). State-Owned and Privately Owned Firms: An Empirical Analysis of Profitability, Leverage, and Labor Intensity. American Economic Review, 320-334. Dewenter, K., & Malatesta, P. (2001). State-Owned and Privately Owned Firms: An Empirical Analysis of Profitability, Leverage and Labor Intensity. The American Economic Review, 320-334. Dewenter, K., & Malatesta, P. (2001). State-Owned and Privately Owned Firms: An empirical analysis of profitability, leverage, and labor intensity. The American Economic Review, 320-334. Dollar, D., & Wei, S.-J. (2007). Das (Wasted) Kapital: Firm Ownership and Investment Efficiency in China. NBER Working Paper 13103. D'souza, J., & Megginson, W. L. (1999). The financial and operating performance of privatized firms during the 1990s. The journal of Finance, 54(4),, 1397-1438. EBRD. (2020). The State Strikes Back: Transition Report 2020-21. Ferro, E., & Patiño-Peña, F. (2023). Private sector performance under the pressure of SOE competition - The case of Ecuador. World Bank Manuscript. Funkhouser, R., & MacAvoy, P. W. (1979). A sample of observations on comparative prices in public and private enterprises. Journal of Public Economics,11(3), 353-368. Goldeng, E., Grünfeld, L., & Benito, G. (2008). The Performance Differential between Private and State Owned Enterprises: The Roles of Ownership, Management and Market Structure. Journal of Management Studies. Grout, P. A., & Stevens, M. (2003 ). he assessment: financing and managing public services. . Oxford Review of economic policy, 19(2), , 215-234. . Guan, J., Gao, Z., Tan, J., Sun, W., & Shi, F. (2021). Does the mixed ownership reform work? Influence of board chair on performance of state-owned enterprises. Journal of Business Research Vol. 122, 51–59. Gunasekar, S., & Sarkar, J. (2019). Does autonomy mater in state owned enterprises? Evidence from performance contracts in India. The Economics of Transition and Institutional Change 27 (3), 763-800. Halkos, G. (2002). State owned enterprises, privatization and the public interest: evidence of SOE performance in the Greek manufacturing. Hermuningsih, Kusuma, Fasa, Iqbl, & Roymon. (2020). Inside Ownership, Financial Performance and Firm Size as Moderating Variables. Utopía y Praxis Latinoamericana, 220-231. Retrieved from https://www.redalyc.org/journal/279/27964799024/27964799024.pdf Hermuningsih, S., Kusuma, H., & Anzalia Cahyarif, R. (2020). Corporate Governance and Firm Performance: An Empirical Study from Indonesian Manufacturing Firm. Journal of Asian Finance, Economics and Business, 827-834. 38 IMF. (2019). Reassessing the role of state-owned enterprieses in Central, Eastern, and Southeastern Europe. IMF Europan Department (Series). Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. . Journal of Financial Economics. Kabaciński, B., Kubiak, J., & Szarzec, K. (2020). Do State-owned Enterprises Underperform Compared to Privately owned Companies? An Examination of the Largest Polish Enterprises. Emerging Markets Finance and Trade, 3174-3192. Kalemli-Ozcan, S., Sorensen, B., Villegas-Sanchez, C., Volosovych, V., & Yesiltas, S. (2015). Kalemli-Ozcan, S., Sorensen, B., Villegas-Sanchez, C., Volosovych, V., & Yesiltas, S. National Bureau of Economic Research. Kalemli-Ozcan, S., Sorensen,, B., Villegas-Sanchez, C., Volosovych, V., & Yelsitas, S. (2015). How to Construct Nationally Representative Firm Level Data from the Orbis Global Database: New facts and Aggregate Implications. . NBER Working Papers 21558, National Bureau of Economic Research. Kang, Y., & Kim, B. (2012). Ownership structure and firm performance: Evidence from the Chinese corporate reform,. China Economic Review Vol 23 (2), 471-481. Kathryn, L., Dewenter, L., & Malatesta, P. (2001). State-owned and privately owned firms: An Empirical Analysis of Profitability, Leverage, and Labor Intensity. American Economic Review 9 (1), 320-334. Kočenda, E., & Svejnar, J. (2003). Ownership and firm performance after large-. William Davidson Institute Working Paper (471a). Kornai, J., Maskin, E., & Roland, G. (2003). Understanding the Soft Budget Constraint. Journal of Economic Literature, 1095-1136. Lawson, C. (1994). The Theory of State-owned enterprises in market economies. The Journal of Economic Surveys, 283-309. Lazzarini, S., & Mussacchio, A. (2018). State ownership reinvented? Explaining performance differences between state-owned and private firms. Corporate Governance: An international Review, 255-272. Le, , T., & Buck, T. (2011). State ownership and listed firm performance: a universally negative governance relationshihp? Journal of Management & Governance 15, 227-248. Le, T. (2021). Performance comparison of state-owned enterprises versus private firms in selected emerging Asian countries. Journal of Asian Business and Economic Studies. Le, T., & Buck, T. (2009). State ownership and listed firm performance: a universally negative governance relationship? Journal of Management & Governance 15, 227-248. Liljeblom, E., Maury, B., & Hörhammer, A. (2020). Complex state ownership, comopetition, and firm performance - Russian evidence. International Journal of Emerging Markets 15 (2) , 189-221. 39 Liljeblom, E., Maury, B., & Hörhammer, A. (2020). Complex state ownership, competition, and firm performance – Russian evidence. International Journal of Emerging Markets, 189- 221. Lin, S., & Rowe, W. (2006). Determinants of the profitability of China's regional SOEs. China Economic Review, 120-141. Lu, C. Q., Siu, O. L., Au, W., & Leung, S. S. (2009). Manager's occupational streess in state- owned and private enterprises in the People's Republic of China. The International Journal of Human Resource Management 20 (8), 1670-1682. Matuszak, P., & Kabaciński, B. (2021). Non-commercial goals and financial performance of state-owned enterprises – some evidence from the electricity sector in the EU countries. Journal of Comparative Economics Vol 9 (4) , 1068-1087. McColgan, P. (2001). Agency theory and corporate governance: a review of the literature from a UK perspective. Megginson, Nash, & Randenborgh, V. (1994). The Financial and Operating Performance of Newly Privatized Firms: An International Empirical Analysis. In The Journal of Finance, 403–452. Megginson, W., & Netter, J. (2001). From State to Market: A Survey of Empirical Studies on Privatization. Journal of Economic Literature, 321-289. My Tran, N., Nonneman, W., & Jorissen, A. (2014). Government Ownership and Firm Performance: The Case of Vietnam. International Journal of Economics and Financial Issues Vol 4 (3), 628-650. Nahadi, B., & Suzuki, Y. (2012 ). Partial Privatization and Performance of Privatized Soes: The evidence from Indonesia. Journal of Economics and Sustainable Development 3 (14) . Nguyen , T., Moslehpour, M., Van Vo, T., & Wong, W.-K. (2020). State Ownership and Risk- Taking Behavior: An empirical approach to get better profitablity, investment, and trading strategies for listed corporates in Vietnam. Economies 8 (2). Nguyen, B., Do, H., & Le, C. (2021). How much state ownership do hybrid firms need for better performance? Small Bus Econ. OECD. (2020). The COVID-19 crisis and state ownership in the economy: issues and policy considerations. Okhmatovskiy, I. (2010). Performance Implications of Ties to the Government and SOEs: A Political Embeddedness Perspective. Journal of Management Studies. Panizza, U. (2021). State-owned Commercial Banks. Graduate Institute of International and Development Studies - Working Paper Series. Parida, M., & Madheswaran, S. (2021). Does ownership matter? Empirical evidence from the performance of Indian state and private coal mining companies. Resources Policy Vol 74. 40 Phi, N., Taghizadeh-Hesary, C. A., Tu, N., Yoshino, N., & Kim, C. J. (2019). Performance differential between private and state-owned enterprises: and analysis of profitability and leverage. ADBI Working Paper Series. Picot, A., & Kaulmann, T. (1989). Comparative performance of government-owned and privately-owned industrial corporations-Empirical results from six countries. . Journal of Institutional and Theoretical Economics (JITE)/Zeitschrift für die gesamte Staatswis. Plane, P. (1992). Productive efficiency of public enterprises: a macroeconomic analysis based on cross-section estimation of a neoclassical production function. . Applied Economics, 24(8), , 833-844. Pratuckchai, W., & Patanapongse, W. (2012). THE STUDY OF MANAGEMENT CONTROL SYSTEMS IN STATE OWNED ENTERPRISES: A PROPOSED CONCEPTUAL FRAMEWORK. International Journal of Organizational Innovation Vol. 5 (2), 83-115. Savas. (1977). The Organization and Efficiency of Solid Waste Collection. Shahab, Y., Ntim, C., & Ullah, F. (2019). The brighter side of being socially responsible: CS ratings and financial distress among Chinese state and non-state owned firms. Applied Economic Letters 26 (3), 180-186. Shang, D., Yuan, D., Li, D., & Fan, L. (2022). The Effects of non-state shareholder on the excess perquisites: evidence from Chineses mixed ownership reform. Emerging markets finance and trade, 4001-4013. Shirley, M., & P., W. (2000). Public versus Private Ownership: The Current State of the Debate. Sun, Q., Zhang, A., & Li, J. (2005). A study of optimal state shares in mixed oligopoly: Implications for SOE reform and foreign competition. China Economic Review 16, 1-27. Szanyi, M. (2016). The revesal of the privatisation logic in central european transition economies. . Acta Oeconomica 66 (1): 33-55. Szarzec, K. T., & Piątek, D. H. (2022). ow do politicians capture a state? Evidence from state- owned enterprises. . East European Politics and Societies, , 36(1), 141-172. Telegdy, A. (2016). Employment adjustment in the global crisis. Economics of Transition and Institutional Change, 683-703. Tihanyi, L., Aguilera, R., Heugens, P., Van Essen, M., Sauwewald, S., Duran, P., & Turturea, R. (2019). State Ownership and Political Connections. Journal of Management 45 (6), 2293- 2321. Totleben, B. S., & Kardziejonek, A. (2019). Rent-seeking by politicians in state-owned enterprises. Ekonomia i Prawo (Vol. 18, Issue 4, p. 515). Uniwersytet Mikolaja Kopernika/Nicolaus Copernicus University. , 515. doi:https://doi.org/10.12775/eip.2019. Uddin, M. (2016). Effect of Government Share Ownership on Corporate Risk Taking: The case of the United Arab Emirates. Research in International Business and Finance, 322-39. 41 UNDP. (2000). Joint Venture Public-Private Partnerships for Urban Environmental Services. New York. Vining, A. R., & Boardman, A. E. (1992). Ownership versus competition: Efficiency in public enterprise. Public choice, 73(2),, 205-239. Wang, X., & Colin, L. (2004). State-owned enterprises going public - The case of China. Economics of Transition and Institutional Change Vol 12 (3), 467-487. WBG-IEG. (2018). World Bank Group Support for the Reform of State-Owned Enterprises, 2007-2018: An IEG Evaluation. Approach Paper, World Bank, IFC. MIGA. WBG. World Bank. (2022). Who's the BOSs: Shedding new light on the state footprint in markets. World Bank. World Bank. (2023). Business of the State. World Bank. World Bank Group. (Forthcoming). Markets and Competition Policy Assessment Toolkit - MCPAT. WBG. Wu, G.-L. (2011). Can minority state ownership influence firm value? Universal and contingency views of its governance effects. Journal of Business Research Vol 64 (8). , 839-845. Yu, M. (2013). State ownership and firm performance: Empirical evidence from Chinese listed companies. China Journal of Accounting Research. Zaim, O., & Taskin, F. (1997). The comparative performance of the public enterprise sector in Turkey: a malmquist productivity index approach. . Journal of comparative Economics, 25(2), , 129-157. Zheng Zhou, K., Yong Gao, G., & Zhao, H. (2017). State Ownership and Firm Innovation in China: An integrated view of institutional and efficiency logics. Administrative Science Quarterly Vol. 62 (2), 375-404. 42 ANNEXES Annex 1. Sample distribution BOS vis-a-vis POEs by country and year Country/year POEs BOSs Total firms Share BOS in total firms BA 66,474 2,707 69,181 3.9% 2011 5,753 263 6,016 4.4% 2012 6,160 268 6,428 4.2% 2013 5,920 271 6,191 4.4% 2014 5,888 270 6,158 4.4% 2015 6,958 270 7,228 3.7% 2016 7,058 271 7,329 3.7% 2017 6,619 264 6,883 3.8% 2018 7,288 278 7,566 3.7% 2019 7,359 279 7,638 3.7% 2020 7,471 273 7,744 3.5% BG 633,625 3,435 637,060 0.5% 2011 58,607 336 58,943 0.6% 2012 48,895 331 49,226 0.7% 2013 60,190 345 60,535 0.6% 2014 46,666 323 46,989 0.7% 2015 70,470 347 70,817 0.5% 2016 76,482 368 76,850 0.5% 2017 70,884 353 71,237 0.5% 2018 71,053 347 71,400 0.5% 2019 69,440 343 69,783 0.5% 2020 60,938 342 61,280 0.6% EE 633,425 1,242 634,667 0.2% 2011 36,100 120 36,220 0.3% 2012 41,135 120 41,255 0.3% 2013 46,574 113 46,687 0.2% 2014 53,290 117 53,407 0.2% 2015 60,300 121 60,421 0.2% 2016 69,016 126 69,142 0.2% 2017 76,756 129 76,885 0.2% 2018 83,339 137 83,476 0.2% 2019 89,552 138 89,690 0.2% 2020 77,363 121 77,484 0.2% ES 2,887,762 13,239 2,901,001 0.5% 2011 229,148 1,236 230,384 0.5% 2012 243,484 1,260 244,744 0.5% 2013 261,307 1,305 262,612 0.5% 2014 274,198 1,362 275,560 0.5% 2015 292,731 1,377 294,108 0.5% 43 2016 302,414 1,388 303,802 0.5% 2017 318,349 1,434 319,783 0.4% 2018 354,189 1,424 355,613 0.4% 2019 354,233 1,365 355,598 0.4% 2020 257,709 1,088 258,797 0.4% HR 323,093 4,988 328,081 1.5% 2011 20,124 389 20,513 1.9% 2012 23,440 420 23,860 1.8% 2013 25,284 439 25,723 1.7% 2014 27,459 482 27,941 1.7% 2015 30,568 509 31,077 1.6% 2016 33,058 522 33,580 1.6% 2017 36,551 529 37,080 1.4% 2018 40,514 569 41,083 1.4% 2019 43,339 563 43,902 1.3% 2020 42,756 566 43,322 1.3% IT 4,011,180 34,475 4,045,655 0.9% 2011 334,203 3,033 337,236 0.9% 2012 342,294 3,086 345,380 0.9% 2013 352,885 3,260 356,145 0.9% 2014 357,327 3,285 360,612 0.9% 2015 374,763 3,376 378,139 0.9% 2016 402,606 3,533 406,139 0.9% 2017 434,584 3,668 438,252 0.8% 2018 477,612 3,845 481,457 0.8% 2019 502,279 3,885 506,164 0.8% 2020 432,627 3,504 436,131 0.8% LT 145,494 951 146,445 0.6% 2011 10,406 87 10,493 0.8% 2012 11,037 92 11,129 0.8% 2013 11,902 90 11,992 0.8% 2014 12,773 96 12,869 0.7% 2015 13,227 96 13,323 0.7% 2016 12,934 97 13,031 0.7% 2017 14,572 97 14,669 0.7% 2018 16,542 102 16,644 0.6% 2019 21,542 100 21,642 0.5% 2020 20,559 94 20,653 0.5% LV 193,516 2,706 196,222 1.4% 2011 12,148 258 12,406 2.1% 2012 14,387 270 14,657 1.8% 2013 15,111 268 15,379 1.7% 2014 16,393 261 16,654 1.6% 44 2015 17,831 269 18,100 1.5% 2016 19,812 277 20,089 1.4% 2017 22,630 275 22,905 1.2% 2018 24,654 277 24,931 1.1% 2019 25,257 276 25,533 1.1% 2020 25,293 275 25,568 1.1% MK 100,179 1,255 101,434 1.2% 2011 4,199 104 4,303 2.4% 2012 6,288 113 6,401 1.8% 2013 6,790 121 6,911 1.8% 2014 8,667 126 8,793 1.4% 2015 9,337 131 9,468 1.4% 2016 10,385 129 10,514 1.2% 2017 11,605 132 11,737 1.1% 2018 12,933 131 13,064 1.0% 2019 14,269 133 14,402 0.9% 2020 15,706 135 15,841 0.9% PL 1,281,569 27,029 1,308,598 2.1% 2011 71,857 2,366 74,223 3.2% 2012 76,215 2,424 78,639 3.1% 2013 82,781 2,482 85,263 2.9% 2014 89,293 2,534 91,827 2.8% 2015 96,612 2,608 99,220 2.6% 2016 98,817 2,678 101,495 2.6% 2017 170,900 2,916 173,816 1.7% 2018 196,999 3,055 200,054 1.5% 2019 210,179 3,063 213,242 1.4% 2020 187,916 2,903 190,819 1.5% PT 647,077 4,058 651,135 0.6% 2011 46,436 366 46,802 0.8% 2012 49,158 376 49,534 0.8% 2013 52,428 389 52,817 0.7% 2014 56,348 405 56,753 0.7% 2015 60,836 412 61,248 0.7% 2016 65,896 420 66,316 0.6% 2017 71,424 427 71,851 0.6% 2018 77,761 425 78,186 0.5% 2019 82,355 440 82,795 0.5% 2020 84,435 398 84,833 0.5% RO 2,138,839 7,617 2,146,456 0.4% 2011 131,179 640 131,819 0.5% 2012 150,175 694 150,869 0.5% 2013 168,376 720 169,096 0.4% 45 2014 184,002 748 184,750 0.4% 2015 191,534 765 192,299 0.4% 2016 206,311 790 207,101 0.4% 2017 238,208 810 239,018 0.3% 2018 264,322 787 265,109 0.3% 2019 287,607 830 288,437 0.3% 2020 317,125 833 317,958 0.3% RS 337,883 5,941 343,824 1.7% 2011 8,751 515 9,266 5.6% 2012 12,639 545 13,184 4.1% 2013 23,341 568 23,909 2.4% 2014 23,014 553 23,567 2.3% 2015 20,392 459 20,851 2.2% 2016 43,715 700 44,415 1.6% 2017 34,305 655 34,960 1.9% 2018 35,434 653 36,087 1.8% 2019 66,996 649 67,645 1.0% 2020 69,296 644 69,940 0.9% RU 4,531,932 133,689 4,665,621 2.9% 2011 204,886 9,616 214,502 4.5% 2012 250,743 10,588 261,331 4.1% 2013 302,107 11,144 313,251 3.6% 2014 451,988 12,506 464,494 2.7% 2015 634,318 14,170 648,488 2.2% 2016 779,576 15,123 794,699 1.9% 2017 404,852 14,723 419,575 3.5% 2018 444,966 15,522 460,488 3.4% 2019 457,351 14,890 472,241 3.2% 2020 601,145 15,407 616,552 2.5% SI 504,828 3,636 508,464 0.7% 2011 37,951 331 38,282 0.9% 2012 44,796 347 45,143 0.8% 2013 45,892 352 46,244 0.8% 2014 49,820 349 50,169 0.7% 2015 50,535 354 50,889 0.7% 2016 50,086 351 50,437 0.7% 2017 52,756 362 53,118 0.7% 2018 54,290 388 54,678 0.7% 2019 57,405 405 57,810 0.7% 2020 61,297 397 61,694 0.6% UA 1,253,115 26,373 1,279,488 2.1% 2011 113,335 2,874 116,209 2.5% 2012 115,208 2,880 118,088 2.4% 46 2013 110,952 2,792 113,744 2.5% 2014 106,719 2,502 109,221 2.3% 2015 116,283 2,492 118,775 2.1% 2016 121,696 2,445 124,141 2.0% 2017 111,997 2,444 114,441 2.1% 2018 118,481 2,413 120,894 2.0% 2019 168,826 2,819 171,645 1.6% 2020 169,618 2,712 172,330 1.6% Annex 2. Robustness checks: estimates for 5-year window (2016-2020) A.2.1 Labor productivity (1) (2) (3) (4) (5) (6) (6) (8) (9) VARIABLES BOS -0.524*** -0.488*** -0.481*** -0.481*** -0.481*** -0.476*** -0.476*** -0.476*** -0.476*** (0.00411) (0.00420) (0.00420) (0.00420) (0.00420) (0.00432) (0.00433) (0.00766) (0.0340) Size 0.373*** 0.392*** 0.393*** 0.393*** 0.393*** 0.385*** 0.385*** 0.385*** 0.385*** (0.000476) (0.000497) (0.000498) (0.000499) (0.000498) (0.000504) (0.000505) (0.000786) (0.0120) Age -0.0134*** -0.0133*** -0.0133*** -0.0133*** -0.0133*** -0.0133*** -0.0133*** -0.0133*** (6.02e-05) (6.01e-05) (6.01e-05) (6.00e-05) (6.00e-05) (6.02e-05) (0.000101) (0.00107) Listed status -0.569*** -0.568*** -0.568*** -0.551*** -0.552*** -0.552*** -0.552*** (0.0128) (0.0128) (0.0128) (0.0128) (0.0129) (0.0230) (0.0411) Constant 5.596*** 5.537*** 5.521*** 5.523*** 5.523*** 5.624*** 5.622*** 5.622*** 5.622*** (0.00652) (0.00660) (0.00661) (0.00661) (0.00660) (0.00668) (0.00669) (0.0103) (0.150) Observations 2,974,035 2,895,615 2,895,615 2,895,615 2,895,610 2,895,534 2,892,917 2,892,917 2,892,917 R-squared 0.667 0.671 0.671 0.671 0.673 0.693 0.694 0.694 0.694 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 A.2.2 Profit margin (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS -6.969*** -7.275*** -7.276*** -7.265*** -7.252*** -7.479*** -7.486*** -7.486*** -7.486*** (0.0814) (0.0835) (0.0835) (0.0835) (0.0834) (0.0854) (0.0857) (0.133) (0.517) Size -0.569*** -0.676*** -0.676*** -0.679*** -0.678*** -0.648*** -0.652*** -0.652*** -0.652*** (0.0103) (0.0107) (0.0108) (0.0108) (0.0107) (0.0109) (0.0109) (0.0155) (0.0786) Age 0.0476*** 0.0476*** 0.0478*** 0.0477*** 0.0363*** 0.0363*** 0.0363*** 0.0363*** (0.00109) (0.00109) (0.00109) (0.00108) (0.00110) (0.00110) (0.00162) (0.0108) Listed status 0.137 0.180 0.175 0.636** 0.669** 0.669 0.669 (0.285) (0.285) (0.283) (0.285) (0.284) (0.426) (0.427) 47 Constant 7.906*** 7.558*** 7.560*** 7.566*** 7.564*** 7.638*** 7.652*** 7.652*** 7.652*** (0.0371) (0.0387) (0.0388) (0.0390) (0.0388) (0.0391) (0.0391) (0.0546) (0.191) Observations 2,925,822 2,851,305 2,851,305 2,851,305 2,851,299 2,851,235 2,848,595 2,848,595 2,848,595 R-squared 0.078 0.073 0.073 0.074 0.082 0.109 0.115 0.115 0.115 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 A.2.3 Current ratio (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS -0.215*** -0.340*** -0.341*** -0.339*** -0.334*** -0.338*** -0.337*** -0.337*** -0.337*** (0.0272) (0.0282) (0.0282) (0.0282) (0.0282) (0.0296) (0.0297) (0.0451) (0.0785) Size -0.478*** -0.580*** -0.580*** -0.585*** -0.588*** -0.588*** -0.591*** -0.591*** -0.591*** (0.00386) (0.00412) (0.00413) (0.00414) (0.00414) (0.00425) (0.00427) (0.00602) (0.0163) Age 0.0566*** 0.0566*** 0.0568*** 0.0567*** 0.0544*** 0.0544*** 0.0544*** 0.0544*** (0.000392) (0.000392) (0.000392) (0.000392) (0.000403) (0.000404) (0.000597) (0.00232) Listed status 0.0929 0.0990 0.104 0.222*** 0.230*** 0.230** 0.230*** (0.0691) (0.0691) (0.0692) (0.0718) (0.0719) (0.114) (0.0742) Constant 5.060*** 4.582*** 4.583*** 4.597*** 4.609*** 4.644*** 4.653*** 4.653*** 4.653*** (0.0147) (0.0149) (0.0150) (0.0150) (0.0150) (0.0153) (0.0153) (0.0214) (0.0555) Observations 2,950,077 2,875,396 2,875,396 2,875,396 2,875,392 2,875,326 2,872,670 2,872,670 2,872,670 R-squared 0.051 0.056 0.056 0.057 0.057 0.070 0.072 0.072 0.072 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 A.2.4 Solvency ratio (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS 6.627*** 4.993*** 4.946*** 4.961*** 4.974*** 4.710*** 4.710*** 4.710*** 4.710*** (0.135) (0.139) (0.139) (0.139) (0.139) (0.143) (0.143) (0.249) (0.468) Size -0.848*** -2.219*** -2.239*** -2.271*** -2.278*** -2.164*** -2.175*** -2.175*** -2.175*** (0.0207) (0.0212) (0.0213) (0.0213) (0.0213) (0.0217) (0.0218) (0.0357) (0.112) Age 0.765*** 0.764*** 0.765*** 0.765*** 0.731*** 0.731*** 0.731*** 0.731*** (0.00197) (0.00197) (0.00197) (0.00197) (0.00201) (0.00202) (0.00337) (0.0114) Listed status 4.650*** 4.687*** 4.687*** 5.913*** 5.895*** 5.895*** 5.895*** (0.368) (0.368) (0.368) (0.378) (0.379) (0.678) (0.516) 48 Constant 44.13*** 37.42*** 37.48*** 37.58*** 37.60*** 37.72*** 37.76*** 37.76*** 37.76*** (0.0734) (0.0756) (0.0758) (0.0759) (0.0760) (0.0769) (0.0771) (0.123) (0.383) Observations 2,943,260 2,868,110 2,868,110 2,868,110 2,868,105 2,868,038 2,865,387 2,865,387 2,865,387 R-squared 0.076 0.118 0.118 0.119 0.120 0.144 0.146 0.146 0.146 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 A.2.5 Return on Assets (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS -5.623*** -5.599*** -5.565*** -5.561*** -5.557*** -5.663*** -5.667*** -5.667*** -5.667*** (0.0629) (0.0651) (0.0652) (0.0652) (0.0651) (0.0681) (0.0684) (0.0923) (0.147) Size -0.704*** -0.627*** -0.613*** -0.609*** -0.600*** -0.537*** -0.535*** -0.535*** -0.535*** (0.0104) (0.0109) (0.0110) (0.0110) (0.0110) (0.0112) (0.0113) (0.0151) (0.0378) Age -0.0530*** -0.0525*** -0.0527*** -0.0527*** -0.0636*** -0.0636*** -0.0636*** -0.0636*** (0.00103) (0.00103) (0.00103) (0.00103) (0.00105) (0.00105) (0.00140) (0.00398) Listed status -3.327*** -3.269*** -3.287*** -2.772*** -2.746*** -2.746*** -2.746*** (0.163) (0.163) (0.163) (0.172) (0.172) (0.246) (0.293) Constant 8.934*** 9.502*** 9.456*** 9.447*** 9.415*** 9.365*** 9.362*** 9.362*** 9.362*** (0.0383) (0.0408) (0.0409) (0.0410) (0.0409) (0.0416) (0.0417) (0.0544) (0.151) Observations 2,948,616 2,873,275 2,873,275 2,873,275 2,873,271 2,873,206 2,870,563 2,870,563 2,870,563 R-squared 0.045 0.045 0.046 0.047 0.052 0.069 0.075 0.075 0.075 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 A.2.6 Return on Equity (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS -17.73*** -16.95*** -16.86*** -16.86*** -16.83*** -17.04*** -17.05*** -17.05*** -17.05*** (0.284) (0.294) (0.294) (0.294) (0.294) (0.309) (0.309) (0.365) (0.520) Size -2.234*** -1.344*** -1.306*** -1.265*** -1.242*** -1.040*** -1.034*** -1.034*** -1.034*** (0.0483) (0.0510) (0.0512) (0.0513) (0.0512) (0.0527) (0.0528) (0.0635) (0.0872) Age -0.522*** -0.521*** -0.522*** -0.520*** -0.550*** -0.549*** -0.549*** -0.549*** (0.00483) (0.00483) (0.00483) (0.00482) (0.00493) (0.00493) (0.00581) (0.0199) Listed status -8.741*** -8.600*** -8.641*** -8.151*** -8.024*** -8.024*** -8.024*** (0.654) (0.653) (0.653) (0.707) (0.703) (0.851) (0.888) Constant 26.62*** 31.67*** 31.55*** 31.43*** 31.32*** 31.09*** 31.07*** 31.07*** 31.07*** 49 (0.172) (0.185) (0.185) (0.186) (0.185) (0.189) (0.189) (0.223) (0.330) Observations 2,723,996 2,652,758 2,652,758 2,652,758 2,652,749 2,652,669 2,649,938 2,649,938 2,649,938 R-squared 0.026 0.030 0.030 0.030 0.034 0.041 0.047 0.047 0.047 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Annex 3. Robustness checks: estimates for 3-year window (2018-2020) A.3.1 Labor productivity (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS -0.523*** -0.487*** -0.481*** -0.481*** -0.481*** -0.480*** -0.480*** -0.480*** -0.480*** (0.00595) (0.00608) (0.00608) (0.00608) (0.00608) (0.00626) (0.00627) (0.00837) (0.0517) Size 0.373*** 0.392*** 0.393*** 0.393*** 0.393*** 0.385*** 0.385*** 0.385*** 0.385*** (0.000680) (0.000710) (0.000711) (0.000712) (0.000711) (0.000719) (0.000720) (0.000896) (0.0178) Age -0.0133*** -0.0132*** -0.0132*** -0.0132*** -0.0131*** -0.0131*** -0.0131*** -0.0131*** (8.39e-05) (8.39e-05) (8.39e-05) (8.36e-05) (8.37e-05) (8.38e-05) (0.000109) (0.00160) Listed status -0.562*** -0.562*** -0.562*** -0.545*** -0.545*** -0.545*** -0.545*** (0.0185) (0.0185) (0.0184) (0.0186) (0.0186) (0.0249) (0.0574) Constant 5.578*** 5.527*** 5.512*** 5.514*** 5.511*** 5.620*** 5.618*** 5.618*** 5.618*** (0.00933) (0.00943) (0.00945) (0.00945) (0.00944) (0.00954) (0.00955) (0.0118) (0.222) Observations 1,518,928 1,477,624 1,477,624 1,477,624 1,477,621 1,477,518 1,476,309 1,476,309 1,476,309 R-squared 0.661 0.664 0.665 0.665 0.666 0.687 0.687 0.687 0.687 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 A.3.2 Profit margin (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS -7.343*** -7.637*** -7.638*** -7.637*** -7.632*** -7.860*** -7.859*** -7.859*** -7.859*** (0.119) (0.121) (0.121) (0.121) (0.121) (0.124) (0.124) (0.154) (0.768) Size -0.600*** -0.702*** -0.703*** -0.701*** -0.692*** -0.671*** -0.671*** -0.671*** -0.671*** (0.0147) (0.0153) (0.0153) (0.0153) (0.0152) (0.0153) (0.0153) (0.0183) (0.102) Age 0.0507*** 0.0507*** 0.0508*** 0.0510*** 0.0380*** 0.0379*** 0.0379*** 0.0379** (0.00154) (0.00154) (0.00154) (0.00153) (0.00156) (0.00156) (0.00188) (0.0154) Listed status 0.0829 0.0847 0.0803 0.589 0.611 0.611 0.611 50 (0.419) (0.419) (0.416) (0.420) (0.419) (0.503) (0.641) Constant 7.767*** 7.347*** 7.348*** 7.341*** 7.309*** 7.438*** 7.441*** 7.441*** 7.441*** (0.0526) (0.0550) (0.0551) (0.0551) (0.0548) (0.0551) (0.0550) (0.0652) (0.240) Observations 1,502,347 1,462,794 1,462,794 1,462,794 1,462,792 1,462,688 1,461,512 1,461,512 1,461,512 R-squared 0.086 0.082 0.082 0.083 0.093 0.122 0.126 0.126 0.126 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector- Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 A.3.3 Current ratio (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS -0.215*** -0.348*** -0.349*** -0.349*** -0.349*** -0.347*** -0.347*** -0.347*** -0.347*** (0.0388) (0.0401) (0.0402) (0.0402) (0.0402) (0.0423) (0.0424) (0.0519) (0.109) Size -0.496*** -0.600*** -0.600*** -0.600*** -0.600*** -0.604*** -0.605*** -0.605*** -0.605*** (0.00528) (0.00562) (0.00564) (0.00564) (0.00564) (0.00579) (0.00580) (0.00699) (0.0223) Age 0.0583*** 0.0583*** 0.0582*** 0.0582*** 0.0558*** 0.0558*** 0.0558*** 0.0558*** (0.000548) (0.000548) (0.000548) (0.000548) (0.000564) (0.000565) (0.000685) (0.00330) Listed status 0.0206 0.0204 0.0211 0.135 0.142 0.142 0.142 (0.0921) (0.0921) (0.0921) (0.0941) (0.0942) (0.117) (0.0963) Constant 5.158*** 4.646*** 4.646*** 4.648*** 4.649*** 4.698*** 4.701*** 4.701*** 4.701*** (0.0201) (0.0204) (0.0204) (0.0204) (0.0204) (0.0208) (0.0209) (0.0249) (0.0725) Observations 1,516,720 1,476,958 1,476,958 1,476,958 1,476,956 1,476,866 1,475,667 1,475,667 1,475,667 R-squared 0.050 0.056 0.056 0.056 0.056 0.071 0.072 0.072 0.072 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 A.3.4 Solvency ratio (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS 6.275*** 4.598*** 4.552*** 4.550*** 4.554*** 4.332*** 4.333*** 4.333*** 4.333*** (0.193) (0.197) (0.197) (0.197) (0.197) (0.203) (0.203) (0.270) (0.648) Size -0.954*** -2.309*** -2.328*** -2.329*** -2.324*** -2.221*** -2.224*** -2.224*** -2.224*** (0.0287) (0.0294) (0.0294) (0.0294) (0.0294) (0.0300) (0.0300) (0.0390) (0.149) Age 0.766*** 0.765*** 0.765*** 0.765*** 0.732*** 0.732*** 0.732*** 0.732*** (0.00273) (0.00273) (0.00272) (0.00272) (0.00278) (0.00279) (0.00362) (0.0159) 51 Listed status 4.611*** 4.613*** 4.601*** 5.779*** 5.772*** 5.772*** 5.772*** (0.524) (0.524) (0.524) (0.537) (0.538) (0.727) (0.745) Constant 44.77*** 37.75*** 37.80*** 37.81*** 37.79*** 37.95*** 37.96*** 37.96*** 37.96*** (0.101) (0.105) (0.105) (0.105) (0.105) (0.106) (0.106) (0.136) (0.501) Observations 1,510,209 1,470,243 1,470,243 1,470,243 1,470,240 1,470,144 1,468,951 1,468,951 1,468,951 R-squared 0.075 0.119 0.119 0.119 0.120 0.145 0.146 0.146 0.146 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 A.3.5 Return on Assets (1) (2) (3) (4) (5) (6) (7) (8) (9) BOS -5.867*** -5.847*** -5.815*** -5.816*** -5.812*** -5.893*** -5.898*** -5.898*** -5.898*** (0.0901) (0.0930) (0.0931) (0.0931) (0.0929) (0.0976) (0.0978) (0.112) (0.211) - Size -0.713*** -0.644*** -0.631*** -0.629*** -0.618*** -0.560*** -0.559*** -0.559*** 0.559*** (0.0147) (0.0154) (0.0155) (0.0155) (0.0154) (0.0158) (0.0158) (0.0183) (0.0479) - Age -0.046*** -0.046*** -0.046*** -0.045*** -0.056*** -0.056** -0.056*** 0.056*** (0.00146) (0.00146) (0.00146) (0.00145) (0.00148) (0.00148) (0.00169) (0.00524) - Listed status -3.265*** -3.262*** -3.278*** -2.692*** -2.687*** -2.687*** 2.687*** (0.236) (0.236) (0.236) (0.248) (0.248) (0.295) (0.379) Constant 8.777*** 9.303*** 9.261*** 9.253*** 9.210*** 9.171*** 9.173*** 9.173*** 9.173*** (0.0536) (0.0574) (0.0575) (0.0575) (0.0573) (0.0583) (0.0583) (0.0666) (0.194) Observations 1,511,792 1,471,766 1,471,766 1,471,766 1,471,764 1,471,670 1,470,483 1,470,483 1,470,483 R-squared 0.053 0.053 0.054 0.055 0.062 0.081 0.085 0.085 0.085 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 A.3.6 Return on Equity (1) (2) (3) (4) (5) (6) (7) (8) (9) VARIABLES BOS -18.12*** -17.37*** -17.28*** -17.27*** -17.25*** -17.45*** -17.47*** -17.47*** -17.47*** (0.413) (0.426) (0.427) (0.427) (0.426) (0.448) (0.448) (0.482) (0.782) Size -2.117*** -1.298*** -1.263*** -1.257*** -1.230*** -1.022*** -1.016*** -1.016*** -1.016*** (0.0692) (0.0733) (0.0735) (0.0735) (0.0733) (0.0753) (0.0753) (0.0823) (0.125) Age -0.489*** -0.488*** -0.488*** -0.484*** -0.513*** -0.513*** -0.513*** -0.513*** (0.00684) (0.00684) (0.00684) (0.00683) (0.00698) (0.00698) (0.00750) (0.0286) 52 Listed status -8.594*** -8.585*** -8.625*** -8.176*** -8.116*** -8.116*** -8.116*** (0.955) (0.954) (0.955) (1.028) (1.025) (1.124) (1.183) Constant 25.31*** 30.27*** 30.16*** 30.12*** 29.98*** 29.74*** 29.72*** 29.72*** 29.72*** (0.245) (0.263) (0.264) (0.264) (0.264) (0.269) (0.269) (0.291) (0.486) Observations 1,397,036 1,359,285 1,359,285 1,359,285 1,359,280 1,359,168 1,357,960 1,357,960 1,357,960 R-squared 0.031 0.034 0.034 0.034 0.038 0.048 0.052 0.052 0.052 Country FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Country-year FE No No No Yes Yes Yes Yes Yes Yes Sector-Year FE No No No No Yes Yes Yes Yes Yes Country-Sector FE No No No No No Yes Yes Yes Yes Country-Sector-Year FE No No No No No No Yes Yes Yes Clustered errors No No No No No No No Firm Sector 2d Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Annex 4. Test of statistical differences and significance of coefficients Table 17 Ho: BOS at x% participation = BOS at y% participation (Joint hypothesis F- test). Competitive Natural Monopoly Part. Contestable Indicator p-value p-value p-value F-value F-value F-value Labor productivity 70.07*** 0.0000 40.17*** 0.0000 38.24*** 0.0000 Profit margin 70.27*** 0.0000 138.56*** 0.0000 23.86*** 0.0000 Current ratio 18.74*** 0.0000 20.89*** 0.0000 4.13*** 0.0071 Solvency ratio 45.78*** 0.0000 147.54*** 0.0000 12.80*** 0.0000 ROA 50.85*** 0.0000 40.30*** 0.0000 45.83*** 0.0000 ROE 45.83*** 0.0000 43.20*** 0.0000 14.33*** 0.0000 53