Policy Research Working Paper 9998 FDI, Market Power, and Markups Evidence from Vietnam Yue Li Ryan Kuo Mauricio Pinzon-Latorre Mark Albertson Finance, Competitiveness and Innovation Global Practice April 2022 Policy Research Working Paper 9998 Abstract To date, the impact of foreign direct investment on market firms individually charging higher markups on average than power and consumer welfare in developing countries has their domestic competitors. The findings further show that been relatively understudied. Utilizing a firm survey dataset while the markups of both foreign- and domestic-owned from Vietnam, this paper first calculates firm-level markups private firms tend to decrease with greater foreign direct for manufacturing firms and then analyzes the impact of investment, state-owned enterprises may be relatively insu- foreign direct investment and foreign ownership on firm lated from foreign direct investment driven competitive markups. Overall, the findings show that increases in the pressures. These results are robust to the inclusion or exclu- presence of foreign firms in a given industry are associated sion of potential outliers and the potential non-random with decreases in markups in that industry, despite foreign selection of firms acquired by foreign investors. 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 authors may be contacted at rkuo@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 FDI, Market Power, and Markups Evidence from Vietnam Yue Li, Ryan Kuo, Mauricio Pinzon-Latorre, and Mark Albertson1 Keywords: Foreign Direct Investment, Competition Economics, Market Power, State-Owned Enterprises JEL classification: F21, F23, L11 1 We are grateful for feedback and comments from Trang Tran, Jesica Torres Coronado, and Seidu Dauda. I. Introduction Robust competition is core to well-functioning markets that work for development. Pressures from existing and prospective competitors force firms to lower prices, increase quality, and innovate new offerings, in turn benefitting consumer welfare. At the macroeconomic level, competition boosts productivity growth, fosters labor reallocation, and favors high-productivity firms. In the absence of such competition, firms are able to extract economic rents by leveraging their market power to raise prices, in turn hurting consumers and decreasing demand for labor (De Loecker, Eeckhout, & Unger, 2020). Concerns over market power and competition have increased in recent years. Over the past several years, scholars have documented an increase in the firms’ market power across countries and regions. For instance, De Loecker and Eeckhout (2020) find that the aggregate global markup has increased from close to 1.15 in 1980 to around 1.6 in 2016, leading to distributional consequences in terms of larger profits and lower labor shares. Moreover, recent challenges posed by the rise of digital platforms and a wave of pandemic-driven bankruptcies by smaller firms may also worsen the trend of increased market power in the future (Akcigit et al., 2021). Enriching policy makers’ and researchers’ understanding of the drivers of and barriers to competition is thus critical to development. Given that foreign investment is a key driver of economic growth in developing economies and the recent trend of rising global markups, it is critical to evaluate the effect of FDI on the markups of host economies and the consequences for consumer welfare. Theoretically, the impacts of FDI on market power are ambiguous because the presence of multinational enterprises (MNEs) could have conflicting effects on incumbent firms' market power: FDI could lower prices due to increased competitive pressure, but MNEs themselves may possess higher market power if they possess significant economies of scale to undercut local firms, force firm exit, and eventually reduce competition. To date, data availability has limited existing studies to advanced economies, and conclusions have been widely heterogeneous based on the country of interest. To address this gap, this paper evaluates the effect of FDI on market power within industries using firm- level panel data for Vietnam. We first apply the methodological work by De Loecker and Warzynski (2012) to estimate firm-level production functions and compute markups. Then, we estimate the impact of FDI on markups via a difference-in-differences estimation. We also analyze the impact of foreign ownership on individual firms’ markups by looking at differences in markups between foreign- and domestic-owned firms and analyzing the impact of acquisition and divestment by foreign investors. Our findings suggest that the presence of MNEs in a given industry decreases markups in that industry. We find that a 10 percent increase in foreign-owned firms’ share of output within a given two-digit manufacturing sector is associated with a 0.4 to 1.2 percentage point decrease in the ratio of price to marginal cost. This pro-competitive finding holds despite MNEs charging higher markups on average than domestic competitors. Furthermore, the results show that, while the markups of both foreign- and domestic-owned private firms tend to decrease with greater FDI presence, state-owned enterprises (SOEs) may be relatively insulated from FDI-driven competitive pressures. Our contribution is twofold. First, our work contributes to the scarce literature that evaluates the effects of FDI on market power in host countries, especially low- and middle-income contexts. Most country-level studies on this topic rely on accounting-based price-cost margins, concentration, or entry and exit dynamics to measure market power and focus on high-income settings (Abolhassani & Danakol, 2019; 2 Bottasso & Sembenelli, 2001; Chung, 2001; Forte & Sarmento, 2012). While some recent studies compute firm-level markups following De Loecker and Warzynski (2012), many such studies are cross-country in nature, raising questions of comparability across contexts (Bitzer & Görg, 2009; Rutkowski, 2006; Weche, 2018). Our paper builds upon this base by analyzing economic markups in a developing country setting. Second, we explore heterogeneity in the effect of foreign presence on individual firm markups depending on the firm’s ownership, specifically looking at potential differences between foreign-owned firms, domestic-owned private firms, and SOEs. The focus on SOEs in particular is critical to developing country contexts such as Vietnam, where SOEs play a major role in the economy (Dang, Nguyen, & Taghizadeh- Hesary, 2020). SOEs may distort competitive environments due to their complex ownership structures, weak management, and unclear financial and debt obligations. Thus, studying the potential (or lack thereof) for FDI-driven gains in SOE productivity and efficiency is of particular interest. In the next section, we summarize the related literature on FDI and market power. Then, we discuss the methodology in Section III before presenting the database and descriptive statistics in Section IV. Next, we report the results and robustness tests in Sections V and VI, respectively. Finally, we provide policy implications and discuss the future research agenda in Section VII. II. Literature review Conventional theory proposes two potentially conflicting outcomes of FDI in developing economies. One claims that the presence of foreign firms should increase the level of competition and, therefore, lower aggregate market power. Another suggests that foreign firms will have significant economies of scale and consequently undercut local firms, potentially raising market power for foreign firms (Amiti & Khandelwal, 2013; Khandelwal, Schott, & Wei, 2013; Nocke & Yeaple, 2008; Norbäck & Persson, 2007). There is extensive literature exploring the role of firm market power and the macroeconomic implications, but relatively few studies have explored this relationship in the context of FDI and the associated entry of multinational enterprises, especially in developing markets.2 The majority of country-level studies exploring the impact of FDI on market power and associated outcomes focus on advanced economies and come to conflicting conclusions. Barrios, Gorg, and Strobl (2005) examine FDI in Ireland and find that the concentration of foreign firms in a given market harms net entry, implying that FDI may have a negative impact on competition and, ultimately, consumers. Similarly, utilizing data from the Netherlands, Abolhassani and Danakol (2019) find that FDI increases concentration by raising the minimum efficient scale. In contrast, Forte and Sarmento (2012) and Bottasso and Sembenelli (2001) find that the presence of foreign firms reduces concentration when looking at Portugal and Italy, respectively. In the United States, Chung (2001) finds that FDI decreases markups at the industry level, with a 10 percent increase in foreign company sales leading to a 2.4 percent reduction in markups. A common theme among these studies is the authors’ emphasis that the results may be limited to the specific country observed and that additional research is needed to fully understand the impacts of FDI on market power. Cross-country studies by Rutkowski (2006), Weche (2018), and Bitzer and Gorg (2009) find some evidence that FDI increases competitiveness and reduces markups, but the results vary across countries, with some experiencing lower markups and others experiencing higher markups. Additionally, when there is 2 Studies of the impact of rising market power such as De Loecker, Eeckhout, & Unger (2020) have identified key implications such as declining labor market dynamism as well as falling labor and capital shares. 3 statistical significance, it is often weak and depends on model specifications. One possible explanation for the mixed results offered by Weche (2018) is that the positive spillovers and competitive impacts of FDI cancel each other out or vary in magnitude across countries. In one of the few studies focusing on a developing economy, Sivadasan (2009) explores the relaxation of FDI restrictions in India and finds a significant long-term negative impact on output prices, possibly indicating higher levels of competition from FDI. He also finds a positive impact on productivity and decomposes the productivity gains to show that they are likely the result of technological adoption and not shifts in market power toward more productive firms. Together, these findings indicate that FDI in developing economies may increase competition and reduce markups while still having positive technological spillovers. In another study of the Indian market, Stiebale and Vencappa (2018) analyze how acquisitions by domestic and international firms have differential impacts on target firms and product markets. They find that acquisitions by foreign firms lead to increases in quality and markups as well as a fall in quality-adjusted prices. This suggests that foreign acquisitions increase markups of acquired firms, but positive spillovers in quality are passed to consumers at the same time. The findings suggests that the quality effect dominates the markup effect but the authors do not rigorously test this conclusion in aggregate terms. Finally, Forte (2016) provides a comprehensive survey of the literature pertaining to FDI’s impact on concentration as well as its impact on domestic firm entry and exit. She finds mixed results regarding concentration with different studies finding both positive and negative impacts of FDI. When looking at domestic firm entry, she finds a positive relationship but observes mixed results around exit and survival. Her study outlines vast heterogeneity between countries and methodologies that have been observed in the literature and highlights the importance of further research, particularly in developing countries. The remainder of this paper attempts to address this gap by examining the impact of FDI on market power within the developing country context of Vietnam. III. Methodology This paper focuses on markups as our primary measure of market power. .3 For the purposes of this paper, we modify this expression slightly to P/MC (price divided by marginal cost) in line with De Loecker and Warzynski (2012) due to the ability to estimate this measure using firm-level data, but the interpretation remains the same. In conceptual terms, markups measure the ability of firms to exert their market power and price goods above the socially optimal outcome of prices equaling marginal costs under perfect competition. Thus, positive markups indicate market power and loss of social welfare. Nevertheless, in practice, firms may also charge prices above marginal costs in response to other incentives different than market power. For instance, firms may increase prices to cover high fixed costs, leading to higher markups in the market. Market power can also be assessed by concentration measures such as the Herfindahl-Hirschman Index (HHI) or firm entry and exit data, which are commonplace among legal practitioners and regulators in part 3 The standard definition of markups is commonly termed the Lerner index as it is built upon the work of Lerner (1934). In practice, there are certain reasons why the Lerner index may not be a completely perfect measure of market power, but in general it provides a closer measure than more indirect approaches of studying market power such as measures of concentration. For more information, refer to the discussion in Elzinga and Mills (2011). 4 because they are less data-intensive.4 We choose markups as our proxy to market power instead of concentration measures because markups acknowledge the role of both output and cost information in leading market power, while measurements such as HHI only use one dimension, particularly revenues. Additionally, we do not analyze the firm’s entry and exit barriers because we do not trust firms to enter and exit our data set in the years in which they actually enter and exit the market, based on an analysis of the correspondence between listed founding dates and years of entry into the data set. The first step of our methodology involves estimating markups, which are the dependent variable across all our regression specifications. We compute markups following the method proposed by De Loecker and Warzynski (2012).5 This method uses optimal input demand conditions from the standard cost minimization problem and includes a control function for unobserved input prices and a routine to recover the allocation of inputs. The empirical approach relies on the insight that the output elasticity of a variable production factor free of adjustment costs is only equal to its expenditure share in total revenue when price equals the marginal cost of production. Therefore, any divergence between an input’s revenue share and its output elasticity is a firm’s markup if competition is imperfect. It is important to mention that De Loecker and Warzynski’s (2012) methodology provides consistent estimates of the output elasticities while allowing some inputs to face adjustments costs. In practice, firm-level production functions are first estimated via a two-stage procedure. The first stage involves generating OLS estimates of output (defined as deflated turnover), assuming both Cobb-Douglas and translog gross output production functions. We define capital as deflated total assets, labor as number of employees, and materials as deflated cost of goods sold less labor cost. The residuals are then captured to proxy for unobserved productivity shocks. The residuals from the first stage are plugged into the second stage, which utilizes GMM to identify production function coefficients. Finally, in line with De Loecker and Warzynski (2012), markup estimates are derived by taking the ratio of output elasticity of materials to materials’ cost share of revenue. Once we finish estimating markups, our main empirical strategy involves estimating the following equation to analyze the impact of FDI on markups: ,, = 0 + 1 ,−1 + 2 ,−1 + 4 ,, + + + ,, (1) where ,, denotes markup; ,−1 represents foreign-owned firms’ share of output in the relevant two-digit sector; ,−1 is the export share of turnover in the relevant two-digit sector; ,, is a dummy variable reflecting whether a firm is foreign-owned; denotes firm fixed effects, denotes year fixed effects; i indexes firms; s indexes two-digit sectors as defined by the Vietnam Standard Industrial Classification System (VSIC); and t indexes years. Lagged explanatory variables are used in response to endogeneity concerns. Results from estimations of (1) are reported in Table 2. The presence of firm and time fixed effects allows us to control for time-invariant, firm-level unobservables and nonlinear, firm-invariant time trends, effectively making 1 a difference-in-differences 4 Wnder certain set of assumptions, concentration measures may suggest changes in the firms’ ability to charge higher markups (Cowling & Waterson, 1976). 5 Other methods of estimating markups have been used such as in Diewert and Fox (2008), where markups are calculated as revenues divided by total costs. The main drawback to this method is its requirement for data on capital costs which are rarely available. The method from De Loecker and Warzynski (2012) does have a drawback in its need to estimate production functions but is better suited to the available data in this study. 5 estimate of the impact of FDI presence on markups. In addition, ,−1 is included to control for the export intensity of a given sector. Sectors that primarily serve global export markets may have lower average markups as they must be competitive globally, although the relationship may be reversed at the individual firm level within sectors because firms’ productivity—and therefore markups—may increase through “learning by exporting” (Dauda, Nyman, & Cassim, 2019; De Loecker & Warzynski, 2012).6 Similarly, including ,, allows us to control for firm-level MNE status, which may impact markups through MNEs’ increased productivity or brand premia that MNEs are able to charge. In addition, in further estimations, we account for differential impacts based on export intensity of the sectors in which firms operate. Additional FDI within an export-intensive sector may have a smaller impact on firms’ markups in the host country because firms are already competing with producers around the world. In other words, they are likely already in competition with MNEs, and the significance of MNE entrants into the host country is likely to be smaller in global market share terms. Conversely, firms within a domestic market-oriented sector are more likely to be in direct competition with MNE entrants as the domestic market is served by firms operating in the host country and imports. To account for this possibility, we estimate the following equation: ,, = 0 + 0 ,−1 + 2 ,−1 + 3 ,−1 ∗ ,−1 + 3 ,, (2) + + + ,, To examine differences between the markups of foreign-owned and domestic-owned firms, we estimate the following equation: ,, = 0 + 1 ,, + ,,−1 + , + ,, (3) where ,,−1 denotes a vector of lagged firm-level controls (typically total factor productivity of revenue, logged output, and logged capital) and , denotes sector-time fixed effects. 1 can be interpreted as the difference in markups between foreign-owned and domestic firms after controlling for ,,−1 and sector- time trends. Results from estimations of (3) are reported in Table 4. However, estimations of (3) only provide a view of differences between foreign- and domestic-owned firms, but they do not necessarily analyze the impact of foreign ownership per se as they do not control for firm-level unobservables. Thus, we further analyze the impact of foreign control by estimating the following equation: ,, = 0 + 1 ,, + ,,−1 + + + ,, (4) where, once again, denotes firm fixed effects, and denotes year fixed effects. This difference-in- differences approach with firm-level fixed effects allows us to interpret 1 as the impact of foreign control as 1 reflects the impact on markups of firms changing ownership due to acquisitions by foreign firms (if the sample is restricted to firms that started domestic-owned in our sample period), divestments by foreign firms (if the sample is restricted to firms that started foreign-owned), or both (if the regression is run on the full sample). Results from estimations of (4) are reported in Table 5. 6 Data constraints prevent us from analyzing firm-level export status as the data for firm-level exports are incomplete in terms of coverage across years. 6 IV. Data This study uses a firm-level unbalanced panel dataset from the Vietnam Economic Survey over the 2009- 2016 period. The Vietnam Economic Survey is conducted annually by the Government of Vietnam’s General Statistics Office (GSO) and covers all manufacturing and services enterprises in the country with more than 20 employees and a representative sample of firms with fewer than 20 employees. Tax codes were used as a time-invariant identifier of firms. Several steps were taken to clean the data prior to analysis. First, Only manufacturing firms were considered for analysis given that the cost of goods sold (COGS) is more easily interpretable as materials costs within manufacturing than in services. In addition, firms with missing or invalid financial data (namely turnover, total assets, cost of goods sold, and employment) were excluded, as were firms with invalid or missing tax codes. Third, Micro-firms with fewer than 10 employees were excluded as production function estimation is not reliable at such low levels of employment. Finally, firms with outlier or unreasonable markup estimates (less than zero or over three) were removed from the data set.7 The resulting data set thus includes 117,363 observations with markup estimates across 35,659 firms over the 2010-2014 period and 2016. Information from 2015 was excluded due to incomplete cost of goods sold data, while markup estimates could not be made for 2009 (the earliest year of the data set) as markup estimates rely on lagged data (see discussion in the methodology section). In addition, the data from the Vietnam Economic Survey were supplemented with further national and global data for analytical purposes. Input and output deflator data at the Vietnam Standard Industrial Classification (VSIC) two-digit level were sourced from the GSO, while input-output tables for Vietnam were sourced from the Asian Development Bank (ADB). Finally, fixed capital deflators were obtained from the Food and Agriculture Organization (FAO). Descriptive statistics Table 1 displays summary statistics of various firm characteristics used for markup estimation and measures of markups and firm profitability. For firm i in year t, ,, is defined as deflated revenue; ,, denotes total employees at the end of year; ,, denotes deflated total assets at the beginning of year; ,, denotes deflated cost of goods sold (COGS); ,, is a dummy variable equal to one if the firm is foreign-owned; ,, is a dummy variable equal to one if the firm is state-owned; and ,, is markup defined as price divided by marginal cost. According to the legal information reported in the survey, a firm is considered foreign-owned when it is a joint venture between foreign owners and other private proprietors or is 100% foreign-owned. Firm-year observations vary widely in terms of their size whether measured in terms of revenue, employees, capital, or materials cost. Average estimated markups are greater than one, signifying that 7 Negative markups are implausible as they require prices or marginal costs to be negative. The high-end cutoff of three was selected as it implies very high gross margins in excess of 75 percent that are seldom seen in practice at this level of aggregation. While this cutoff is somewhat arbitrary, we show in Section VI that our results are robust to different levels of cutoffs. Overall, such outliers account for 2 percent of markup estimates assuming a Cobb- Douglas production function and under 1 percent of estimates assuming a translog production function. Outliers correspond to 1 percent or fewer of all translog estimates, although they are somewhat more common for within individual sectors for Cobb-Douglas estimates, with a maximum of 11 percent for the leather goods sector. 7 the average firm is able to price above marginal cost. About 20 percent of the firm-year observations in our sample are foreign-owned firms, while 2 percent are state-owned firms.8 Also, the sector-year statistics show that the average manufacturing sector accounts for 44% of its output in foreign-owned firms and has an export share of turnover of around 27%. However, there is significant dispersion with standard deviations around 25% and 17% for foreign-owned firms’ share of output and the export share of turnover, respectively. Table 1. Summary statistics Variable Obs. Mean Std. Dev. Min Max log (,, ) 120,029 5.15 1.86 -2.06 15.32 log (,, ) 120,029 4.05 1.35 2.30 11.35 Firm log( ,, ) 120,029 5.32 1.66 -2.20 14.28 characteristics log( ,, ) 120,029 4.65 2.08 -6.13 15.12 ,, 120,029 0.20 0.40 0.00 1.00 ,, 120,029 0.02 0.15 0.00 1.00 ,, (Cobb-Douglas) 117,363 1.11 0.42 0.05 3.00 Markups ,, (translog) 113,647 1.16 0.28 0.00 3.00 Sector , 168 0.44 0.25 0.01 0.99 characteristics , 144 0.27 0.17 0.00 0.72 Over time, both foreign presence and markups increased over our sample period across most two-digit manufacturing sectors within Vietnam. Figure 1 displays the evolution of foreign-owned firms’ share of industry output between 2010 and 2016 by two-digit manufacturing sector. The repair and maintenance, machinery manufacturing, and textiles sectors saw the largest increases in foreign presence over the 2010-2016 period, while the motor vehicles subsector saw the largest—albeit still modest—decrease. Figure 2 displays the evolution of mean markups by two-digit manufacturing sector between 2010 and 2016. Average markups increased across most two-digit sectors between 2010 and 2016, especially for motor vehicles and apparel. In contrast, the tobacco subsector saw a decrease in average markups. The increase in markups over time does not necessarily imply that firms have more market power. For instance, an increase in fixed costs or new product varieties may increase markups without necessarily implying higher profits. Unfortunately, due to data restrictions associated with the firm's cost structure, we cannot perform further analysis of profits. Therefore, we limit our conclusions to the rise in markups without emphasizing the potential market power rises. 8 In the data, we define SOEs as firms which have >50 percent state ownership based on their legal form. This corresponds with the legal forms for “Central State”, “Local State”, and “Joint Stock Co. having state capital >50%”. 8 Figure 1. Evolution of foreign presence across two-digit manufacturing sectors Figure 2. Evolution of markups across two-digit manufacturing sectors In general, before controlling for various firm- and sector-level factors, two-digit sectors which saw larger increases in foreign presence saw lower growth in markups over the 2010-2016 period (Figure 3). 9 Figure 3. Change in foreign presence versus change in markups by two-digit manufacturing sector Note: Points denote two-digit manufacturing sectors The relationship between sector-level mean markups and foreign share of revenue comes despite foreign firms generally charging higher markups than their domestic counterparts. Figure 4 displays the distribution of estimated markups for all valid firm-year observations over the sample period. Regardless of whether a Cobb-Douglas or translog production function is assumed, the distribution of estimated markups for foreign firms is to the right of the distribution for domestic firms. Figure 4. Distribution of estimated markups V. Results Impact of foreign presence on sector-level markups Our results show that the presence of foreign firms decreases markups at the sector level. Results from estimations of (1) and (2) are displayed in Table 2 in Columns 1-2 and 3-4, respectively, depending on the 10 production function assumed. Our preferred specifications in Columns 3-4 estimate (2) to account for heterogeneous impacts based on the export intensity of the sector for the theoretical reasons outlined previously. Under our preferred specifications, we find a negative and statistically significant coefficient on ,−1 , regardless of whether a Cobb-Douglas or translog production function is assumed. In other words, greater FDI intensity is associated with decreases in markups in a given sector. This result is largely in line with the findings of Chung (2001), Sivadasan (2009), Forte and Sarmento (2012) and Bottasso and Sembenelli (2001). Specifically, we find the point estimate implies that a 10-percentage point increase in FDI presence is associated with a 1.5 to 2.6 percentage point reduction in the ratio of price to marginal cost in sectors that sell entirely to domestic markets. Similarly, the reduction on markups due to an increase in FDI of the same magnitude in the median sector of the export distribution is between 0.04 to 1.2 percentage points. The coefficient on ,−1 is also negative across estimations of (1) that exclude the interaction term, although they are not always statistically significant. Table 2. Impact of FDI presence on markups (1) (2) (3) (4) Markup Markup Markup Markup VARIABLES (CD) (TL) (CD) (TL) ,−1 -0.0339 -0.111** -0.153* -0.261** (0.0446) (0.0459) (0.0796) (0.0955) ,−1 -0.116 -0.245* -0.323* -0.505*** (0.101) (0.127) (0.157) (0.155) ,−1 ∗ ,−1 0.371* 0.464 (0.199) (0.281) ,, 0.0860*** 0.125*** 0.0861*** 0.125*** (0.0273) (0.0263) (0.0273) (0.0261) Constant 1.144*** 1.260*** 1.203*** 1.335*** (0.0308) (0.0459) (0.0466) (0.0523) Firm fixed effects Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Observations 100,956 97,823 100,956 97,823 R-squared 0.810 0.824 0.810 0.824 Robust standard errors clustered at the industry level in parentheses. Dependent variable displayed at the top of each column: CD denotes markups derived assuming a Cobb-Douglas production function, while TL denotes markups derived assuming a translog production function. *** p<0.01, ** p<0.05, * p<0.1 Our results provide some evidence that the impact of foreign presence is lower in export-intensive sectors. The coefficients on the interaction term between FDI intensity and export intensity are generally positive in Table 2 and marginally statistically significant in Column 3, while the coefficient in Column 4 just misses the threshold for statistical significance. In addition, the coefficient on the export intensity term is negative across all specifications and at least marginally statistically significant in all but one specification. Taken together, these results suggest that FDI is associated with reduced markups primarily in domestic market-oriented sectors, potentially because firms in export-oriented sectors already face competition from producers around the world, making the contribution of MNE entrants to competition more negligible. 11 In addition, the impact of foreign presence on markups may vary depending on incumbent firms’ ownership structures. To study differential impacts of FDI presence depending on the ownership structure of incumbent firms, we estimate a modified version of (2) with interaction terms with dummies for foreign ownership (,, ) and state ownership (,, ). The results from the resulting regressions are summarized in Table 3. The coefficient on the interaction of FDI presence and MNE status tends to be statistically insignificant. In contrast, interactions between a dummy for SOE and foreign presence tend to have positive and at least marginally statistically significant coefficients. This finding suggests that FDI may introduce competition for both foreign- and domestic-owned private firms but that SOEs may be insulated from these competitive pressures through other mechanisms, e.g., preferential financing, subsidies, or regulatory advantages. Table 3. Heterogeneous impact of FDI presence on markups depending on ownership structure (1) (2) (3) VARIABLES Markup (TL) Markup (TL) Markup (TL) ,−1 -0.275** -0.268** -0.283** (0.103) (0.0960) (0.103) ,−1 -0.538*** -0.509*** -0.544*** (0.158) (0.156) (0.159) ,−1 ∗ ,−1 0.574* 0.476 0.590* (0.302) (0.281) (0.300) ,, 0.167*** 0.126*** 0.164*** (0.0575) (0.0261) (0.0572) ,, ∗ ,−1 -0.0345 -0.0263 (0.119) (0.119) ,, ∗ ,−1 -0.0143 -0.00719 (0.147) (0.145) ,, ∗ ,−1 * ,−1 -0.120 -0.136 (0.291) (0.287) ,, -0.0672* -0.0696** (0.0345) (0.0323) ,, ∗ ,−1 0.140* 0.145* (0.0753) (0.0715) ,, ∗ ,−1 0.174 0.194 (0.133) (0.119) ,, ∗ ,−1 * ,−1 -0.233 -0.272 (0.254) (0.226) Constant 1.336*** 1.337*** 1.338*** (0.0543) (0.0528) (0.0549) Firm fixed effects Yes Yes Yes Year fixed effects Yes Yes Yes Observations 97,823 97,823 97,823 R-squared 0.824 0.824 0.824 Robust standard errors clustered at the industry level in parentheses. Dependent variable displayed at the top of each column: CD denotes markups derived assuming a Cobb-Douglas production function, while TL denotes markups derived assuming a translog production function. *** p<0.01, ** p<0.05, * p<0.1 12 Markups of foreign firms vis-à-vis domestic firms Despite foreign-owned firms increasing competition and decreasing markups at the sector level, we also find evidence that individual foreign-owned firms generally possess higher ability to price above marginal cost relative to their domestic-owned counterparts. Table 4 displays estimates of (3). Across specifications, the coefficient on ,, is consistently positive and statistically significant, reflecting how foreign-owned firms are able to charge higher markups than their domestic-owned counterparts. This finding is robust to controls for (lagged) firm productivity, suggesting that the higher markups are driven by higher market power and not just improved efficiency. In addition, our results indicate that foreign-owned firms are less able to charge markup premiums in sectors already characterized by high competition from global trade or other foreign firms. In Table 4, the coefficient on the interaction term ,, * ,−1 is negative across specifications and statistically significant in our preferred specification with markup estimates assuming a translog production function. A similar pattern exists for ,, * ,−1 . This finding suggests that foreign-owned firms are less able to charge markup premiums in the face of increased competition from other firms or with respect to global markets. Table 4. Comparison of markups of foreign-owned vs. domestic-owned firms (1) (2) (3) (4) (5) (6) Markup Markup Markup Markup Markup Markup VARIABLES (CD) (TL) (CD) (TL) (CD) (TL) ,, 0.193*** 0.0954*** 0.227*** 0.227*** 0.230*** 0.202*** (0.0287) (0.0192) (0.0529) (0.0297) (0.0513) (0.0425) ,, * ,−1 -0.106 -0.404*** (0.0981) (0.0989) ,, * ,−1 -0.0753 -0.212** (0.0705) (0.0870) log (),,−1 -0.166*** -0.0297*** -0.166*** -0.0293*** -0.166*** -0.0294*** (0.00750) (0.00544) (0.00756) (0.00526) (0.00755) (0.00546) log (),,−1 0.0727*** 0.00954* 0.0723*** 0.00889* 0.0725*** 0.00906* (0.00631) (0.00470) (0.00665) (0.00474) (0.00633) (0.00454) (),,−1 -0.239 -0.259 -0.247 (0.275) (0.286) (0.274) (),,−1 0.133*** 0.115*** 0.118*** (0.0359) (0.0348) (0.0343) Constant 1.619*** 1.235*** 1.627*** 1.238*** 1.622*** 1.238*** (0.106) (0.0406) (0.111) (0.0387) (0.106) (0.0395) Firm fixed effects No No No No No No Industry-year fixed effects Yes Yes Yes Yes Yes Yes Observations 78,596 75,673 78,596 75,673 78,596 75,673 R-squared 0.280 0.415 0.280 0.423 0.280 0.418 Robust standard errors clustered at the industry level in parentheses. Dependent variable displayed at the top of each column: CD denotes markups derived assuming a Cobb-Douglas production function, while TL denotes markups derived assuming a translog production function. *** p<0.01, ** p<0.05, * p<0.1 13 Our results also provide evidence that foreign control per se is associated with an increase in markups via our estimations of (4). In Table 5—across specifications that analyze the impact of acquisitions of domestic companies by foreign firms (Columns 4-6), divestments of foreign-owned firms to domestic buyers (Columns 7-9), and both (Columns 1-3)—the coefficient on ,, is consistently positive and statistically significant. With firm fixed effects, these results can be interpreted as evidence that switching from domestic to foreign control is associated with an increase in markups, while switching from foreign to domestic control is associated with a decrease in markups. The coefficients on interactions with export intensity and foreign presence are also consistently negative—suggesting that foreign presence and export intensity may decrease foreign-owned firms’ markup premiums—although only one of these coefficients is statistically significant (Column 9, which examines the impact of divestments by foreign shareholders). Table 5. Impact of foreign control on markups (1) (2) (3) (4) (5) (6) (7) (8) (9) Markup Markup Markup Markup Markup Markup Markup Markup Markup VARIABLES (TL) (TL) (TL) (TL) (TL) (TL) (TL) (TL) (TL) ,, 0.116*** 0.157*** 0.140*** 0.0965* 0.302 0.132 0.105*** 0.145*** 0.195*** (0.0285) (0.0362) (0.0276) (0.0533) (0.213) (0.0954) (0.0259) (0.0507) (0.0480) ,, * ,−1 -0.115 -0.541 -0.106 (0.0746) (0.485) (0.115) ,, * ,−1 -0.0544 -0.0823 -0.212** (0.0434) (0.223) (0.0930) log (),,−1 0.00267 0.00281 0.00274 0.00479* 0.00479* 0.00479* -0.00759 -0.00759 -0.00772 (0.00261) (0.00258) (0.00260) (0.00263) (0.00262) (0.00263) (0.0148) (0.0148) (0.0147) log (),,−1 -0.00380 -0.00403 -0.00387 -0.00296 -0.00300 -0.00297 -0.0113** -0.0113** -0.0113** (0.00267) (0.00267) (0.00267) (0.00292) (0.00292) (0.00292) (0.00510) (0.00509) (0.00506) (),,−1 0.0162 0.0158 0.0159 -0.00450 -0.00497 -0.00460 0.0450*** 0.0450*** 0.0448*** (0.0213) (0.0210) (0.0211) (0.0151) (0.0148) (0.0151) (0.00834) (0.00836) (0.00835) Constant 1.134*** 1.134*** 1.135*** 1.136*** 1.137*** 1.137*** 1.212*** 1.206*** 1.235*** (0.0176) (0.0169) (0.0176) (0.0124) (0.0124) (0.0125) (0.100) (0.0955) (0.0904) Sample All firms All firms All firms Started Started Started Started Started Started domestic domestic domestic foreign foreign foreign Firm fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Industry-year fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Observations 69,621 69,621 69,621 53,453 53,453 53,453 16,060 16,060 16,060 R-squared 0.833 0.833 0.833 0.807 0.807 0.807 0.900 0.901 0.901 Robust standard errors clustered at the industry level in parentheses. Dependent variable displayed at the top of each column: TL denotes markups derived assuming a translog production function. *** p<0.01, ** p<0.05, * p<0.1 VI. Robustness tests Exclusion and inclusion of potential outliers Outlier sectors and years may potentially drive the results for both the impact of foreign presence and foreign control on markups. To account for this possibility, we rerun our preferred specifications from Table 2 (Column 4) and Table 5 (Column 1), excluding two-digit sectors and years one at a time. Our findings are robust to dropping individual sectors and years, with regression coefficients preserving their 14 signs, statistical significance, and approximate magnitude regardless of which individual sector or year is dropped.9 In addition, it is possible that the inclusion or exclusion of extreme values may be driving our observed results. To account for this possibility, we reintroduce potentially valid markup observations that had been dropped (i.e., those markups greater than three, although markups less than zero remain excluded as they are not plausibly valid) and experiment with different levels of winsorization and then rerun our preferred specifications. Again, the regression coefficients largely preserve their signs, statistical significance, and approximate magnitudes across specifications, as seen in Table 6 and Table 7. Table 6. Impact of FDI presence on markups (different levels of winsorization) (1) (2) (3) (4) (5) (6) Markup Markup Markup Markup Markup Markup VARIABLES (CD) (TL) (CD) (TL) (CD) (TL) ,−1 -0.127 -0.257** -0.145 -0.261*** -0.156* -0.267*** (0.120) (0.0992) (0.100) (0.0904) (0.0771) (0.0797) ,−1 -0.192 -0.449*** -0.256 -0.488*** -0.296* -0.454*** (0.261) (0.156) (0.195) (0.147) (0.150) (0.121) ,−1 ∗ ,−1 0.239 0.419 0.299 0.470* 0.343* 0.546** (0.264) (0.274) (0.217) (0.255) (0.185) (0.206) ,, 0.155*** 0.150*** 0.126*** 0.129*** 0.0910*** 0.111*** (0.0543) (0.0493) (0.0355) (0.0294) (0.0221) (0.0241) Constant 1.254*** 1.324*** 1.244*** 1.331*** 1.219*** 1.311*** (0.0793) (0.0521) (0.0630) (0.0498) (0.0467) (0.0424) Level of winsorization None None 1% 1% 5% 5% Firm fixed effects Yes Yes Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Yes Yes Observations 103,283 98,051 103,283 98,051 103,283 98,051 R-squared 0.654 0.793 0.799 0.822 0.826 0.834 Robust standard errors clustered at the industry level in parentheses. Dependent variable displayed at the top of each column: CD denotes markups derived assuming a Cobb-Douglas production function, while TL denotes markups derived assuming a translog production function. *** p<0.01, ** p<0.05, * p<0.1 Table 7. Impact of foreign control on markups (different levels of winsorization) (1) (2) (3) (4) (5) (6) Markup Markup Markup Markup Markup Markup VARIABLES (CD) (TL) (CD) (TL) (CD) (TL) ,, 0.107** 0.140*** 0.0955** 0.123*** 0.0757** 0.104*** (0.0493) (0.0456) (0.0436) (0.0314) (0.0313) (0.0269) log (),,−1 -0.0209 0.00286 -0.0154* 0.00302 -0.0144* 0.00265 (0.0124) (0.00336) (0.00827) (0.00285) (0.00756) (0.00236) 9 Regression results from these specifications are not reported in this paper but are available from the paper’s authors upon request. 15 log (),,−1 -0.0354* -0.00405 -0.0191*** -0.00397 -0.0149*** -0.00266 (0.0171) (0.00296) (0.00514) (0.00274) (0.00381) (0.00220) (),,−1 -0.0423** 0.00658 -0.0220 -0.00673 -0.0105 -0.00369 (0.0191) (0.0268) (0.0167) (0.0164) (0.0165) (0.0126) Constant 1.447*** 1.134*** 1.306*** 1.137*** 1.253*** 1.130*** (0.153) (0.0234) (0.0466) (0.0181) (0.0385) (0.0170) Level of winsorization None None 1% 1% 5% 5% Firm fixed effects Yes Yes Yes Yes Yes Yes Industry-year fixed effects Yes Yes Yes Yes Yes Yes Observations 73,768 69,770 73,768 69,770 73,768 69,770 R-squared 0.599 0.809 0.783 0.830 0.819 0.840 Robust standard errors clustered at the industry level in parentheses. Dependent variable displayed at the top of each column: CD denotes markups derived assuming a Cobb-Douglas production function, while TL denotes markups derived assuming a translog production function. *** p<0.01, ** p<0.05, * p<0.1 Propensity score matching and the impact of foreign acquisition The non-random selection of acquisition targets by foreign acquirors is a potential source of endogeneity that may bias estimates under our baseline specification: Foreign acquirors may disproportionately target firms with greater market power or, more precisely, firms on more upward trajectories with respect to market power. To control for the non-random selection of acquired firms, we apply propensity score matching and reweighting to construct a counterfactual by matching acquired firms with similar firms that were not acquired. By combining propensity score reweighting with the difference-in-differences estimator, we estimate the average treatment effect on the treated (ATT), which compares the actual post-acquisition markup of a target firm and the situation had the firm not been acquired (Stiebale & Vencappa, 2018). A probit regression is used to calculate propensity scores for the likelihood of acquisition by a foreign acquirer controlling for industry and year fixed effects. According to the coefficients from the probit estimation reported in Table 8, output is a significant predictor of foreign acquisition regardless of specification. Total factor productivity is also associated with a higher propensity to be acquired when the specification accounts for all available information before the acquisition (Columns 3-4). 16 Table 8. Propensity score estimation (1) (2) (3) (4) For. Acq = For. Acq = For. Acq = For. Acq = VARIABLES 1 1 1 1 log (),,−1 0.170*** 0.165*** 0.172*** 0.163*** (0.0424) (0.0423) (0.0254) (0.0253) log (),,−1 -0.023 -0.026 0.032 0.028 (0.0447) (0.0442) (0.0270) (0.0267) (),,−1 0.398 0.753*** (0.4634) (0.2517) (),,−1 0.197 0.402*** (0.2017) (0.1220) Constant -4.342*** -4.180*** -4.278*** -3.986*** (0.3035) (0.2171) (0.1688) (0.1226) Industry fixed effects Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Observations 69,891 69,891 70,840 70,840 Standard errors in parentheses. Lagged covariates correspond to the value one year before the acquisition in Columns (1) and (2) and all observations before the acquisition in Columns (3) and (4). The dependent variable is an indicator variable equals 1 if the domestic establishment was acquired by a foreign firm, 0, otherwise. TFPR (CD) denotes total factor productivity assuming a Cobb-Douglas production function, and TFPR (TL) represents total factor productivity assuming a translog production function. *** p<0.01, ** p<0.05, * p<0.1 After estimating propensity scores, we calculate weights to approximately preserve proportions between treatment and control groups and test the balancing property between treated and control groups. Table 9 summarizes these results by showing the difference in the means of key variables between treatment and control groups. The evidence suggests no statistically significant differences in output, capital, and TFP between treatment and control groups, regardless of whether we assume a Cobb-Douglas or translog production function. Table 9. Balance between treatment and control groups (1) (2) (3) (4) (5) Mean t-test VARIABLES Treated Control Diff t Pr(|T| > |t|) N = 70,840 log (),,−1 0.0207 0.0238 -0.0032 1.05 0.295 log (),,−1 0.0203 0.0234 -0.0031 1.06 0.287 (),,−1 0.0009 0.001 -0.0001 0.91 0.361 (),,−1 0.0000 0.0000 -0.0000 0.14 0.886 ,,−1 0.0037 0.0045 -0.0008 1.73 0.360 Table shows the mean values of covariates for the reweighted sample. TFPR (CD) denotes total factor productivity assuming a Cobb-Douglas production function, and TFPR (TL) represents total factor productivity assuming a translog production function. 17 Finally, using the matched sample, the difference-in-differences estimation shows that target firms’ markups increase by about 7 percent after acquisition by a foreign acquiror (Columns 3 and 4). Table 10 shows that coefficients associated with the dummy variable for a firm that has been acquired by a foreign investor are positive and statistically significant, regardless of whether a Cobb-Douglas or translog production function is assumed. For consistency with the baseline results, we show estimates at levels, too (Columns 1 and 2). Table 10. Impact of foreign acquisition on markups (propensity score matching estimation) (1) (2) (3) (4) Markup (CD) ln(Markup) Markup (TL) ln(Markup) (TL) VARIABLES (CD) ,, 0.089*** 0.089*** 0.071*** 0.071*** (0.0244) (0.0244) (0.0178) (0.04) Constant 1.127*** 1.125*** 0.094*** 0.097*** (0.0007) (0.008) (0.003) (0.22) Firm fixed effects Yes Yes Yes Yes Industry-year fixed effects Yes Yes Yes Yes Observations 51,011 51,011 51,011 51,011 R-squared 0.8632 0.8562 0.902 0.889 Robust standard errors in parentheses. , is an indicator variable equal to 1 if the domestic establishment was acquired by a foreign firm and 0 otherwise. CD and TL denote markups estimated assuming Cobb-Douglas and translog production functions, respectively. The table shows the ATT based on reweighted regressions at the firm-level. *** p<0.01, ** p<0.05, * p<0.1 VII. Policy implications and agenda for future research Overall, evidence from Vietnam suggests that FDI and, by extension, pro-FDI reforms are pro-consumer. Within Vietnam, greater foreign firm presence is associated with lower markups. Thus, further opening sectors to FDI is likely to be beneficial for competition and domestic consumers. Such reforms could be especially important in the current context of industries undergoing consolidation due to pandemic-driven bankruptcies. At the same time, FDI does not decrease markups as much in export-intensive sectors, one area where higher markups are more of a ‘good’ thing for the host country. These findings hold at the sector level despite individual foreign-owned firms charging higher markups than domestic-owned firms at the firm level, perhaps owing to foreign-owned firms’ ‘premium-ness’ (e.g., via brand names or reputation). However, FDI may need to be accompanied by complementary reforms—particularly with respect to SOEs—for countries to reap the full benefits of increased competition from FDI. We find suggestive evidence that FDI intensity does not decrease the markups of SOEs as much as those of other firms, suggesting that SOEs are insulated from the competitive pressures introduced by FDI. This finding may be driven by SOEs receiving preferential financing, subsidies or other policy-related advantages. Thus, insofar as SOEs operate in areas important to domestic consumers, host-country governments may wish to package FDI reforms with efforts to level the playing field between SOEs and private firms or even privatization. Such SOE reforms could include ensuring neutrality between SOEs and private enterprises with respect to product market regulations, public procurement processes, and taxation (World Bank, 18 2019) and are likely to be especially important in contexts like Vietnam’s, where SOEs continue to play a significant role in the local economy (Dang, Nguyen, & Taghizadeh-Hesary, 2020). Future research could provide further evidence on the impact of FDI on competition and market power in developing countries as well as analyze heterogeneity. As mentioned earlier, most of the evidence on the relationship between FDI, competition, markups, and market power come from studies of high-income countries. Expanding to further countries beyond Vietnam would thus help to deepen the evidence base for developing countries. In addition, studies across different country settings sometimes draw conflicting conclusions regarding the impact of FDI on markups and market power. Analyzing why the impact of FDI may differ across contexts (e.g., based on policy environment or industry structure) would enrich policymakers’ and researchers’ understanding of the trade-offs of FDI and how to maximize the benefits of FDI. Further research on whether there are differential impacts based on mode of FDI entry (e.g., mergers and acquisitions versus greenfield) would also allow for more tailored policies for maximizing the development impact of FDI. Finally, if additional data were available, analysis of FDI’s impact on competition could be expanded to cover services, especially with respect to digital platforms. 19 References Abolhassani, M., & Danakol, S. H. (2019). Wage and competition channels of foreign direct investment and new firm entry. Small Business Economics, 53(4), 935-960. Akcigit, U., Chen, W., Diez, F. J., Duval, R. A., Engler, P., Fan, J., . . . Villegas-Sanchez, C. (2021). Rising Corporate Market Power: Emerging Policy Issues. Washington: IMF. Amiti, M., & Khandelwal, A. K. (2013). Import Competition and Quality Upgrading. The Review of Economics and Statistics, 95(2), 476-490. Barrios, S., Gorg, H., & Strobl, E. (2005). Foreign direct investment, competition and industrial development in the host country. European Economic Review, 49(7), 1761-1784. Bertrand, O. H.-J. (2012). Should countries block foreign takeovers of R&D champions and promote greenfield entry? Canadian Journal of Economics, 45(3), 1083-1124. doi:https://doi.org/10.1111/j.1540-5982.2012.01728.x Bitzer, J., & Görg, H. (2009). Foreign Direct Investment, Competition and Industry Performance. The World Economy, 32(2), 221-233. Bottasso, A., & Sembenelli, A. (2001). Market power, productivity and the EU Single Market Program: Evidence from a panel of Italian firms. European Economic Review, 45(1), 167-186. Chung, W. (2001). Mode, size, and location of foreign direct investments and industry markups. Journal of Economic Behavior and Organization, 45, 185-211. Cowling, K., & Waterson, M. (1976). Price–cost margins and industry structure. The Economic Journal, 43, 267-274. Dang, L. N., Nguyen, D. D., & Taghizadeh-Hesary, F. (2020). State-Owned Enterprise Reform in Viet Nam: Progress and Challenges. ADBI Working Paper 1071. Dauda, S., Nyman, S., & Cassim, A. (2019). Product Market Competition, Productivity, and Jobs: The Case of South Africa. World Bank Policy Research Working Paper 9084. De Loecker, J., & Eeckhout, J. (2020). Global Market Power. Working Paper. De Loecker, J., & Warzynski, F. (2012). Markups and Firm-Level Export Status. American Economic Review, 102(6), 2437-2471. De Loecker, J., Eeckhout, J., & Unger, G. (2020). The Rise of Market Power and the Macroeconomic Implications. Quarterly Journal of Economics, 135(2), 561-644. Elzinga, K. G., & Mills, D. E. (2011). The Lerner Index of Monopoly Power: Origins and Uses. American Economic Review, 101, 558-564. Forte, R. P. (2016). Multinational firms and host country market structure: A review of empirical literature. The Journal of International Trade and Economic Development, 25(2), 240-265. Forte, R., & Sarmento, P. (2012). Foreign Presence and Market Concentration: The Case of Portuguese Manufacturing Industries. FEP Working Papers, 444. 20 Hall, R. (1988). The Relation between Price and Marginal Cost in U.S. Industry. Journal of Political Economy, 96(5), 921-947. Hall, R. (2018). New Evidence on the Markup of Prices over Marginal Costs and the Role of Mega-Firms in the US Economy. NBER Working Paper. Khandelwal, A. K., Schott, P. K., & Wei, S.-J. (2013). Trade Liberalization and Embedded Institutional Reform: Evidence from Chinese Exporters. American Economic Review, 103(6), 2169-2195. Lerner, A. (1934). The Concept of Monopoly and the Measurement of Monopoly. Review of Economic Studies, 1, 157-175. Nocke, V., & Yeaple, S. (2008). An Assignment Theory of Foreign Direct Investment. Review of Economic Studies, 75(2), 529-557. Norbäck, P.-J., & Persson, L. (2007). Investment liberalization -- Why a restrictive cross-border merger policy can be counterproductive. Journal of International Economics, 72(2), 366-380. Rutkowski, A. (2006). Inward FDI, concentration, and profitability in the CEECs: Were the domestic firms crowded out or strengthened. Transnational Corporations, 15(3), 107-141. Sivadasan, J. (2009). Barriers to Competition and Productivity: Evidence from India. The B.E. Journal of Economic Analysis & Policy, 9(1), 1-66. Stiebale, J., & Vencappa, D. (2018). Acquisitions, markups, efficiency, and product quality: Evidence from India. Journal of International Economics, 112(C), 70-87. Weche, J. P. (2018). The heterogeneous competitive effects of trade and foreign direct investment: Firm-level evidence for European countries. The World Economy, 41, 801-830. World Bank. (2019). State-owned enterprises: Understanding their market effects and the need for competitive neutrality. Washington: World Bank. 21