key: cord-0760698-9cwsbjaw
authors: Boubakri, Narjess; Chen, Ruiyuan Ryan; El Ghoul, Sadok; Guedhami, Omrane; Nash, Robert
title: State ownership and stock liquidity: Evidence from privatization
date: 2020-10-22
journal: nan
DOI: 10.1016/j.jcorpfin.2020.101763
sha: ac0e5f5e5ed3802c6ff4b8b8be8efccf9424e575
doc_id: 760698
cord_uid: 9cwsbjaw

We provide unique firm-level evidence of the relation between state ownership and stock liquidity. Using a broad sample of newly privatized firms (NPFs) from 53 countries over the period 1994–2014, our study identifies a non-monotonic association between state ownership and stock liquidity. The inverse U-shaped relation is consistent with trade-offs between costs and benefits of state ownership and suggests an optimal level of government shareholdings that maximizes stock liquidity of NPFs. We further identify that the inflection point from the cost/benefit trade-off is contingent upon characteristics of the nation's institutional environment.

Government bailout programs during the global financial crisis (GFC) led to a significant increase in state ownership around the world, giving rise to what is now called State Capitalism.

This phenomenon was perceived as an overturn of decades of privatizations (i.e., divestitures of government assets) that sought to disengage the economy from state dominance. Governments' equity ownership driven by -reverse privatizations‖ accounted for nearly one-fifth of stock market capitalization worldwide (Borisova et al., 2015; Megginson, 2017) 1 renewing the debate about the role of governments as shareholders. In this vein, recent research examines how and to what extent state ownership affects the valuation of corporate assets and equity (e.g., Holland, 2019) , the cost of equity (Ben-Nasr, Boubakri, and Cosset, 2012) , the cost of debt (Borisova and Megginson, 2011; Borisova et al., 2015) , cash holdings (Chen et al., 2018) , corporate risk-taking (Boubakri, Cosset, and Saffar, 2013) , governance quality (Borisova et al., 2012) , and corporate investment efficiency (Jaslowitzer, Megginson, and Rapp, 2018) . 2 The debate is now being fueled by the bailout programs that governments worldwide have devised to rescue major industries hit by the COVID-19 pandemic (Megginson and Fotak, 2020) . Our paper adds to this literature by investigating how state ownership affects the liquidity of the stock of formerly stateowned firms or newly privatized firms (NPFs). 3 1 Referring to state capitalism, Megginson (2017, p.18) notes, -This policy emphasizes that governments can and should retain control over vital national economic assets and promote the development of national champions in various globally competitive industries, and invest some or most of the national savings through state-owned vehicles‖. Nash (2017) investigates the balance between public and private ownership for 1996-2015 and finds that although privatizations outpaced nationalizations in the late 1990s, the gap narrowed after 2000, especially during the GFC (when the number of nationalizations surpassed the number of privatizations). 2 See Megginson (2017) for a recent survey of the literature examining various aspects of state ownership and the economic role of state-owned enterprises. 3 Following Chen et al. (2018) , we define an NPF as an entity in which state ownership has been recently reduced through privatization. An NPF will thus have a zero or positive level of residual state ownership.

shareholder expropriation, and improve firm performance. 5 To the extent that investors also associate government ownership with poor corporate governance (e.g., Borisova et al., 2012) , they shy away from buying. Consistent with this view, Chung, Elder, and Kim (2010) show that better corporate governance is associated with higher market liquidity, and Kyle (1985) finds that the price impact from trading is greater for firms with poor corporate governance. These arguments thus suggest that higher residual ownership by the state should lead to lower firm liquidity.

On the other hand, the soft-budget-constraint view of state ownership holds that the state, compared to other blockholders, can relax an SOE's budget constraint through implicit government guarantees, preferential access to credit, tax discounts and other forms of support (Kornai et al., 2003; Faccio, Masulis, and McConnell, 2006; Borisova and Megginson, 2011; Borisova et al., 2015; Nash, 2017; Boubakri, El Ghoul, Guedhami, and Megginson, 2018; Boubakri and Saffar, 2019; Holland, 2019) . Consistent with this perspective, Faccio et al. (2006) document that politically connected firms are more likely to be bailed out than non-connected matching firms. Focusing on state ownership, Beuselinck et al. (2017) and Boubakri et al. (2018) find that that SOEs exhibit higher market valuation than non-SOEs, consistent with the benefits of state ownership under the soft budget constraint view. These studies of the financing advantages of state-owned firms argue that the ownership of SOE stocks may thus be more desirable for investors. This effect is especially important in periods of economic distress. These studies contend that higher levels of state ownership increase shareholders' incentives to invest in NPFs, which should contribute to a positive relation between state ownership and firm-level stock liquidity. 5 An abundant literature provides empirical evidence on performance improvements following privatization. For a review of this literature, see Megginson (2017) .

To identify how these two countervailing influences can affect firm-level liquidity, we use a sample of 3,759 firm-year observations representing 473 NPFs from 53 countries over the period 1994-2014. We examine firm-level liquidity within the context of privatization because state divestiture is typically implemented over time, which allows us to consider how variations in levels of government ownership may affect a stock's liquidity. We find that the stock of partially privatized firms exhibits higher liquidity than that of fully privatized firms, consistent with the soft-budget-constraint view of state ownership.

Our results also suggest an inverse U-shaped relation between state ownership and our proxies for stock liquidity. In particular, using a quadratic regression that controls for firm-and country-level factors that could impact stock liquidity, we determine that liquidity is highest at an inflection point of 44% government ownership. Our finding of a liquidity-maximizing mix of public and private ownership is predicted by Shleifer and Vishny (1994) who suggest that privatization decisions are driven by a trade-off between the costs and benefits of state ownership.

In the context of our study, when the government stake in the firm is less than 44%, the benefits from financing advantages and implicit guarantees associated with state ownership (the softbudget-constraint view) appear to outweigh the costs of state ownership due to the pursuit of non-economic objectives (the political view). However, investor fear of the -grabbing hands‖ should increase with the level of state ownership. When the government retains more than 44% of firm shares, the costs of state ownership apparently become higher than the benefits, leading to less demand for NPFs' stocks and hence less trading and reduced liquidity.

Providing additional support for our conclusion, we find that the inflection point shifts in a manner consistent with cross-country differences in the relative costs and/or benefits of state ownership. Focusing first on the costs of state ownership (political view), we find that the J o u r n a l P r e -p r o o f inflection point falls from 44% to 21% in countries with left-wing governments, where the -grabbing hand‖ of the state is more likely (Shleifer and Vishny, 1994; Biais and Perotti, 2002; Megginson, Nash, Netter, and Poulsen, 2004; D'Souza and Nash, 2017; Chen et al., 2020) , and moves up to 55% for countries with center/right-wing governments, where the -grabbing hand‖ of the government is less likely. Also, to assess the effect of relative differences in the benefits of state ownership, we follow Frydman et al. (2000) and Chen et al. (2020) and note that higher levels of state ownership of banks should enhance the financing advantages provided to SOEs (soft-budget-constraint view) . Consistent with the soft-budget-constraint view, we find that the inflection point rises to 49% in countries with greater prevalence of state-owned banks and falls to 40% if state involvement in the banking sector is lower.

Also, we consider the role of the GFC and determine that the effect of state ownership on stock liquidity is stronger during the crisis period. Reflecting one of the major benefits of state ownership (as articulated in the soft-budget-constraint view), this finding suggests that liquidity is enhanced by a strong government presence/influence in the economy and by the resultant greater likelihood of government-led bailout programs during times of financial distress (Faccio et al., 2006; Borisova and Megginson, 2011; Boubakri et al., 2012; Boubakri et al., 2018) .

Our evidence is robust to employing instrumental variables, propensity score matching, and Heckman selection models to address potential endogeneity, and to including additional controls for political risk, income inequality, tradable goods, institutional and foreign ownership, acquisition activities, and changes in trading rules and securities regulation. Furthermore, our findings are not sensitive to excluding China from our sample, employing panel and Tobit regressions, and using alternative measures of stock liquidity. Finally, based on a sample of re-J o u r n a l P r e -p r o o f nationalized firms, we find that re-nationalization decreases stock liquidity, which is consistent with the political view of state ownership.

Our paper makes several contributions to the literature. First, we broaden our understanding of the effects of privatization on financial market development [see Megginson (2005 [see Megginson ( , 2017 for excellent surveys]. Evidence regarding the liquidity implications of privatization has been scarce, with the notable exceptions of Boutchkova and Megginson (2000) and Bortolotti et al. (2007) , who focus on the link between privatization and aggregate liquidity.

Providing an alternative perspective, our paper complements these country-level studies (Boutchkova and Megginson, 2000; Bortolotti et al., 2007) by using firm-level liquidity metrics to identify how state ownership affects firm-level stock liquidity.

Our analysis also enhances our understanding of the association between liquidity and ownership structure (i.e., ownership concentration, family ownership, institutional investors, and foreign investors) and investor base (e.g., Attig et al., 2006; Brockman et al., 2009; Chia et al., 2020) . More closely related to our study, Brockman et al. (2009) argue that block ownership can affect the liquidity of a firm's stock by changing the firm's trading activity or by changing its information environment. We extend Brockman et al. (2009) to examine the impact of a particular (and very different) type of block ownership-state ownership-on stock liquidity.

We focus on the government as a blockholder because, compared to other large shareholders, the state is unique. State blockholders are characterized by multiple objectives (including the pursuit of political and social goals) and are typically plagued by weak monitoring incentives and less effective corporate governance. However, state owners also benefit from the financing advantages stemming from the soft budget constraint. We show that these countervailing influences contribute to a non-monotonic relation between state ownership and J o u r n a l P r e -p r o o f stock liquidity. Furthermore, our study extends Ding and Suardi (2019) , who find that state ownership is associated with higher stock liquidity, in the context of China.

Finally, our results have policy implications for the privatization process. Our evidence indicates that high levels of continued government ownership in NPFs is suboptimal.

Specifically, if substantial amounts of state ownership remain, privatization does little to enhance liquidity at the firm level. This finding suggests that, as Megginson (2017) argues, large ties to the government should be reduced for the most favorable outcomes from privatization to materialize. The economic distortions introduced by state ownership can thus be costly to the economy when governments dominate as residual owners in NPFs.

The remainder of the paper is organized as follows. In Section 2, we review the literature and derive our hypotheses. Section 3 describes our sample and reports descriptive statistics.

Section 4 presents our results. We conclude in Section 5.

2.1. Ownership structure, corporate governance, and stock liquidity Prior studies contend that block ownership affects liquidity through one of two main channels: trading activity and information environment (Bhide, 1993; Bolton and Von Thadden, Overall, the political view of government ownership suggests that residual government stakes in NPFs discourage other shareholders from trading and thus should lead to a negative association between state ownership and stock liquidity.

In contrast, the soft-budget-constraint view holds that government ownership has a number of benefits, including an implicit guarantee of rescue in times of financial distress (Faccio et al., 2006; Borisova and Megginson, 2011; Boubakri et al., 2012) , prolonged and easier access to finance (Cull, Xu, and Zhu, 2009; Chaney, Faccio, and Parsley, 2011) , and availability of subsidies from the state budget or tax concessions (e.g., remission, reduction, or deferral of taxes) as well as other means of indirect support. Faccio et al. (2006) , for instance, show that politically-connected firms are more likely than non-politically-connected firms to be bailed out by the state. Similarly, Boubakri et al. (2012) find that politically-connected firms enjoy a lower cost of equity, especially in countries where the likelihood of a government bailout is higher.

Chaney, Faccio, and Parsley (2011) additionally observe that investors do not penalize politically-connected firms for lower earnings quality by requiring higher returns. This suggests that investors value the benefits that such firms receive by being linked to the government. As a result, investors may be more willing to buy the shares of NPFs, thus increasing their liquidity.

Building on these studies, and to the extent that the state is more inclined to support firms with connections to the government (such as NPFs with residual state ownership), the presence of the state as a blockholder enhances the liquidity of firms with residual government shareholdings.

We also recognize that there are negative implications associated with the soft budget constraint, especially in countries with left-wing governments (which are more inclined to use J o u r n a l P r e -p r o o f Journal Pre-proof SOE resources for political expediency). 6 For example, noting the costs of the soft budget constraint, Megginson and Netter (2001) contend that a driver of post-privatization efficiency improvements is the motivation brought on by the elimination of the -safety net‖ of the soft budget constraint (that was previously provided to state-owned firms but is no longer available following privatization). While certainly contributing to greater motivation and focus, working without a safety net also, by definition, increases risk. The findings of Boubakri et al. (2018) indicate that the risk-reduction benefits of having the safety net (i.e., the soft budget constraint) outweigh the efficiency improvements that result from removing it. Specifically, Boubakri et al. (2018; p.52) identify that investors assign greater importance to the benefits of the soft budget constraint and conclude that -easier and sustained access to financial resources provides government owned firms with a significant comparative advantage‖. Therefore, following Boubakri et al. (2018) , we contend that the comparative advantage from the soft budget constraint should contribute to greater liquidity for the stock of state owned firms.

Accordingly, the two competing views (soft budget constraint and political view) suggest that the relation between stock liquidity and state ownership is ultimately an open question. It is important to note that our hypotheses are based on the countervailing influences of two factors that are both unique to state ownership. Political benefits of ownership are only valuable to owners who are politicians. As an example of the use of an SOE for political purposes, stateowned firms may choose to overstaff in order to create jobs and thus curry political favor with voters. Winning voters is very important to politicians but is not important to other types of blockholders. Therefore, the political benefits (as per the -political view‖) apply to state owners (politicians) but would not apply to other types of blockholders.

Similarly, the soft-budget-constraint refers to funding advantages provided by the state to firms with state ownership. That is, the soft-budget-constraint results in state-owned firms receiving financing advantages that would not be available to firms owned by other types of blockholders. Highlighting the exclusivity of this benefit to state-owned firms, Megginson and Netter (2001) contend that a main driver of post-privatization efficiency improvements is the motivation brought on by the elimination of the safety net of the soft-budget-constraint (that was previously provided to state-owned firms but is no longer available following privatization).

Therefore, the financing benefits as per the -soft-budget-constraint view‖ apply to state owners but would not apply to other blockholders. 7

In the following empirical analysis, we consider how this specific trade-off (political view vs soft-budget-constraint view) affects the relation between state ownership and stock liquidity. 8 We formalize our primary predictions as follows:

H1: The level of state ownership affects stock liquidity.

H1a: Under the political view, residual state ownership is negatively related to stock liquidity. 7 Given these unique views of state ownership, substantial literature exclusively focuses on the impact of state ownership (Ben-Nasr et al., 2012; Borisova and Megginson, 2011; Borisova et al., 2012 Borisova et al., , 2015 Boubakri et al., 2013; Chen et al., 2018; Holland, 2019; Jaslowitzer et al., 2018, among others) . 8 It would be interesting to compare the liquidity implications of changes in state blockholdings to the liquidity implications of changes in other types of blockholdings (e.g., changes in ownership blocks by founding families, private equity, etc.). However, our specific hypotheses, by focusing on factors unique to state blockholders, would not facilitate such a comparison. Nevertheless, we recognize the importance of considering how changes in ownership by different types of blockholders may have different effects on stock liquidity. While beyond the scope of this paper, we identify such comparisons as interesting avenues for future research.

H1b: Under the soft-budget-constraint view, residual state ownership is positively related to stock liquidity.

To empirically examine the relation between state ownership and stock liquidity, we construct a sample of 473 NPFs from 53 countries over the period 1994-2014. Our initial data are from Boubakri et al. (2013) , which we update using Privatization Barometer, Thomson Reuters, the SDC Platinum Global New Issues, and SDC Platinum Mergers & Acquisitions databases. By using these data to track the change in government shareholdings after the first privatization, we investigate how the effect of state ownership on stock liquidity varies over time.

We obtain stock liquidity and financial statement information from Compustat Global and ownership statistics from Boubakri et al. (2013) , firms' annual reports, Bureau van Dijk's Osiris database, and Bloomberg. Because the behavior of financial firms (SIC codes between 6000 and 6999) is heavily influenced by a country's regulatory environment, we exclude these firms from our analysis. After also removing observations with missing data, our final sample contains 3,759 firm-year observations. ************************ Insert Table 1 Here ************************ Following Lesmond et al. (1999) or LOT, we first measure firm-level stock liquidity using the proportion of trading days with zero returns during the year (ZEROS). The denominator of ZEROS is the actual number of a firm's total trading days in a given year on its respective exchange. 11 Securities with lower liquidity are likely to have more zero-volume days and thus more zero-return days. Bekaert, Harvey, and Lundblad (2007) show that zero-return days is a good measure to predict future returns in emerging markets compared with alternative measures such as turnover. Moreover, they argue that transaction data (such as bid-ask spreads) are not widely available in emerging markets, while zero-return days only require a time-series of daily equity returns. Lesmond (2005) presents evidence that the LOT statistic (i.e., ZEROS) captures cross-country liquidity effects better than other metrics. As a robustness check, we measure stock liquidity using Fong et al.'s (2017) variable FHT, a percent-cost proxy that simplifies the LOT measure. Moreover, we adopt an alternative liquidity measure (e.g., Bekaert et al., 2007; 9 Our main findings are not sensitive to sequentially excluding each country from our analysis. 10 All of our inferences continue to hold when we sequentially exclude each industry from our analysis. 11 To account for zero-return days due to holidays or market closures, we calculate ZEROS excluding days when there are more than 5 or 10 consecutive days of zero returns. In unreported tests, we confirm that our main findings are statistically unchanged. We thank an anonymous reviewer for raising this point. Goyenko, Holden, and Trzcinka, 2009), namely, AMIHUD (Amihud, 2002) . This metric is the average across stocks of the daily ratio of absolute stock return to dollar volume. 12

Because ZEROS, FHT, and AMIHUD reflect stock illiquidity, higher values of these metrics indicate lower stock liquidity. We summarize the definitions for these and all other variables in the Appendix.

We emphasize that our metrics focus on firm-level liquidity. Alternatively, Boutchkova and Megginson (2000) and Bortolotti et al. (2007) use the market turnover ratio as a countrylevel measure of liquidity and provide evidence that privatization affects aggregate liquidity.

Specifically, Boutchkova and Megginson (2000) find a positive relation between the turnover ratio of a market and the number of privatizations in the country. Bortolotti et al. (2007) , also using a turnover-based measure, further document an increase in aggregate liquidity for privatized IPOs in a sample of 19 OECD countries between 1985 and 2002. However, the previously used measures of aggregate liquidity (particularly the country-level turnover variables), while offering broad insights regarding country-level liquidity, are less well-suited for assessing firm-level liquidity. First, as noted by Jun, Marathe, and Shawky (2003) , there is an important distinction between the liquidity of an individual stock and the liquidity of the total equity market. 13 Bortolotti et al. (2007) and Boutchkova and Megginson (2000) do an excellent job of addressing the latter, but do not address the former.

Also, as in many empirical endeavors, there are trade-offs regarding the choice of statistical measures. Bortolotti et al. (2007) , citing Pástor and Stambaugh (2003) , concede that the turnover ratio may not always accurately reflect market liquidity. Historically, there have been market environments exemplified by high levels of turnover but low degrees of market liquidity (such as October 1987). Lee and Swaminathan (2000) additionally identify a relation between trading volume and past price momentum and thus warn that turnover measures may provide less reliable assessments of market liquidity. Specifically, Lee and Swaminathan (2000, p.2061) conclude, -This evidence further supports the notion that past turnover is a measure of fluctuating investor sentiment and not a liquidity proxy‖. Accordingly, in our study, we attempt to overcome these potential weaknesses of the previously used measures of aggregate liquidity (i.e., the turnover ratios) by applying more precise measures of firm-level liquidity [such as those developed by Lesmond, Ogden, and Trzcinka (1999) and by Fong, Holden, and Trzcinka (2017)].

We capture state ownership using the percentage of shares held by a government (STATE). Our regressions also include several firm-and country-level control variables to ensure that the relation between state ownership and stock liquidity is not driven by confounding factors.

At the firm level, we follow prior literature (e.g., Lang, Lins, and Miller, 2004; Lang, Lins, and Maffett, 2012; Stoll, 2000) and control for firm size as measured by the log of a firm's market value of equity (LOG MV), book-to-market (BM), return variability (STDRET), transparency as reflected by earnings smoothness (EM), analyst coverage as indicated by the number of analysts forecasting current-year earnings (ANALYST), and an indicator for whether the firm had a loss (LOSS). We also include indicator variables for whether the stock trades in the U.S., either on an exchange (ADR_EX) or on the OTC or PORTAL markets (ADR_NEX). Trading in the U.S. is likely to lead to higher transparency (Lang, Lins, and Miller, 2003) , and it may also draw liquidity from local markets to the extent that shares are less costly to trade in the U.S. (Baruch, J o u r n a l P r e -p r o o f Karolyi, and Lemmon, 2007) . We further control for whether the firm reports under IFRS or U.S. (2000) show that the securities of firms that convert to IAS or U.S. GAAP are associated with higher liquidity; we therefore expect a positive relation between INTGAAP and stock liquidity. Moreover, we control for the stock trading activities (STOCK TURNOVER), stock price (LOG (PRICE)), and stock trading days (LOG (TRADING DAYS)). 14 At the country level, we control for institutions that are likely to influence the extent to which firm-level transparency affects liquidity (e.g., Lesmond, 2005; Lang et al., 2012) .

Specifically, we include the number of listed firms in the country (LISTED) to control for the level of stock market development, the extent of press freedom (MEDIA) to indicate the degree of media penetration, and log GDP per capita (LGDPC) to capture aggregate income. 15

Panel A of Table 2 reports summary statistics. We find that ZEROS has a mean (median) of 0.11 (0.07). Residual state ownership (STATE) has a mean (median) of 0.27 (0.18), in line with a sharp decline in state ownership after privatization (Boubakri, Cosset, and Guedhami, 2005) . Panel B of Table 2 presents Pearson correlation coefficients among key variables. As can be seen, state ownership is negatively correlated with all measures of stock illiquidity, indicating that higher state ownership is associated with higher stock liquidity. ************************ Insert Table 2 Here 14 We thank an anonymous reviewer for suggesting these controls. 15 In countries with better stock market development and higher income, we expect stock liquidity to be higher. Similarly, we expect that when media penetration is poor, corporate governance may be less effective, which may also reduce stock liquidity (e.g., Lang et al., 2012) .

J o u r n a l P r e -p r o o f ************************

In Table 3 , we perform univariate analysis of the relation between state ownership and stock liquidity. In Panel A, we first split the sample of privatized firms into two groups: partially privatized firms (Column 2) and fully privatized firms (Column 3). We find that partially privatized firms have significantly higher stock liquidity than fully privatized firms. These results, which suggest that some residual state ownership in NPFs enhances liquidity, are consistent with the soft-budget-constraint hypothesis.

In Panel B, we examine the relation between partially privatized firms and fully privatized firms during the GFC. We find that partially privatized firms show higher stock liquidity than fully privatized firms, regardless of the timing relative to the financial crisis. This indicates that the residual government ownership, by endowing partially privatized firms with the security of state support, contributes to greater stock liquidity.

Taken together, the results of the univariate analysis provide preliminary evidence that the soft budget constraint associated with state ownership may contribute to higher stock liquidity. In the following section, we further explore these relations with our multivariate analysis. ************************ Insert Table 3 Here ************************

In Table 4 , we first examine the impact of state ownership on stock liquidity by using a continuous measure of state ownership (STATE) as our independent variable of primary interest.

We use ZEROS as the dependent variable and estimate the following model (subscripts omitted for simplicity):

To control for within-firm correlation, we present significance levels based on robust standard errors adjusted for clustering at the firm level.

In Model (1), we find that STATE is negatively associated with ZEROS. This relation is statistically significant at the 5% level. It is also economically significant, with the coefficient on STATE suggesting that, all other variables held constant, increasing state ownership by one standard deviation will result in a 6.1% (= 0.28 × (-0.024) / 0.11) decrease in zero-return days (increase in stock liquidity). Thus, in line with the univariate results, these findings support the soft-budget-constraint view of state ownership. ************************ Insert Table 4 Here ************************ ************************ Insert Figure 1 Here ************************ J o u r n a l P r e -p r o o f Model (2) explores the non-monotonic relation between state ownership and stock liquidity using a quadratic model. We continue to find that STATE loads with a negative coefficient (that is statistically significant at the 1% level). Additionally, we find that the quadratic term STATESQR has a positive coefficient (that is statistically significant at the 1% level). These results confirm the curvilinear relation between state ownership and stock liquidity.

This finding is similar to that of Borisova and Megginson (2011) who find that the cost of debt is non-monotonically related to residual state ownership.

Further, the Model (2) results (illustrated in Figure 1 ) show that stock liquidity is highest at an inflection point of 44% government ownership, a level consistent with the government retaining some influence over the firm. This suggests that reducing state ownership to lower levels may decrease government influence to the point that the benefits of the soft budget constraints are diminished (which reduces the NPF's liquidity). However, when government ownership exceeds 44%, liquidity is also adversely affected. This is consistent with the political view of state ownership, which holds that investors exhibit greater fear of the -grabbing hand‖ of political interference as government ownership increases. Overall, this non-monotonic relation appears to be reflective of a trade-off between the costs (political view) and the benefits (softbudget-constraint view) of state ownership.

In Column (3) of Table 4 , we replace the continuous state ownership metric (STATE) with a dummy variable PARTIAL (which indicates whether a government retains shares in a firm after privatization (i.e., STATE > 0)) as an alternative independent variable. We estimate the following specification (subscripts omitted for simplicity): ZEROS = α + β 1 PARTIAL + β 2 LOG MV + β 3 BM + β 4 STDRET + β 5 EM + β 6 ANALYST + β 7 LOSS + β 8 ADR_EX + β 9 ADR_NEX + β 10 INTGAAP J o u r n a l P r e -p r o o f Journal Pre-proof + β 11 STOCK TURNOVER + β 12 LOG(PRICE) + β 13 LOG (TRADING DAYS) + β 14 LISTED + β 15 MEDIA + β 16 LGDPC + β 17 COUNTRY FIXED EFFECTS

We find that the coefficient on PARTIAL is negative and statistically significant at the 5% level, suggesting that partially privatized firms are associated with higher stock liquidity than fully privatized firms. This finding is also economically significant in that a firm that is fully privatized observes on average 12.7% (= (0-1) × (-0.014) / 0.11) more zero-return days (all other variables constant) and therefore exhibits significantly lower stock liquidity than a firm that is partially privatized. These results indicate that partial privatization is associated with higher stock liquidity. Consistent with the soft-budget-constraint view of state ownership, these results identify that firms have higher liquidity when the government retains some shares in NPFs.

In the context of privatization, a major econometric concern is selection bias. As Megginson and Netter (2001, p.346 ) point out, -sample selection bias can arise from several sources, including the desire of governments to make privatization look good by privatizing the healthiest firms first‖. 16 Also, governments may retain larger stakes in firms with higher liquidity to extract greater private/political benefits. In addition, the relation between state ownership and stock liquidity could be driven by unobserved determinants of liquidity that also explain residual state ownership. We address these issues using several approaches in Table 5 . ************************ Insert Table 5 Here ************************ 16 Bortolotti and Faccio (2009) confirm that governments are more likely to first privatize firms that are more valuable and more profitable. (2) of Panel A reports the results of the second-stage regression. We again find that STATE is significantly negatively associated with ZEROS.

In Model (3), we treat both state ownership and state ownership squared as endogenous variables. Again, in the first-stage regression (Model (1) of Panel A), we regress STATE on STATE COUNTRY together with the full set of control variables. We then use the predicted state ownership and the squared value of predicted state ownership in the second-stage regression of stock liquidity on state ownership. We find that state ownership is associated with fewer zeroreturn days, while state ownership squared is associated with more zero-return days. This confirms the existence of a nonlinear relation between liquidity and state ownership.

In Models (4) and (5), we perform a Heckman two-stage analysis to address sample selection bias. In the first stage, we use a Probit model to predict whether governments retain control over privatized firms. We regress Control (a dummy variable indicating whether 17 Our findings are robust to using the log of the number of employees at a firm and government deficits/GDP as instruments for state ownership (following Borisova and Megginson, 2011).

J o u r n a l P r e -p r o o f governments retain more than 50% of privatized firms) on STATE COUNTRY, the full set of control variables, and country, industry, and year fixed effects. This step allows us to estimate the inverse Mills ratio (Lambda). In the second stage, we include LAMBDA as an additional independent variable in the liquidity regression. The results in Model (4) show that the coefficient on STATE is significantly negative (at the 1% level), indicating that stock liquidity increases as residual state ownership increases. In Model (5), we further find that the coefficient for STATESQR is significantly positive. This reinforces our earlier evidence of a nonlinear relation between state ownership and stock liquidity.

Models (6) and (7) employ propensity score matching, which allows us to randomize the sample selection by using observable firm characteristics to match privatized firms under government control with those that are not. 18 In the first stage, we use the same Probit model as in the Heckman first-stage analysis. We then match state-controlled firms to NPFs (not controlled by the state) with the closest propensity score. In the second stage (Models (6) and (7)), we estimate the regressions using the matched sample. Consistent with our main analysis, the instrumental variables analysis, and the Heckman analysis, we continue to find that state ownership is nonlinearly related to stock liquidity.

In this section, we further consider how soft budget constraints may affect the relation between state ownership and stock liquidity. To do so, in Model (1) of Table 6 we interact state ownership with a measure of the government ownership of banks (GOVBANK). We expect stateowned firms to benefit from preferential access to financing in countries with higher government 18 We obtain statistically similar results when we match firms that are partially privatized (i.e., State Ownership > 0) with firms that are fully privatized (i.e., State Ownership = 0). J o u r n a l P r e -p r o o f ownership of banks (e.g., Chen et al., 2018; Jaslowitzer et al., 2018; Frydman et al., 2000) . We estimate the following model (subscripts omitted for simplicity):

In Model (1), we examine whether superior access to state-owned banks explains the greater stock liquidity of NPFs with larger residual state ownership. Barth, Caprio, and Levine (2013) provide data measuring government ownership of banks (GOVBANK) at the country level.

We find that the coefficient on STATE × GOVBANK is negative and statistically significant at the 1% level. Consistent with the soft-budget-constraint view, this indicates that the liquidityenhancing effects of residual state ownership are stronger in countries with more government ownership of banks. 19 ************************ Insert Table 6 Here ************************ Because the state ownership of banks may amplify the financing advantages provided to SOEs, we expect the benefits of state ownership to be greater in countries with a higher prevalence of state-owned banks (SOBs). In Figure 2A , we measure how the inflection point 19 In unreported results, we also proxy for soft budget constraints using the extent to which foreign banks are allowed to enter a country's banking industry and own domestic banks (LIMFOREIGN). A higher value indicates fewer restrictions on foreign entry and therefore less comparative advantage of state ownership. Additionally, we capture soft budget constraints using the degree to which the supervisory authority is independent of political influence (POLITICAL INDP). In both supplemental regressions, we find that when there is comparative advantage of state ownership in terms of access to finance, state ownership is associated with greater stock liquidity. identify that politically-connected firms have a lower cost of borrowing (even though their reporting quality is poorer). Therefore, to the extent that the soft-budget-constraint view is true, we should observe a stronger relation between state ownership and stock liquidity during the J o u r n a l P r e -p r o o f financial crisis period. We test this conjecture in Model (2) of We find that the coefficient on the interaction term between STATE and DURING CRISIS is negative and statistically significant. This identifies that when state ownership increases, the stocks of these firms became more liquid during the financial crisis. This is consistent with an increase in the liquidity-enhancing effect of the soft budget constraint during the crisis.

In Models (2) and (3) of Table 6 , we more extensively consider how the financial crisis may have affected the relation between state ownership and firm-level stock liquidity.

Specifically, we focus on observations from 2008 to 2010. Supporting our conjecture from the previous analysis, the data indicate that state ownership significantly enhances firm-level stock liquidity during the years of the financial crisis. Interestingly, the coefficient on STATESQR is positive but statistically indistinguishable from zero, suggesting that the non-monotonic relation between state ownership and liquidity weakens during the crisis years. This evidence suggests J o u r n a l P r e -p r o o f that the benefits (soft-budget-constraint view) of state ownership become more valuable during crisis periods and overcome the costs (political view) of state ownership.

Overall, these findings are consistent with our expectation that the specter of the financial crisis substantially heightened investor appreciation of the bailout potential and other fiscal advantages stemming from the state's soft budget constraint. 20 As such, the crisis-induced increase in the benefits of government ownership contributed to greater firm-level liquidity for NPFs with larger residual state shareholdings. 4.5. State ownership and stock liquidity -Political view (i.e., the -grabbing hand‖ effect) So far, we have shown how the soft budget constraint associated with state ownership helps improve stock liquidity. In this section, we further explore the costs of state ownership.

Specifically, the -grabbing hand‖ effect suggests that the costs of state ownership become higher as the government retains a higher stake of privatized firms. Fear of the -grabbing hand‖ leads to less demand for NPFs' stock and reduced liquidity. One may argue that because governments tend to sell more profitable firms, the negative relationship between high state ownership and stock liquidity is simply driven by firms' profitability. In Model (1) of Table 7 , we disentangle the -grabbing hand‖ effect from the profitability effect. To isolate the impact of the -grabbing hand‖, we interact state ownership with earnings management (EM), a proxy for the expected agency costs of expropriation (Haw et al., 2004) . Prior accounting literature suggests that opaque financial reporting, evident in higher earnings management, help controlling shareholders hide their extraction of private benefits of control at the expense of minority shareholders (Leuz et al., 20 This result also augments the conclusion from Beuselinck et al. (2017) that state ownership had a favorable valuation effect during the financial crisis. That is, the stronger positive relation between state ownership and firm-level liquidity during the crisis years (that we identify in Table 6) Haw et al., 2004; Kim and Yi, 2006; Gopalan and Jayaraman, 2012) . Consistent with this view, Leuz (2006) , Gopalan and Jayaraman (2012) , and Attig et al. (2020) report higher earnings management in closely-held firms. We find that the interaction term between STATE and EM loads positive and is statistically significant. This suggests that stock liquidity is lower when there is a higher level of earnings management (which is symptomatic of greater intervention by the state). Therefore, the results are consistent with the -grabbing hand‖ effect of government ownership. 21 Moreover, in Model (2) of Table 7 , we replace EM with ROA to examine the effect of profitability on the relationship between state ownership and stock liquidity. The interaction term is statistically insignificant. Taken together, the results in Table 7 support the political view of state ownership. ************************ Insert Table 7 Here ************************ To further validate the -grabbing hand‖ effect, we construct a dummy variable, LEFT, which equals one when the political orientation of a country's ruling executive is communist, socialist, social democratic, or left-wing, and zero otherwise. We obtain this metric from the Database of Political Institutions (Beck et al., 2001) . 22 Justification for this variable is based on the political interference hypothesis (Boycko et al., 1996; Shleifer, 1998; Shleifer and Vishny, 21 Using the country median ratio of the firm-level standard deviations of income and cash flow as an alternative proxy for earnings management (Leuz et al., 2003) , we find our results remain statistically the same. Shleifer and Vishny (1994) present evidence that left-wing governments attach greater value to the political benefits obtained by directing SOE resources to favored constituents (such as by creating jobs for public sector employees). Chen et al. (2020) find that left-wing governments are more likely to use SOEs to grant larger amounts of trade credit, which they show is politicallymotivated and value-reducing. Therefore, because left-wing governments will generally be more inclined to use SOE resources for political expediency (as opposed to economic optimality), investors should be more apprehensive about potential expropriation. Providing evidence that left-oriented governments may be willing to sacrifice shareholder wealth maximization in order to achieve political objectives, Holland (2019) (2020) document that efforts to achieve those political goals have negative implications for shareholder value (and those negative implications are significantly more severe in countries with left-wing governments). Overall, these studies identify that politically motivated endeavors by SOEs have real economic costs for minority investors. Minority shareholders may opt to avoid these costs by choosing not to hold or trade the stocks of state-owned firms, which may reduce the liquidity of these securities. Accordingly, we expect that firms with higher state ownership will exhibit lower stock liquidity in countries with left-wing governments. 23 To test this prediction, we estimate the following model (subscripts omitted for simplicity): 23 One concern is that political orientation is highly correlated with the economic development of countries so that our measure of political orientation is capturing the effect of economic development J o u r n a l P r e -p r o o f ZEROS = α + β 1 STATE + β 2 LEFT + β 3 STATE × LEFT + β 4 LOG MV + β 5 BM + β 6 STDRET + β 7 EM + β 8 ANALYST+ β 9 LOSS + β 10 ADR_EX + β 11 ADR_NEX + β 12 INTGAAP + β 13 STOCK TURNOVER + β 14 LOG(PRICE) Table 8 Here ************************ We present results from this specification in Table 8 . We find that the coefficients of STATE × LEFT load positively and are statistically significant. This indicates that stock liquidity is lower when state ownership is higher in countries with left-wing governments.

Importantly, the coefficient of STATE × LEFT is 0.05, suggesting that firms from nations with left-wing governments have 45% (=0.5/0.11) more zero-return days. This finding is consistent with the view that investors are less likely to invest in privatized firms with higher potential for government expropriation.

In additional analysis, we re-estimate the models of Table 8 by splitting our sample into firms from countries with left-wing and center/right-wing governments. We plot the results in Figure 2B . For firms in countries with left-wing governments, we find that the inflection point decreases from 44% to 21%. This suggests that shareholders in countries with left-wing governments are more reluctant to invest once the state retains more than 21% of the firm (due to the greater fear of political intervention in countries with left-wing governments). Interestingly, we find that the inflection point increases to 55% for nations with center/right-wing governments, (rather than the extent of political costs of state ownership). In our sample, the average score of LEFT for developed countries is 0.29, while the average score for developing countries is 0.30. Moreover, developed countries account for 2,243 of our observations (i.e., 57%), while developing countries account for 1,684 observations (i.e., 43%). Taken together, our data are not primarily skewed toward developed or developing countries in terms of government political orientation.

suggesting that investors are more tolerant of higher state ownership if the government is less likely to be involved in economic activity (and is therefore less likely to expropriate minority shareholders). 24 ************************** Insert Figure 2B Here **************************

Another concern with our main analysis is that the relation between state ownership and stock liquidity may have alternative explanations. We investigate several possibilities in Table 9 .

First, Lesmond (2005) finds that political risk helps explain stock liquidity in emerging markets.

Our results could therefore be driven by an omitted variable-political risk. Second, Megginson et al. (2004) show that countries with more equal income distributions have a broader base of potential shareholders, which can contribute to greater country-level liquidity. Third, Sarkissian and Schill (2004) argue that firms producing tradeable goods have wider name recognition, and the stocks of these firms are more warmly received by potential shareholders in those markets and thus may be more liquid.

We test these possibilities in Models (1) and (2) by including the political risk measure of the International Country Risk Guide (ICRG) (POLRISK), the income inequality measure from All the Ginis Dataset (INEQUALITY), and an indicator for firms that produce tradeable goods (TRADEABLE) as additional control variables. 24 We acknowledge that these comparisons of inflection points are descriptive in nature because we cannot test the statistical differences in inflection points across different regressions.

After including these additional controls, the coefficient on STATE remains negative and statistically significant at the 1% level. In Model (2), we continue to find that STATE is negatively and STATESQR is positively associated with zero-return days. ************************ Insert Table 9 Here ************************ Another concern is that our main results regarding the relation between state ownership and stock liquidity are driven by other types of blockholders. Specifically, residual state ownership in partially privatized firms may naturally induce lower liquidity since the state (relative to other blockholders or other investors) may be less inclined to actively trade shares.

Therefore, shares of privatized firms are less frequently traded. 25

To rule out this possibility, in Models (3) and (4) (2000) note significant differences in the share-ownership structure of NPFs and always-private firms. Specifically, after comparing the shareholder rosters of privatized firms to a capitalization-based matched sample of private firms, Boutchkova and Megginson (2000) find that privatized firms generally have a larger number of shareholders than private firms and that the composition of the shareholder base is more likely to change in NPFs. Therefore, even if the residual shares held by the state are likely to trade less frequently, there may a countervailing effect on the liquidity of privatized firms due to the larger number of shareholders and the dynamic nature of the shareholdings of those investors. Furthermore, our findings that differences in the institutional environment contribute to differences in the inflection points (from the cost/benefit trade-off) provide additional evidence to mitigate the potential conjecture that lower stock liquidity in NPFs is simply driven by fewer trading activities by the state. 26 In unreported tests, we examine the effects of ownership blocks by foreign and institutional owners measured at 10%, 20%, and 30% of proportional ownership. We also consider whether foreign and institutional owners are the largest shareholder in a firm. Our models indicate that these ownership J o u r n a l P r e -p r o o f

In Models (5) and (6), we address the concern that acquisition activities may affect stock trading and therefore drive our main results. To control for acquisition activity, we include a dummy variable (ACQUISITION), which is equal to one if a firm has acquisition expenditures that are larger than zero. Our main results are unaffected. 28 In Models (7) and (8), we exclude China from our sample to mitigate the concern that our results are driven by the country with the largest number of observations in our data. Moreover, in Models (9) (2017) find that this metric is useful in cross-country studies. In Models (13) through (16), we variables are not statistically significant, while our inferences on the role of state ownership are not affected. We thank an anonymous reviewer for suggesting this analysis. 27 As a further control, we add the number of shareholders (e.g., Chia et al., 2020) to Model 2 in Table 4 . Although the sample size drops significantly (756 firm-year observations, representing 20% of our full sample), we continue to find that state ownership and its squared term remain statistically significant, consistent with our main evidence. 28 To further confirm the robustness of our results, we account for changes in trading rules that may affect stock liquidity. We identify changes in trading rules and securities regulation from Cumming, Johan, and Li (2011), Bhattacharya and Daouk (2002 ), Edmans, Jayaraman, and Schneemeier (2017 ), and Fauver et al. (2017 . Our findings are unaffected by including these additional controls.

J o u r n a l P r e -p r o o f verify that our results are not sensitive to using AMIHUD or FHT as an alternative measure of liquidity.

Also, in unreported results, we use the Roll illiquidity measure (Roll, 1984) , which is a covariance spread estimator of stock illiquidity, and obtain similar results. ROLL is calculated as 2√

, where P t is the observed closing price on day t and is equal to the stock's true value plus or minus half of the effective spread.

In this section, we validate our main evidence regarding the relation between state ownership and stock liquidity by examining a sample of re-nationalized firms. Specifically, we use SDC Platinum to identify previously-privatized firms that governments have subsequently re-nationalized. Using propensity score matching, we then paired those re-nationalized firms with firms from the private sector (according to all firm and country characteristics ************************ Insert Table 10 Here ************************ J o u r n a l P r e -p r o o f As we report in Table 10 , RE-NATIONALIZATION loads positively and is statistically significant. This indicates that stock liquidity is lower for re-nationalized firms compared to private sector firms. This is consistent with the view that a greater level of investor fear of the -grabbing hand‖ leads to a lower level of stock liquidity.

We investigate the link between state ownership and firm-level stock liquidity. Our study contributes to the privatization literature by presenting unique firm-level evidence regarding the liquidity implications of privatization reforms across a broad sample of countries. 29 In particular, high levels of state ownership in NPFs could dissuade investors who fear the -grabbing hand‖ of the government, which would reduce the liquidity of newly privatized stocks and in turn increase their cost of capital and decrease their value. However, at least from a liquidity perspective, lower levels of state ownership may also be disadvantageous.

That is, especially during times when the scars of the financial crisis are still fresh, the financing advantages (and the implicit and not-so-implicit bailout guarantees) provided by government shareholdings may resonate with investors and thus enhance the liquidity of firms with some state ownership. Overall, consistent with a trade-off between the benefits and the costs of state ownership, our results suggest that there is a level of state ownership that maximizes the liquidity of the stock of NPFs. 29 These findings are especially important because a primary objective of privatization programs in many countries is the development of stock markets by promoting an -equity culture‖ or -people's capitalism‖ among investors (e.g., Megginson and Netter, 2001; Boutchkova and Megginson, 2000; Megginson, Nash, and Van Randenborgh, 1994) . In turn, this equity culture is conducive to a change in the trading behavior of investors, thus affecting stock liquidity. By addressing the liquidity implications of state ownership at the firm level, we provide important insights to policymakers attempting to spur economic development by fostering an equity culture. J o u r n a l P r e -p r o o f 1.00 FHT 0.83 *** 1.00 AMIHUD 0.34 *** 0.37 *** 1.00 STATE -0.08 *** -0.06 *** -0.12 *** 1.00 STATESQR -0.05 *** -0.04 * -0.10 *** 0.95 *** J o u r n a l P r e -p r o o f Table 4 reports regression results relating partial privatization to stock liquidity. The sample comprises 3,759 firmyear observations representing 473 newly privatized firms from 53 countries over the period 1994-2014. The dependent variable is ZEROS, which is the percentage of days during the fiscal year that the stock price does not change and is calculated as ZEROS=ZeroReturnDays/Total Trading Days. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variable definitions and sources. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. Dependent variables: Table 5 reports regression results addressing endogeneity of state ownership using instrumental variables, Heckman two-stage selection, and propensity score matching. In the first-stage regressions, we regress state ownership (STATE) on the country-level state ownership (STATE COUNTRY), which is lagged 3 years, together with all control variables and year and industry fixed effects. The sample comprises 473 newly privatized firms from 53 countries over the period 1994-2014. The dependent variable is ZEROS, which is the percentage of days during the fiscal year that the stock price does not change and is calculated as ZEROS=ZeroReturnDays/Total Trading Days. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variable definitions and sources. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. Table 6 reports regression results relating soft budget constraints, state ownership, and stock liquidity. The sample comprises 3,759 firm-year observations representing 473 newly privatized firms from 53 countries over the period 1994-2014. The dependent variable is ZEROS, which is the percentage of days during the fiscal year that the stock price does not change and is calculated as ZEROS=ZeroReturnDays/Total Trading Days. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variable definitions and sources. tstatistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. Dependent variables: Table 7 reports regression results considering how the relation between government ownership and stock liquidity is potentially affected by risk of expropriation and profitability. The dependent variable is ZEROS, which is the percentage of days during the fiscal year that the stock price does not change and is calculated as ZEROS=ZeroReturnDays/Total Trading Days. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variable definitions and sources. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. Table 8 reports regression results relating government orientation, state ownership, and stock liquidity. The sample comprises 3,759 firm-year observations representing 473 newly privatized firms from 53 countries over the period 1994-2014. The dependent variable is ZEROS, which is the percentage of days during the fiscal year that the stock price does not change and is calculated as ZEROS=ZeroReturnDays/Total Trading Days. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variable definitions and sources. tstatistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. Table 9 reports regression results relating state ownership to stock liquidity using additional controls and alternative dependent variables. The full sample comprises 3,759 firm-year observations representing 473 newly privatized firms from 53 countries over the period 1994-2014. The dependent variable in Models (1) to (12) is ZEROS, which is the percentage of days during the fiscal year that the stock price does not change and is calculated as ZEROS=ZeroReturnDays/Total Trading Days. The dependent variable in Models (13) and (14) is AMIHUD, which is the average stock return over trading volume. The dependent variable in Models (15) and (16) is FHT, which is a liquidity proxy based on low-frequency data and is defined as 2×Sigma×Probit((1+ZEROS)/2), where Sigma=Std (Returns). We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variable definitions and sources. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. Table 10 reports regression results relating government re-nationalization and stock liquidity. The dependent variable is ZEROS, which is the percentage of days during the fiscal year that the stock price does not change and is calculated as ZEROS=ZeroReturnDays/Total Trading Days. We winsorize all financial variables at the 1% level in both tails of the distribution. The Appendix provides variable definitions and sources. t-statistics based on robust standard errors clustered at the firm level are in parentheses below each coefficient. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. Authors' calculation based on Amihud (2002) ROLL A liquidity proxy developed by Roll (1984) , which is: ROLL = 2√ , where P t is the observed closing price on day t and is equal to the stock's true value plus or minus half of the effective spread.

Author's calculation based on Roll (1984)

A dummy variable equal to one if the government remains the largest shareholder in a privatized firm, and zero otherwise. Firm's annual report PARTIAL A dummy variable equal to one if the government retains shares (i.e., STATE >0) in a privatized firm, and zero otherwise. As above

The percentage of state ownership. As above

The square of state ownership.

Author's calculation

The percentage of foreign ownership. Firm's annual report

The percentage of institutional ownership. As above

The log of the market value of equity at year-end. Compustat Global

The book value of common equity divided by the market value of equity. As above

The annual standard deviation of daily stock returns. As above

The standard deviation of income over the standard deviation of cash flows. As above

The number of analysts that follow the firm. I/B/E/S LOSS A dummy variable equal to one if net income before extraordinary items is negative, and zero otherwise. Compustat Global ADR_EX A dummy variable equal to one if the firm trades on a U.S. exchange during the year, and zero otherwise. As above J o u r n a l P r e -p r o o f Journal Pre-proof ADR_NEX A dummy variable equal to one if the firm has an ADR but is not traded on a U.S. exchange during the year, and zero otherwise.

As above

A dummy variable equal to one if the firm reports under IFRS or U.S. GAAP during the year, and zero otherwise. As above

The natural logarithm of total assets. As above

Total debt over total assets. As above

Cash and short-term investment divided by total assets. As above CAPX Capital expenditure divided by total assets. As above

A dummy variable that equals one if dividend payout is greater than zero, and zero otherwise. As above

Income before extraordinary items, plus R&D expenditures and depreciation, all deflated by total assets. As above NWCAP Current assets minus current liabilities, delated by total assets. As above

The natural log of stock price. As above

The natural log of the number of trading days for a firm in a given year. As above

Forecast optimism bias defined as the difference between the one-year-ahead consensus earnings forecast and realized earnings delated by June-end stock price.

Author's calculation based on I/B/E/S data

Inflation rate of a country. World Bank

A dummy variable that equals one for chemicals, consumer goods, electronics, manufacturing, healthcare, mining, oil and gas, and paper industry, and zero otherwise.

A dummy variable equal to one if a firm has acquisition expenses that are larger than zero, and zero otherwise. As above

Stock turnover ratio of each firm, defined as the stock's total trading volume divided by the total outstanding shares. As above J o u r n a l P r e -p r o o f Journal Pre-proof RE-NATIONALIZATION A dummy variable that equals one for previously privatized firms that are acquired by government or government-controlled entities, and zero otherwise.

The number of firms listed on a nation's stock market. World Bank

A variable that rates each country's media freedom from 0 to 100. Transformed to 100 minus the original Freedom House index so that higher values indicate that a country's media are more independent.

LGDPC 

The average level of state ownership by each country (lagged 3 years).

Author's calculation POLRISK A variable measured as an amalgamation of 12 country elements and that ranges from zero to 100. A higher value indicates less political risk.

International Country Risk Guide (ICRG)

The extent to which the banking system's assets are government-owned. Specifically, this metric reflects the percentage of banking system's assets in banks that are 50% or more government-owned (based on surveys conducted by the World Bank in 1999 Bank in , 2003 Bank in , 2007 Bank in , and 2011 .

The extent to which foreign banks may enter a country's banking industry and own domestic banks. A higher value indicates fewer restrictions.

As above

The degree to which the supervisory authority is independent of political influence. A higher value indicates greater independence. 

A dummy variable that indicates time periods before the global financial crisis. Equal to one for period before 2008, and zero otherwise.

The limits of discipline: Ownership and hard budget constraints in the transition economies

Actual share repurchases, timing and liquidity

Private control benefits and earnings management: Evidence from insider controlled firms

Do liquidity measures measure liquidity

Auditor choice in privatized firms: Empirical evidence on the role of state and foreign owners

Ultimate ownership, income management, and legal and extra-legal institutions

Government investment in publicly traded firms

Market liquidity and performance monitoring

State ownership and corporate investment

Liquidity and stock returns in emerging equity markets