Factors Affecting Stock Liquidity : Corporate Governance , ADRs and Economic Crisis

Listing for trading in one of the segments of the BM&FBovespa requiring enhanced corporate governance can be seen as a way to align the interests of agents and principals. One of the supposed benefits of adhesion to these segments is increased stock liquidity. This study analyzes a sample of firms listed on the BM&FBovespa through panel data with Huber-White correction. The hypotheses are a positive relation between listing in one of these segments and liquidity, and that the higher the governance level, the greater the effect on liquidity. The results indicate that in the period before the 2008 crisis, the companies as a whole listed in the special governance segments had more liquid shares. But this result extends only to firms listed in the Level 1 and Novo Mercado segments when the three segments are analyzed individually. The hypotheses could not be confirmed in the entire period analyzed, from 2000 to 2009, possibly because of the effects of the crisis. For companies traded in the Level 2 segment, higher liquidity was not observed in the periods studied. Additionally, companies with ADRs showed higher liquidity in relation to those listed in the enhanced governance segments, independent of the effects of the crisis.


INTRODUCTION
ood corporate governance practices are important for emerging and developed countries alike.In Brazil, the importance of adopting good governance standards has increased with the need to obtain funding from abroad at competitive costs.For this purpose, the São Paulo Stock, Mercantile and Futures Exchange (BM&FBOVESPA) created three trading segments for firms with enhanced governance: Level 1, Level 2 and Novo Mercado ("New Market").These levels are reserved for companies that voluntarily adopt governance safeguards in addition to those required by law.In this respect, these governance practices should give shareholders rights and guarantees that reduce the information asymmetry and agency conflict between minority and controlling shareholders, which can have a strong influence on managers (LA PORTA et al., 2000).
Theoretically, investors with more information about firms will be more willing to invest, meaning greater confidence in and demand for the securities of firms listed in these trading segments, causing an increase in the liquidity of their shares.This relationship between governance and stock liquidity is corroborated by Schadewitz & Blevins (1998, p. 41), who state that "rational investors, realizing the possible risks, avoid ownership in firms whose quantity and quality of disclosures are consistently below expectations."Therefore, increased stock liquidity can generate benefits for firms, such as lower cost of capital, as pointed out by Amihud & Mendelson (1986;2000), who found empirical evidence of this effect.
On the other hand, poor governance practices can result in the use of inside information for the self-benefit of managers, manipulation of earnings and conflicts of interest, as amply demonstrated by the cases of Enron, Tyco, Worldcom and ImClone.In this scenario or scandals, the quality of disclosure by firms has gained importance, and efforts have been made to assure stricter governance standards in many countries, such as the United States with the enactment of the Sarbanes-Oxley Act (SOX).Therefore, the adoption of better corporate governance practices is strongly related to the level of disclosure.This should be a critical factor for the success of listed corporations, based on the assumption that investors are more watchful of this aspect, particularly after cases of fraud or economic crisis.
The demand for information in the market can be affected by the institutional arrangements of countries, such as the legal system (LA PORTA; LOPES-DE-SILANES; SHLEIFER, 1998).In this respect, Brazil has a code law system.According to various studies, code law countries tend to provide less protection to investors and thus have less developed capital markets than do countries with a common law legal tradition (LA PORTA et al., 1997;BUSHMAN;SMITH, 2001).Despite this expectation for code law countries in general, the availability of stock market trading segments with stricter rules on corporate governance, seeking to increase protection to investors, should provide a reward in the form of increased liquidity.In this respect, both the BM&FBovespa and the Brazilian Securities Commission (CVM) state i that listing in one of the enhanced governance segments brings benefits, such as better institutional image; greater demand for shares, meaning higher prices; lower cost of capital; and greater protection of minority shareholders.For the equity market as a whole, these institutions affirm that such special listing increases liquidity and makes it easier to issue shares.These claims are empirically verifiable and justify applied studies to confirm or refute the existence of these benefits.
In this respect, various studies have sought to identify the relationships between governance practices and the benefits of these practices.An example is the technical report of KPMG ( 2009), which applied a checklist to investigate the differences of some governance practices between companies listed in one of the special levels of the BM&FBovespa and firms with ADRs.The results showed that firms with ADRs satisfied more of the items on the checklist.Based on this context, we propose to contribute to research in this area by examining the following question: What is the relationship between listing in the special governance segments of the BM&FBovespa and the liquidity of shares?
Therefore, our general aim is to analyze the relationship between adherence to one of the enhanced governance segment of the BM&FBovespa and the liquidity ii of shares.Our specific objectives are: i) to verify whether there is an association between stock liquidity and adherence to the Level 1 segment; ii) to verify whether there is an association between stock liquidity and adherence to the Level 2 segment; iii) to verify whether there is an association between stock liquidity and adherence to the Novo Mercado segment; and iv) if such a relationship is confirmed, to investigate whether this relationship is different between the three levels, i.e., if the benefit in terms of stock liquidity increases as the listing level rises.
Although our focus is on the relationship between liquidity and governance, we provide some additional considerations regarding firms that issue American Depositary Receipts (ADRs) because of their need to adopt stricter disclosure levels due to the requirements of the The article is organized into five sections including this introduction.Section 2 presents the theoretical foundation and briefly discusses some empirical results of studies on governance and stock liquidity; Section 3 explains the methodology; Section 4 presents and analyzes the results; and Section 5 contains our final considerations.

THE AGENCY PROBLEM AND THE RELATIONSHIP WITH CORPORATE GOVERNANCE
According to Jensen & Meckling (1976, p. 5), the agency relationship is defined as "a contract under which one or more persons (the principal(s)) engage another person (the agent) to perform some service on their behalf which involves delegating some decision making authority to the agent."If both sides try to maximize their utilities, it is logical to believe that the agent will not always act in the best interests of the principal, and instead will make decisions that can expropriate the principal's wealth.The causes of agency conflict were further investigated by Jensen & Meckling (1994), under the hypothesis that human nature is utilitarian and rational, leading people to maximize a utility function of preferences.This hypothesis is based on the fact that an individual tends to be more effective in meeting his own objectives than those of others.However, it must be considered that the agency problem in Brazil is mainly between the controlling and minority shareholders rather than between managers and shareholders as a group, due to the great concentration of voting shares (OKIMURA et al., 2004) and other corporate governance aspects iii , because the controlling shareholders exert strong influence on the managers.
According to Shleifer & Vishny (1997, p. 737), "corporate governance deals with the ways in which suppliers of finance to corporations assure themselves of getting a return on their investment."The term supplier here refers both to the shareholders (suppliers if equity capital) and lenders (suppliers of debt capital), and both groups can increase their funding depending on the level of protection against expropriation.La Porta et al. (2000, p. 3) complement the idea by stating that "corporate governance is, to a large extent, a set of mechanisms through which outside investors protect themselves against expropriation by the insiders."In this sense, Silveira (2002, p. 14) points out that corporate governance corresponds to the "set of incentive and control mechanisms that aim to harmonize the relationship between shareholders and managers, by reducing agency problems, in a situation of separation between ownership and control." Therefore, stricter corporate governance practices tend to align the interests between the shareholders and managers, or between the controlling and minority shareholders.The adhesion of firms to a special governance listing segment can also be seen as a solution to the adverse selection and moral hazard problems, since by doing so companies are signaling to the market they want to reduce the information asymmetry between market participants and managers and convey a sense of greater security to investors.

THE LEVELS OF CORPORATE GOVERNANCE OF THE BM&FBOVESPA
In December 2000, the BM&FBovespa iv created three trading segments reserved for firms with enhanced governance levels.The basic idea was to increase the value of the companies and the liquidity of their shares, because adhesion to higher governance standards tends to provide greater security to shareholders and higher quality and transparency of the information disclosed.With regard to the benefits of good governance practices, Iudícibus & Lopes (2004, p.182) add that a good corporate governance system aims to conciliate the interests of agents and principals.Besides this, the adhesion to stricter governance levels can signal to the market that the company wants to reduce information asymmetry and transmit greater security to investors.
The characteristics of these three trading segments can be summarized as follows: i) Level 1 requires companies to adopt practices that favor transparency and access to information, by disclosing more information than the minimum required by law, and to maintain more dispersed ownership, through a minimum free float of 25%; ii) Level 2 requires satisfying the rules of Level 1 and others, such as extending to all holders of common shares the same conditions obtained by the controllers upon the sale of control of the company, and at least 80% of this value to holders of preferred shares (tag along), as well as voting rights to preferred shares; and iii) Novo Mercado has the same rules and Level 2 but only allows the issuance of voting shares.Therefore, the Novo Mercado segment is the most rigorous in terms of governance, and Level I is the least rigorous.
Because the creation of these three trading segments by the BM&FBovespa aims to improve the quality of disclosure, reduce information asymmetry and broaden the shareholding base, theoretically adherence to these rules should have an effect on the liquidity of those firms' shares.This justifies the interest in analyzing whether migration to one of the enhanced governance levels affects the liquidity of these companies' shares.It is also interesting to verify if there is rising liquidity according to the corporate governance level, given the different rules, mainly related to the rights of shareholders, as discussed before.

EMPIRICAL EVIDENCE OF THE EFFECT OF CORPORATE GOVERNANCE ON STOCK LIQUIDITY
Various authors have found a positive relationship between the level of corporate governance and stock liquidity, such as Attig et al. (2006), Kanagaretnam, Lobo & Whalen (2007), Chen et al. (2007), Goh, Ng & Young (2008), Chung, Elder & Kim (2010) and Dumitrescu (2010). In Brazil, Carvalho (2003) found that adhesion to one of the three special segments of the Bovespa had a positive impact on the stock liquidity.Similar results were reported by Martins, Silva & Nardi (2006), Camargos & Barbosa (2006) and Chaves & Silva (2006).On the other hand, Aguiar, Corrar & Batistella (2004) did not observe a change in liquidity for companies listed in the Level 1 segment.Silveira & Barros (2008) investigated the determinants of governance of listed Brazilian companies and concluded that adhesion to enhanced governance levels of the BM&FBovespa did not appear to influence their level of governance.In light of the above, it can be noted that the relationship between corporate governance and stock liquidity is not clear and needs further study.Therefore, since the adoption of higher levels of governance implies practices that reduce information asymmetry and increase shareholders' rights, our hypotheses are: H1: adherence to the Level 1 trading segment of the BM&FBovespa causes an increase in stock liquidity; H2: adherence to the Level 1I trading segment of the BM&FBovespa causes an increase in stock liquidity; H3: adherence to the Novo Mercado trading segment of the BM&FBovespa causes an increase in stock liquidity; and H4: the stricter the requirements (information disclosure, shareholder rights, etc.) a firm must satisfy in a determined segment are, the greater will be the liquidity of its shares.
These hypotheses are justified by the fact that adhesion to one of the special trading segments is associated with giving more rights to shareholders and greater transparency to the market, meaning a possible reduction of information asymmetry.Because there are rising governance standards for the three trading segments, we assume that adherence to each one implies distinct changes in stock liquidity.

METHODOLOGY
The sample is composed of all firms listed on the BM&FBovespa.In the case of those with both common and preferred shares, we excluded the type with lower liquidity.We also excluded stocks that were never traded, leaving 521 firms.The distribution of the companies according to trading segment was as follows: 4% for Level 2; 9% for Level 1; 19% for Novo Mercado; and 69% for the traditional segment.Since our main objective is to verify whether adhesion to one of the special governance segments has a differentiated impact stock liquidity, we created four dummies to capture this effect: one for Level 1 (L1), another for Level 2 (L2), a third for Novo Mercado (NM) and a fourth for any of these three segments (CG).In each of the dummies, we attributed a value of 1 for companies in the particular segment(s) and 0 otherwise.The control variables are described below.
(i) We included a dummy variable for firms with ADRs, which must meet the criteria set by the SEC and the listing rules of the NYSE.In theory these are more rigorous than those required by the BM&FBovespa.Lima et al. (2011) found that firms with ADRs had a smaller bid-ask spread vi than other Brazilian firms.Hence, we expect to find evidence of a positive relationship between the ADR dummy and stock liquidity.
(ii) We also included a dummy variable for firms in regulated sectors (REG), among them telecommunications, electricity and insurance, as well as companies in the financial (iii) To control for the size effect, we used the logarithm of total assets (SIZE) (Chung, Elder & Kim;2010).
(iv) We included financing (FIN), defined by the logarithm of the ratio between total liabilities and stockholders' equity.According to Chen et al. (2007), this variable has a negative relationship with quality of disclosure.Assuming that lesser information asymmetry reduces the risk of adverse selection, investors should have stronger demand for shares of companies when there is less asymmetry, positively affecting those shares' liquidity.
Therefore, we expect a negative relationship between FIN and stock liquidity.
(v) We included return on assets (ROA), calculated as the net profit over total assets, to represent the economic performance of the firms.According to Chen et al. (2007), there is a positive relationship between return and disclosure.
(vi) We also included a measure of volatility, calculated as the standard deviation of ROA (VROA).Volatility is known as a measure of risk.Assuming that ROA is a proxy for firms' economic performance, the more volatile the performance indicators are, the greater the risk will be.Therefore, we expect a negative relationship between this variable and stock liquidity.
(vii) To capture the evolution of assets over time and the cyclical behavior of assets and the reflections on liquidity, we included a temporal variable (TIME), calculated as the mean of the logarithm of total assets in each year for the companies.
(viii) Finally, we included a dummy to identify periods in which a firm had negative equity (NEGEQ), assigning a value of 1 for negative equity and 0 otherwise equity.
We used the following empirical model to verify the behavior of stock liquidity during the sample period: Where: Rit representes the dummy variables for governance described above: L1, L2, NM or CG.For example, in section 4.2, the first model contains only CG, while the second model contains the variables L1, L2 and NM.The use of the panel data method is justified by our interest in observing the effect of the governance variables on liquidity from January 2000 to December 2009 rather than in a determined year, which could be done in a cross-sectional regression model.According to Wooldridge (2006, p. 416), the basic reason to employ the panel data method is that it takes into account that the unobserved effects are correlated with the explanatory variables.We ran the statistical tests with the STATA version 9.2 software.
Observation of the average of LIQ during the quarters of the study period shows there was growth from the first quarter of 2000 to the start of 2007 and liquidity declined in [2008][2009], which includes the worst part of the crisis.The median of LIQ increased in the quarters of 2007, which is coherent with the reduction of average liquidity that year.In turn, the behavior of the standard deviation at the start of 2009 indicates there was an increase in volatility around the mean.
The average of the SIZE variable generally increased during the entire period, including that of the crisis, while its standard deviation also increased during the series due to the increased number of observations far from the mean.
There was wide fluctuation for FIN.The increase in its average in 2002 can be explained by the depreciation of the exchange rate, causing a sharp rise in the debt of companies with foreign loans.The ROA variable showed a negative average for nearly the entire period.

ANALYSIS OF THE RESULTS FOR THE PERIOD FROM 2000 TO 2007
We estimated two panel regression models, one for the general governance dummy variable and the other for each of the governance variables separately, with the other variables the same.The results are in Table 2.
To increase the robustness of the models, we carried out the following steps: i) In analyzing the residuals generated by the regression, we checked for the presence of heteroscedasticity vii and autocorrelation viii , besides whether or not the data were normally distributed; and ii) To adjust the results for problems of heteroscedasticity and autocorrelation, we used the Huber-White or Sandwich adjustment (Huber, 1967;White, 1980) to estimate the covariance matrix of the estimated parameters.According to Cameron & Trivedi (2010, p. 334), this adjustment is adequate for panel data.
The first model refers to the dummy CG which represents listing in any of the three governance segments indiscriminately.The results are shown below.According to the above results, the CG variable has a positive effect on stock liquidity of 0.1014, which is coherent with our expectation, that the shares of companies with enhanced corporate governance practices will tend to be more liquid than those of firms in the traditional trading segment.This result is also in line with the findings of Martins, Silva & Nardi (2006), Camargos & Barbosa (2006) and Carvalho (2003), of the benefits in terms of liquidity of establishing higher governance standards.
However, the ADR variable has a higher coefficient and significant at the 1% level, implying an even stronger effect on liquidity of issuing ADRs.
The second model contains the dummies L1, L2 and NM.In this case the coefficients of L1 and NM are positive and statistically significant at 1%, in contrast to coefficient of L2, which is negative and significant at 5%.Also, the effect on liquidity from listing in the Novo Mercado segment is stronger (0.1815) than that of Level 1 (0.1384).With respect to L2, the estimates indicate that companies of the Level 2 segment on average have lower liquidity of 0.1653 in comparison with companies in the traditional segment.This result runs counter to the expectation of hypothesis 4, according to which there should be a stronger positive impact of governance on liquidity for L2 than L1 firms.Nevertheless, the findings for NM and L1 are in line with hypothesis 4. The coefficient estimated for ADR is still positive and significant, demonstrating a higher impact on liquidity in relation to adherence to any of the three special governance segments.
An explanation for the differences observed between the governance and ADR dummies is that companies that issue ADRs (which can also be listed in one of the special trading segments) must meet the strict governance and other standards of the SEC.A study by KPMG (2009) applied a checklist to find differences in some corporate governance practices between companies listed in the special segments of the BM&FBovespa and those with ADRs.The result indicated that firms with ADRs satisfied more items on the checklist.
Therefore, ceteris paribus, those differences should imply changes in stock liquidity, in line with the results presented here.
In general, the result observed for CG is coherent with the findings of other studies.
Nevertheless, the main contribution of this paper is the analysis of liquidity broken down into the three governance levels.The results provide evidence that the shares of NM firms are more liquid than those of L1 companies, but those of L2 firms are less liquid than the other firms.This result explains the higher coefficient for L1 (second model) in relation to CG (first model), because L2 companies are considered in the aggregate analysis.The third model, in contrast to the first, shows no added liquidity of the shares, because the coefficient of CG is not statistically significant.
The crisis unleashed in 2008 may have influenced the relationship between the dependent variable LIQ and the independent ones CG (third model), L1 and NM (fourth model) ix .The overall decline in liquidity of the stock market occurred due to various factors, among them the outflow of foreign investors' capital.We believe this, among other factors, affected the market's liquidity.This being the case, one of the variables that was important to explain stock liquiditythe governance dummyceased being so.However, this might not be a On the other hand, the L2 variable continues having the same result as in the 2000-2007 period, albeit contrary to the expectation.In other words, the shares of companies in the Level 2 segment, in the pre-crisis period and during the crisis, are less liquid than the shares of the other firms.
The interactive variables show the effects of the governance levels during the crisis.In the case of CGxCRISIS, the estimated coefficient is negative and statistically significant at 1%.Besides this, the variables L1xCRISIS, L2xCRISIS and NMxCRISIS have negative signs, although only the last one is significant.These results indicate that enhanced corporate governance practices during the crisis period provided no liquidity premium.
However, the ADR dummy is statistically significant in all the models, with a higher coefficient in comparison with the 2000-2007 period.This indicates that the issuance of ADRs came to have a stronger impact on liquidity.A possible explanation is that against the backdrop of the higher systemic risk in the crisis period, demand increased for the shares of firms perceived as safer investments due to the higher disclosure required by the SEC's rules.
Therefore, the results for the variables L1 and NM do not allow any conclusions regarding hypotheses H1 and H3, since those variables were statistically significant in the 2000-2007 period and not in the 2000-2009 period.The likely explanation for this change in the relationship of the variables is the crisis.However, to draw more solid conclusions on the effects on liquidity of listing in the Level 1 and Novo Mercado Segments, later studies with a longer time frame will be necessary.
On the other hand, according to the results of this study, hypothesis H2 can be rejected, since the result runs counter to the expectation in periods with our without crisis.
Nevertheless, we must stress the low number of firms listed in the Level 2 segment in the period studied in relation to the total of 521 firms, as can be seen in the appendix.Since Level 2 firms follow the same rules as Level 1 companies plus others, we expected to find liquidity at least the same as that of Level 1 firms, assuming all else constant.However, other factors associated with the companies, such as sector, size, past performance, age, analyst coverage and specific events, might explain the contrary result.
Another observation that deserves note is the consistently positive effect on liquidity of issuing ADRs before and during the crisis, something that did not occur for L1 and MN firms.In other words, these variables ceased being relevant to explain stock liquidity, unlike the ADR variable.With this, assuming that ADR, L1 and NM all represent enhanced governance factors, it can be said that Brazilian companies that adapted to the rules of the SEC were perceived more favorably by investors during the crisis period than those following the governance rules of the L1 and NM segments.It is important to stress that nearly all the companies issuing ADRs in the sample belong to Level II or III of the NYSE, the most demanding in terms of SEC regulations, and also are subject to SOX.Consequently, the governance standards of these firms are naturally stricter, requiring greater rights for investors and better transparency.These aspects explain the greater liquidity premium for firms with ADRs in relation to L1 and NM firms.
An explanation for that result can be the different governance rules the companies must satisfy.For example, firms with securities traded in the American market must abide by the SOX Act, which according to Andrade & Rossetti (2006, p. 183), "established comprehensive regulations applicable to corporate life, based on good governance practices."Besides this, we can mention the role of institutional investors with influence on governance.
Another aspect to consider is the legal setting of the two countries.Brazil has a code law system while the United States follows the common law tradition.Among the various characteristics that differentiate the two systems is the generally weaker enforcement in code law counties due to institutional factors.According to La Porta, Lopez-de-Silanes & Shleifer (1998), a legal system with strong enforcement can offset the existence of weak rules, as long as the courts and administrative authorities are active in protecting investors.This aspect can explain the greater trust by investors in monitoring the application of governance rules and the degree of investor protection, helping explain the difference in the estimated coefficients for the governance and ADR dummies.
In summary, the differences in terms of governance rules and the legal environment are reflected in the liquidity premium, which favors issuers of ADRs over other firms.
Hypothesis H4 also can be rejected, since although the coefficient of NM is higher than that of L1 for the pre-crisis period, that effect evaporates for the full period.Furthermore, the coefficient of L2 is negative in the two periods.i As disclosed at the respective sites: <www.bovespa.com.br> and <www.cvm.gov.br>.

APPENDIX
ii One of the motivations to study stock liquidity is the study by Vieira, Ceretta & Fonseca (2011).According to them, "the influence of liquidity on the return of assets has been widely studied in recent years, both from the standpoint of individual stocks and market liquidity."Specifically, the work contributes to the study of liquidity from the perspective of the influence of the variation of liquidity on stock prices.
The study period covers the financial crisis of 2008.Assuming that the crisis caused structural breaks in the stock market, we first analyzed the 2000-2007 period and then the entire 2000-2009 period.This treatment is justified by the possibility that the crisis affected the liquidity of the market as a whole in2008 and 2009, which  would influence the relationship between governance and liquidity.The period considered for the financial crisis is July 2008 to May 2009.The reason for this delineation is that the main market-tracking index, the Ibovespa, started to fall sharply just before July 2008 and the began recovering in May 2009, as can be seen in the graph.Silva, Nardi, Martins, Barossi-Filho BBR, Braz.Bus.Rev. (Engl.ed., Online), Vitória, v. 11, n. 1, Art. 1, p. 1 -24, jan.-feb.2014 www.bbronline.com.brGraph 1 -Behavior of the Ibovespa, period 2008-2009.Source: Prepared by the authors.The information on the companies that also have ADRs was obtained from the site of the New York Stock Exchange (NYSE), and the dates of adhesion to one of the special trading segments were obtained from the site of the BM&FBovespa.Most of the remaining data were gathered from the Economática database, including the numbers for computing the stock liquidity proxy (LIQ) v .
from 2000 to 2007, the results indicate an increase in liquidity for the companies with enhanced governance as a group, but only for those in the Level 1 and Novo Mercado segments with when examining each segment individually.With the period extended to 2009 to include two years impacted by the financial crisis of 2008, no benefits of governance on stock liquidity were observed.Finally, companies with ADRs showed higher liquidity in relation to those listed in the three governance segments both before and during the crisis.These results contribute to the literature by allowing a comparison of the effect of corporate governance on stock liquidity in a period before and during the crisis, based on interactions between the governance levels and the crisis.

Table 3 -Regression models with panel data, estimated with fixed effects, period from 2000 to 2009.
Only future studies, after the effects of the crisis dissipate, will be able to resolve this point.