7.1 Summary Statistics and Correlations
Panel A of Table 1 shows the descriptive statistics of our five liquidity measures and their control variables. Our sample period is from January 1 2002– December 31 2002, 252 trading days. The mean of quoted half-spread (QSP) is about 2.2046 cents per share, and its range is about 3.522 cents. The mean of proportional quoted half-spread (PSP) is around 0.0716 percent with the range from 0.0281 percent to 0.3220 percent. The average effective spread (ESP) is 1.6166 cents with the range about 2.9434 cents, and is about 73 percent of the quoted half-spread. The finding that the average effective spread is less than the average proportional quoted half-spread is consistent with the argument suggested by Lin, Sanger, and Booth (1995)12. The relative effective spread (RESP) has the mean about 0.0519 percent which is about 72 percent of the proportional quoted half-spread, and its range is between 0.0218 percent and 0.2170 percent. The information asymmetry cost of the effective spread (INF) has the average value about 0.6577 cents, and its range is from 0.2172 cents to 1.3402 cents. The average closing price (CLP) for our sample is approximately $38.01, and the range is between $4.56 and $121.73. The mean of daily dollar volume (DOLVOL) is around $66.64 millions and the sample range is from $ 4.41 millions to $610.33 millions. The return volatility (RETSTD) has the average value about 0.0264, and the sample range is from 0.0138 to 0.0768. The Pearson correlation coefficients of our five liquidity measures and their control variables are shown in Panel B of Table 2. There is one thing catching our attention: the information asymmetry component, quoted half-spread, and effective spread are strongly positive correlated, implying that the higher information asymmetry cost induces higher equity spread under the fact that order processing cost is fixed.
12 Lin, Sanger, and Booth (1995) argue that demanders of immediacy services rarely received prices which were less favorable than prevailing quotes on the NYSE.
The descriptive statistics of the S&P T&D final rankings and their control variables are shown in the Panel A of Table 2. The mean of composite basis T&D final ranking (CFR) is about 7.55 with the range between 7 and 9. The average of annual basis T&D final ranking (AFR) is around 4.78 with the range from 1 to 8. Taking notice of the difference between these two rankings, the annual basis rankings have lower mean but larger range while the composite basis rankings have higher mean but smaller range. This characteristic is consistent with the argument of Pantel and Dallas (2002). They suggest that the annual basis rankings which only focus on firms’ annual reports could be viewed as firms’ voluntary disclosures. On the contrary, the composite basis rankings which include annual reports, 10-Ks, and other proxy statements might be regarded as regulatory disclosure practices. Thus, due to strict laws of investor protections and severe disclosure regulations in U.S., the firms reveal consistently higher rankings on composite basis and smaller differences between firms’ composite basis rankings than their annual basis rankings. Panel B of Table 2 presents Pearson correlation coefficients of the S&P T&D final rankings and their control variables. The firm size (SIZE) and asset-in-place (AIP) reveal positive correlations to both composite and annual basis T&D final rankings, but the positive correlation between firm size and these two rankings are insignificant. The return standard deviation (RETSTD) is significantly negative correlated with annual basis T&D final ranking, but is insignificantly negative correlated to composite basis T&D final ranking. This finding is consistent with the results of the first-stage regression. We argue that the reason of return volatility presenting more power to explain annual basis final ranking than composite basis final ranking is due to annual basis final ranking relate more closely to firm’s voluntary disclosure than do composite basis final ranking.
7.2 OLS, 3SLS and GMM estimation results of the quoted half-spread, effective spread and two S&P T&D final rankings
We first examine the relationship between the quoted half-spread and S&P T&D rankings by applying 3SLS and GMM estimation to system (20) and system (21). Table 3 presents the estimation results of the quoted half-spread (QSP) and composite basis final ranking (CFR) while Table 4 presents the estimation results of the quoted half-spread (QSP) and annual basis final ranking (AFR). The OLS estimation results are also provided for comparison in Table 3 and Table 4. The p-value of each coefficient estimate is provided in the parenthesis below it.
The composite basis final ranking (CFR) reveals a significantly negative relation to the quoted half-spread (QSP) under both 3SLS and GMM estimations of the first equation, and the results support our hypothesis that firms with better disclosure practice have relatively better market liquidity of their stocks. Comparing 3SLS and GMM results with OLS result, we find that the negative relation between composite basis final ranking and the quoted half-spread is not statistical significant in the first equation under OLS estimation. Moreover, the simultaneous estimation of the second equation shows that the quoted half-spread is not significantly related to composite basis final ranking, indicating that there is probably no simultaneity existing in the determination of the quoted half-spread and composite basis final ranking. Because there are big differences between OLS estimation and other two instrumental variables estimations without significant simultaneous problem in our model, we argue that there may be other endogenous problem due to measurement error of the composite basis T&D final ranking in assessing firm’s disclosure practice. All control variables of the quoted half-spread present significant coefficient estimates and the signs of them are consistent with our expectation. The instruments of composite basis final ranking reveal predicted signs of their coefficient estimates, and all of them are statistically significant at
common confident level.
Table 4 shows the simultaneous estimation results of the quoted half-spread (QSP) and annual basis T&D final ranking (AFR). The results are very similar to the former. The annual basis final ranking also presents a significantly negative relation to the quoted half-spread under both 3SLS and GMM estimations of the first equation, and the results support our hypothesis that firms with better disclosure practice have relatively better market liquidity of their stocks. But the negative relation between composite basis final ranking and the quoted half-spread is not significant in the first equation under OLS estimation. Again, the simultaneous estimation of the second equation shows that the quoted half-spread is not significantly related to annual basis final ranking, implying that there might be no simultaneity existing in the determination of the quoted half-spread and annual basis final ranking. All control variables of the quoted half-spread except return standard deviation (RETSTD) present significant coefficient estimates and the signs of them are consistent with our expectation. The instruments of annual basis final ranking reveal predicted signs of their coefficient estimates, and all of them are statistically significant at common confident level.
Table 5 and Table 6 report the estimation results of the effective spread and two S&P T&D final rankings. The empirical results are similar to results of the quoted half-spread and two S&P T&D final rankings. Both composite and annual basis T&D rankings reveal significantly negative relations to the effective spread (ESP) under 3SLS and GMM estimations of the first equation, and the results support our hypothesis that firms with higher T&D rankings have relatively lower effective spreads of their stocks. We also find that the negative relation between two final rankings and the effective half-spread are not statistically significant in the first equation under OLS estimation. Moreover, the simultaneous estimation of the second equation shows that the effective spread is insignificanly related to two final rankings, indicating that there is probably no simultaneity existing in the determination of the
effective spread and disclosure practice. All control variables of the effective spread present significant coefficient estimates and the signs of them are consistent with our expectation. The instruments of two final rankings reveal predicted signs of their coefficient estimates, and all of them are statistically significant at common confident level.
7.3 OLS, 3SLS and GMM estimation results of the proportional quoted half-spread, relative effective spread and two S&P T&D final rankings
Table 7 to Table 10 report the OLS, 3SLS and GMM estimation results of the proportional quoted half-spread (PSP), relative effective spread (RESP) and two S&P T&D final rankings. The differences between these estimation results and the former estimation results are that the estimated coefficients of composite and annual basis rankings in the first equation under 3SLS and GMM estimations do not reveal significantly negative relations to both proportional quoted half-spread and relative effective spread. There are several possible reasons for these differences. First, previous studies consider the effect of the institutional feature that the spreads are quoted in eighths or sixteenths, and thus use the proportional spread measures to catch the fact that low priced stocks have higher relative spreads in practice. But the studying period of this study is the year 2002, when the spreads are no longer quoted in eighths or sixteenths. Second, we argue that because the quoted half-spread and the effective spread represent the dollar value of the actual cost and revenue obtained by liquidity suppliers, and the S&P T&D ranking is related to information asymmetric risk of the stock, the dollar measures of liquidity should be more significantly correlated with S&P T&D ranking than the proportional measures of liquidity. In addition, the quoted half-spreads and the effective spreads are not only very smaller than quoted price, but their ranges are also narrower than the range of quoted price. Therefore, when the quoted half-spreads and the effective spreads are divided by quote midpoint, the stock price will dominate these two proportional measures and cause our insignificant results of the negative effect of S&P T&D
ranking on these two proportional measures.
7.4 OLS, 3SLS and GMM estimation results of the information asymmetry component and two S&P T&D final rankings
The information asymmetry component of the effective spread represents the information asymmetry cost faced by market liquidity suppliers when trading with informed traders and therefore reflects market perception of the firm’s information asymmetry risk.
Further, the S&P T&D ranking measures the extent of firm’s corporate governance and is predicted to be directly related to firm’s information asymmetry risk. Therefore, in this section, we examine the relationship between the dollar value of information asymmetry component and S&P T&D ranking by applying 3SLS and GMM estimation to show whether better corporate governance is associated with better equity liquidity.
Table 11 shows the simultaneous estimation results of the dollar value of information asymmetry component (INF) and composite basis T&D final ranking (CFR), and Table 12 presents the results of the dollar value of information asymmetry component (INF) and annual basis T&D final ranking (AFR). We find that both composite and annual basis T&D final rankings have significantly negative relation to information asymmetry component in the first equation under 3SLS and GMM estimations, but the negative partial relation is not statistically significant under OLS estimation. Furthermore, information asymmetry component does not reveal significant relation to both composite and annual basis T&D final rankings in the second equation under all three estimation methods, indicating that there might be no simultaneity existed in the determination of information asymmetry component and these two S&P T&D final rankings. Again, the significantly negative relationships between information asymmetry component and two S&P T&D final rankings in the first equation under 3SLS and GMM estimation support our hypothesis that firms with better
disclosure practice have better corporate governance and thus have lower information asymmetry components of their stocks. We also argue that the two S&P T&D final rankings may have measurement error in measuring the extent of firm’s disclosure practice and information asymmetry, and thus cause the inconsistent estimation result in OLS estimation in the first equation of the simultaneous system.