Examining the Effect of Stock liquidity on the Relationship between Stock Split and Stock Market Performance

The effect of stock liquidity on the relation between stock split and stock market performance is puzzling. This paper examines the factors that affect the relationship between stock split and stock market performance. The data are gathered from the Egyptian Stock Exchange listed companies, based on their market capitalization from all sectors in Egypt during 2010 to 2020. This event study employs multiple regression analysis. Liquidity is measured by volume and number of transactions. Announcement date is considered for stock split as independent variable. Also, firm size, split factor, Industry type as control variables have been tested in the model. The research is event study; The time window is based on twenty days and five days. Results indicate that liquidity as moderator is positively affect the relationship between stock split and stock market performance for five- and twenty-days’ time windows. A robustness check has been performed for every regression model. Showing significant effect of liquidity as moderator, measured by volume of transactions, on the relationship for the sample period between 2010 and 2019 for twenty days’ time window. Results support the easiness and enhancing the process of stock split. For example, Financial Regulator Authority could waive its approval for stock split to companies’ general assembly as the market would correct itself for disturbance in liquidity after stock split announcement. Also, Minimizing the number of companies that don't execute stock split affect the investors’ behavior which affects the relationship of stock split announcement and stock market performance.


Introduction
Theoretically, Increasing the numbers of shares that are outstanding is a decision of company's management by issuing more shares to current shareholders. Fundamentally, companies undertake stock splits when the company is confident that the momentum of the company's earnings will continue to push the stock prices upward. Therefore, companies that are interested in lowering the trading cost will lower their cost of capital because of liquidity. In which stock splits will attract small investors and increase the liquidity of the stocks and automatically the cost of capital will be reduced (Company's beta). According to Copeland (1979), stock liquidity is less than before the stock split announcement date. Hence, this finding contradicts the street lore that stock split increase liquidity with no plausible explanation. relationship between stock split and stock market performance in the EGX? The research question could be addressed as follows: Does liquidity affect the relationship between stock split and stock market performance?
Following this introduction, the literature review will be discussed in section 2, though variables and developing hypotheses are shown in section 3. Section 4 illuminates the descriptive and diagnostic statistics, hypotheses' testing in section 5 while concluded remarks are referred in section 6.

Literature Review
The concern of many scholars is with the effect of liquidity on the relationship between stock split and stock market performance. Copeland (1979) studied the liquidity change after stock split announcement arguing that stocks that are traded with the optimal price range should be more liquid. Also, Copeland (1979) shed a light that there are other factors that affect liquidity other than the number of shareholders for example, the psychological aspect in which every trader sees the opportunity. However, the factors which were considered in his study was the share volume trading and the changes in transactions costs as a percentage of value trade. Copeland (1979) found that stock liquidity is less than before the stock split announcement date. Hence, this finding contradicts the street lore that stock split increase liquidity. There was no plausible explanation for the finding; however, the only explanation that companies' performance maybe extraordinary before the split. Dennis (1998) studied the effect of stock split on stock liquidity after announcement date, arguing that the some of the past research assumed increase in liquidity after stock split announcement date. They found no noteworthy change in trading volume after stock split announcement date. Which contradicts Desai et al. (1998) study which examined the effect of stock split on liquidity, or the bid-ask spread taking those two triggers for enhancement in trading which are liquidity and information. Desal et al. (1998) found that the bid-ask spread increases directly after stock split which means higher liquidity than before stock split in addition to number of trades.
In consideration that the signaling theory cannot affect index. Hence, Dennis (2003) studied stocks liquidity of a tracking index (NASDAQ 100) to separate the signaling aspect from the liquidity aspect. Accordingly, the Index will be affected only by the liquidity aspect. He argued that stock split can lead to liquidity for either of the two reasons; the first reason is the lower price can encourage small investors to trade in the split stock, while the second reason is that stock liquidity may increase/decrease as a result of signaling aspect. Dennis (2003) found that the arithmetic average daily turnover was 23.95% prior to the split and 22.8% following the split. On the other side, it has been found that the number of trades following the split is below twice prior to the split. Along with share volume, the average is nearly twice before the split. In other words, the number of small investors increase after stock split while the turnover was not affected but when classifying trading volume and share volume to trade size, the liquidity appears to be increased to small investors only.
Also, Dennis and Strickland (2003) examined the relationship between ownership structure and abnormal return at announcement date in addition to the alteration in liquidity after stock split. Their concern wasn't the reason behind the management decision regarding stock split rather than the effect of stock split on stock volume and abnormal return. The data was gathered from NYSE, AMEX, and NASDAQ between Jan. 1990and Dec. 1993. Nevertheless, Dennis and Strickland (2003 studied a time series analysis of liquidity for their sample. Additionally, a turnover as a measure of liquidity were used which is the monthly volume divided by the number of outstanding shares. They argued that turnover measure equalizes the measurement between different companies while considering their size. Dennis and Strickland (2003) found that institutional ownership share in the ownership structure increase post stock split and this maybe a reason for the increase in liquidity after split. Additionally, the level of liquidity increase is dependent on the percentage of institutional ownership structure.
Recently, Goyenko et al. (2006) studied the extent in which stock split improve liquidity on longer term time windows by shading a light that past research studied shorter term windows. They measured liquidity by the average cost of trade which is the percentage of spread. Also, they studied the stock performance after the split for 6 years before the split and the first year after the split. They used the matched sample methodology to measure liquidity, in which they search for a matching stock and study liquidity of the two stocks. Additionally, they used Famaa and MacBeth (1973) regression approach which relates the stock liquidity to different factors and control variables. Goyenko et al. (2006) found that the deterioration in liquidity is not permanent and it may continue up to the first year only while they found gains in liquidity based on longer terms.

Measuring Variables and Developing Hypotheses
The independent variables have been measured by "stock split announcement date", on fiveand twenty-days' time windows. while the dependent variable has been measured by log of sector index of EGX. According to previous analysis, the big majority of past research tested stock split on stock performance, while taking different ijef.ccsenet.org International Journal of Economics and Finance Vol. 14, No.5; 2022 proxies of stock performance. While The reason for considering sector index instead of normal Egyptian stock Indices, such as: EGX30, 70, and 100, is that there are stock split cases that are OTC and because there is no index for OTC, log of sector index has been used to indicate the market performance. Moderator is measured by volume and number of transactions. Other control variables are firm size, split factor, Industry type. stock markets of Egypt are the source of data on daily basis during the period from Jan 1, 2010, to July 24, 2020 noting that two companies listed in EGX30 was excluded due to data unavailability. Table 1 shows the research variables, as follows: The number of stock indices and stock splits used to estimate the variables and moderator are shown in Table 2 as follows: This paper seeks at testing the importance of the liquidity on the relationship between stock split and stock market performance in the EGX. This has been accompanied by testing the following hypothesis: -There is no significant impact of liquidity on the relationship between stock split and stock market performance.
This means that null hypothesis Ha: β = 0 and the alternative hypothesis Hb: β # 0, where β is the regression coefficient as the subsequent function: LOGSector i,t = α + β 1 AnnouncementDatei,t + β2 LiqiuidityNumber + β3 Liquidity Volume + β4 AnnouncementDatei,t * Liquidity Number + β5 AnnouncementDatei,t * Liquiidty Volume We have tested the impact of each moderator and independent variable in a multiple regression model, as shown in table 3.

Descriptive and Diagnostic Statistics
The whole research period and for the sub-periods are shown in table 3 of the correlation between the independent variables, as follows: ijef.ccsenet.org International Journal of Economics and Finance Vol. 14, No.5; 2022 Table 3 shows that correlation coefficients between of liquidity measured by volume which is negative and moderator liquidity measured by number of transactions which is positive for five days' time window. For twenty days' time window, correlation coefficients between of liquidity measured by volume and number of transactions is positive. While Tables 4 explains the descriptive statistics variables for the research variables, as  follows:   Table 4. Descriptive statistics Source outputs of data processing using Stata version-17.
Since the value of "F-Test" for five days' time window is 62.401 with significance level at zero for five days' time windows and twenty days "F-Test" 242.098 with significance level at zero too; then the researcher concludes that the moderator and stock split announcement date affect the stock market performance. It would be useful in determining the value of the dependent variable coefficient and moderators' coefficient in the regression model.
In order to ensure the non-existence of relationship between the dependent variable, VIF (Variance Inflation Factor) was needed to measure the multicollinearity. The larger the variance inflation factors, the more severe the multicollinearity. Some statisticians argue that if any variance inflation factor exceeds 10, then multicollinearity is a problem. Other statisticians argue this value to be liberal and suggest that the variance inflation factors should not exceed 4 or 5. The value of VIF is less than 10, then the model is not suffering from multicollinearity problem for five days' time windows. On the other hand, for twenty days' time windows, the value of VIF is less than 10, then the model is not suffering from multicollinearity problem too.

Testing Hypotheses
This paper aims to test the significance of the effect of liquidity on the relationship between stock split announcement date and stock market performance in EGX. Using multiple regression model in this event study for two-time windows which is five and twenty days, Results indicated that liquidity as independent variable and moderator accepted in the model and explain 74.5% at α = 0.05 significance for five days' time window while 74.1% at α = 0.05 significance for twenty days' time window. This means that the moderator affects the relation between the independent variable and dependent variable and explain the change in the dependent variable by 74.1% for twenty days and 74.5% for five days' time window.
In five days', time window heteroskedasticity Chi is 10.51 and its prob is 0.0012 while Durbin Winston is 0.339. While In twenty days' time window heteroskedasticity Chi is 39.19 and its prob is zero While Durbin Winston is 0.103. since a value of 2.0 means that there is no autocorrelation detected in the sample. Values from 0 to less than 2 indicate positive autocorrelation and values from 2 to 4 indicate negative autocorrelation. Hence, auto autocorrelation is positive and heteroskedasticity is out of range that's why FGLS regression model was also applied.
Looking at the sample from a more holistic approach, since the value of "F-Test" for five days' time window is 62.4 at significance level zero and 242.09 for twenty days' time window at significance level zero too. then it can be concluded that the moderator and stock split announcement date affect the stock market performance. It would be useful in determining the value of the dependent variable coefficient, moderators, and control variables' coefficient in the regression model. Since the significance value of the test statistic lower than 0.05, then the null hypothesis would be rejected. After robustness check has been conducted and splitting the research period into two-time windows two regression models have been tested as shown from tables from as follows:  Source outputs of data processing using Stata version-17.
As shown in Table 5, that liquidity measured by calumet of transaction is significant as moderator in addition to all the control variables in the relationship between stock split announcement date and stock market performance measured by sector index. After a robustness check has been conducted for each time window all through the entire period, showing significant impact of liquidity on the relationship between stock split and stock market performance in the EGX. showing significant effect in the regression model significant effect of liquidity on the relationship between stock split and stock market performance. So, for the hypothesis, we can reject the null hypothesis and accept the alternative one.

Summary and Concluded Remarks
This paper aims to develop a model, using regression model through OLS method to examine liquidity effect the relationship between stock split and stock market performance. Two time-windows were used to examine short term and long-term impact on stock market performance, where it is measured by proxy, which is sector index. While announcement date is a dummy variable for this event study. Liquidity as moderator, is measured by volume and number of transactions. This has been applied on EGX on daily basis throughout the period from Jan 1, 2010, to July 24, 2020.
Liquidity measured by volume of transactions is significant in twenty days' time windows while number of transactions is insignificant. This indicates that stock split affects stock market performance by the volume of transactions as a result that current investors increase their investments in stock split share after announcement date in twenty days' time window which is consistent with (Dennis, 2003). While results for five days' time window shows that liquidity is insignificant in the relationship between stock split and stock market performance which indicates that investors don't respond to stock split announcement in five days' time window which in consistent with (Copeland, 1979). On the other hand, control variables were significant in this relationship.
Results indicate that liquidity is insignificant on stock market performance around the announcement date. There are three explanations: first, the explanations maybe that the stock price still too high for small investors. Second, explanations could be that investors experienced same companies that announce stock split but don't execute as a trial for stock price increase. Third, market response around announcement date maybe affected by investor's relations manager inefficiency to market well for investors about the reasons behind stock split.
There are other factors that may affect the relationship between stock split and stock market performance other than liquidity. for example, the psychological aspect in which every trader sees the opportunity. However, Future researchers may consider the following points First, Exclude the companies that didn't execute stock split. Second, exclude companies that performed stock split execution during your time window. Third, the number of days between stock split announcement date and execution date can be used as control variable. Fourth, Future research may consider another control variable which is the weight of the stock in its index in a time around the stock split announcement as for Asquith et al. (1989). Fifth, future researchers could also consider carrying out a similar study in a different stock exchange to assess any variation in responses.