Signals of market and firm characteristics and asymmetric information

In addition, this study also confirms a high level of growth opportunity and policy on widening price limit range have the positive and statistically significant effects on private infor

Signals of Market and Firm Characteristics and Asymmetric Information

Phan Bui Gia Thuy

Nguyen Tat Thanh University, Vietnam

Nguyen Tran Phuc Ngo Vi Trong

Banking University HCMC, Vietnam

Abstract

This study aims to determine the effects of market and firm characteristics on asymmetric information This study finds that signals of the market and firm characteristics are likely to significantly explain to proxies for asymmetric information measured by two econometric model, trade-indicator model and serial covariance model Analyzing 174 firms listed on HOSE from 2010 to 2016 with 1102 observations, this study finds that liquidity of stock and debt financing inversely and significantly impact on adverse selection component Interestedly, unlike the extant literature, a negative relationship between volatility and adverse selection problem is found In addition, this study also confirms a high level of growth opportunity and policy on widening price limit range have the positive and statistically significant effects on private information

Keywords: Asymmetric information, adverse selection component, liquidity, volatility, price limit

JEL Classifications: D82; G18

1 Introduction

Asymmetric information is a form of market failure, causing shocks to interest rates and banking crises (Mishkin, 1990) It is also the cause of the financial crisis in the US from 2007 to 2009, which has frozen the economy for years not only in the United States but also in other countries (Ashcraft and Schuermann, 2008) For the stock market, information asymmetry refers to informed investors have superior information related to firm business activities while uninformed investors do not (Chae, 2005) Informed investors use this particular information to gain on the losses of uninformed investors These losses are adverse selection costs that uninformed investors have to burden

The results of factors affecting information asymmetry have not reached a common consensus in some studies In particular, some studies find a positive relationship between growth opportunities and asymmetric information (Hegde and McDermott, 2004; Fosu et al., 2016), while Van Ness et al (2001) find no this relationship In addition, the debt ratio negatively effects on asymmetric information (Elbadry et al., 2015) while this effect is unclear (Cai et al., 2006) Furthermore, policy on widening the price limit range is likely to make the price move quickly to the equilibrium point (Kim and Rhee, 1997) and attract more investors joining

in the stock market (Anshuman and Subrahmanyam, 1999), but increase information risk (Berkman and Lee, 2002) Moreover, price limit range is not an effective tool to limit asymmetric information (Chan et al., 2005; Kim and Yang, 2008)

In Vietnam, the research mentioned above is highly rare Therefore, this study aims to identify determinants of asymmetric information between informed and uninformed investors when performing stock

trading This study is motivated by the empirical work of Van Ness et al (2001), Hegde and McDermott (2004), Kim and Yang (2008), Narayan et al (2015) and Fosu et al (2016) who explore the effects of asymmetric information among different investors on the specific stock market The results of this study show that there is a negative and statistically significant effect of liquidity of stock, volatility and debt financing on adverse selection component, while growth opportunity and policy on adjusting price limit range positively impact on information risk Our results are very useful for policymakers to consider whether to adjust policy on price limit range and for stakeholders to predict information risk of stock trading

The remainder of the paper is organized as follows In Section 2, this study discusses the extant literature

on asymmetric information and factors of market and firm characteristics related to information risk Section

3 presents the data sample and discusses the methods used in our empirical estimator Section 4 presents the empirical results, while Section 5 discusses these results Section 6 emphasizes our main findings and concludes the paper

2 Literature review

2.1 Asymmetric information

Asymmetric information reflects an object or group of objects that own superior firm-specific information related to future public disclosures not available to uninformed (Chae, 2005), and it arises from private information between informed and liquidity investors (Barakat et al., 2014) Informed traders make a profit from performing securities transactions on private information that uninformed ones do not, which is attributed to adverse selection problem (Copeland and Galai, 1983; Glosten and Milgrom, 1985) When the stock market has a severe adverse selection problem between different investors, uninformed investors could leave the market

According to signaling theory and market microstructure theory, specific signals of market and firm characteristics could predict the level of asymmetric information in several ways For example, frequent trading of stock and financial structure would negatively impact on adverse selection problem (Acker et al., 2002; Degryse and Jong, 2006), while volatile stock price and growth opportunity are likely to positively effect

on adverse selection risk (Chung et al., 2010; Fosu et al., 2016) However, many empirical studies have shown the mixed results, consistent with the literature and inconsistent, because these results depend on the characteristics of each country, the period of study and the research methodology

2.2 Determinants of asymmetric information

Based on the signaling theory, market microstructure theory and empirical research involved, the factors

of liquidity, volatility, growth opportunity, debt financing and price limit affecting asymmetric information are reviewed below

2.2.1 Liquidity of stock and asymmetric information

Trading volume is likely to serve as a proxy for liquidity of a stock during a particular trading period Shares with large and frequent trading quantities are considered more liquid; otherwise, they are considered less liquid Chae (2005) has shown that there is an amount of significant asymmetric information between informed investors and uniformed before the period of earning disclosure Uninformed investors often choose

to limit their trading activities except for the urgent need for liquidity

Acker et al (2002) show that stocks with high trading volume and frequency would have lower adverse selection component than those with less liquid Draper and Paudyal (2008) also found that stock liquidity was negatively correlated with asymmetric information Therefore, the hypothesis is as follows:

H1: Trading volume has a negative effect on asymmetric information

2.2.2 Volatility of stock price and asymmetric information

Volatility of stock price usually increases during the upward or downward market trend and is considered

a risk factor when trading The more private information reflects the stock price, the larger its volatility would become (Bhushan, 1989; Moyer et al., 1989) According to Wang (1993), there are different levels of stock price volatility because of private information held by investors, and the higher changing in stock price, the greater profit informed traders could make Therefore, the volatility of stock price positively relates to asymmetric information Supporting these perspectives, Chung et al (2010) and Barakat et al (2014) find that there is a positive impact of volatility on information risk

On the other hand, the positive effect of volatility on asymmetric information is inconsistent with other empirical studies For instance, Li and Wu (2006) do find no explanation of volatility leading to adverse selection component The authors suppose that volatility of stock price includes a noise signal that dissociates from asymmetric information Moreover, Chordia et al (2001) and Narayan et al (2015) find a negative effect

of volatility on spread

It can be seen that there are still mixed results suggesting that volatility could explain private information Many studies, however, have confirmed that price volatility positively effects on information problem This discussion leads to the following hypothesis:

H2: There is a positive relationship between volatility of stock price and asymmetric information

2.2.3 Growth opportunity and asymmetric information

Companies with high growth opportunities would have a high level of asymmetric information (Myers and Majluf, 1984) There are two approaches to explain this phenomenon, including the information approach and the behavioral finance Drawing from the information approach, in companies with high growth opportunities, inside managers have private information about new investment projects or cash flows from assets in place while outside investors could not afford to observe the behavior of the manager (Smith and Watts, 1992) In addition, drawing from the behavioral finance approach, shareholders who invest in companies with high growth opportunities are often overconfident and tend to overreact to vague or unverifiable information (Daniel and Titman, 2006)

Consistent with these perspectives, Hegde and McDermott (2004), Fosu et al (2016) find that the companies with a high growth opportunity positively relate to the asymmetric information Therefore, according to the information approach, behavioral finance approach and related empirical research, the hypothesis of the relationship between growth opportunities and asymmetric information is as follows:

H3: There is a positive relationship between growth opportunity and asymmetric information

2.2.4 Debt and asymmetric information

Studying the relationship between debt ratio and asymmetric information opens up different perspectives According to Stulz (1990), many companies with poor performance often use debt financing to offset their operating cash flow Moreover, the excessive use of debt could incur financial costs for the company The debt ratio, hence, is likely to diminish company performance rather than asymmetric information For this argument, Hegde and McDermott (2004), Cai et al (2006) find no relationship between debt ratio and asymmetric information

However, agency theory, signaling theory and pecking order theory underline a positive outlook of debt financing Debt financing conveys a positive signal to shareholders and creditors about the effectiveness of monitoring the behavior of management (Jensen and Meckling, 1976), improving transparent disclosure (Ross, 1977; Jensen, 1986) and declining the managerial discretion and private information (Degryse and Jong, 2006)

It also positively signals to investors about a future value-for-money perspective (Myers and Majluf, 1984) In addition, debt financing is useful for firms to enhance their performance by exploiting tax shield effectively This discussion leads to the following hypothesis:

H4: There is a negative relationship between debt and asymmetric information

2.3.5 Price limit and asymmetric information

Price limit set the maximum permitted price variation around a base price Changing the price limit range will have different impacts on the stock market Specifically, narrowing price limit range may reduce the volatility of stock price (Chen, 1993, Lee and Kim, 1995), assist the index of stock market not fall deeper during the crisis (Rhee and Chang, 1993) and restrict price manipulation in countries with high level of corruption and low-quality public enforcement (Kim et al., 2010) However, the disadvantage of the price limit is that the stock price movement slows to equilibrium (Kim and Rhee, 1997) Extending the price limit range is useful for attracting more investors, but its negative aspect causes adverse selection cost (Anshuman and Subrahmanyam, 1999) Lee and Chou (2004) study intraday price limit on the TSE and find that the firms whose stock price hits a ceiling of price limit would have a higher level of asymmetric information than the firms whose stock price fluctuates within the limit

The price limits have been adjusted many times in Vietnam’s stock market In the period 2010-2016, the price limit range of the firms listed on HOSE was adjusted an increase from 5% to 7% since January 15, 2013 Widening the price limit range could increase the volatility of stock price and decline trading volume (Berkman and Lee, 2002) This discussion leads to the following hypothesis:

H5: The firms after the enactment of the legislation increasing the price limit range would have the higher

level of asymmetric information than those before this enactment

3 Data and methodology

3.1 Data collection

This study collected statistical data of trading prices and orders of companies listed on HOSE during the first quarter from January 1 to March 31 in the period 2010-2016 to measure asymmetric information, stock liquidity, volatility and growth opportunity The first quarter is the time when the listed firms disclose information about annual reports and audited financial statements at the end of year related to firm performance, and there is a significant information risk between insider and outsider or between informed investors and uninformed investors In addition, the sample does not include banks, financial institutions, insurance companies and investment funds because of specific activities as well as specific legal regulations for these organizations The reason for choosing this period is that VNIndex has a negligible fluctuation Figure

1 below reveals this index which is likely to present the market volatility

Figure 1 below illustrates VN-Index in the period 2007-2016 Obviously, VN-Index declined dramatically, from 943 points in December 2007 to 261 points in March 2009 At the beginning of April 2009, the Vietnam Prime Minister issued Decision No 443/QĐ-TTg on giving interest rate with an interest rate of 4%/year, so the VN-Index rose again from 263 points in March 2009 to 589 points in October 2009 However, after this period, October 2009 until March 2016, VN-Index had a tendency to move sideways within the resistance range of 400-600 points This evidence shows that although Vietnam has overcome the global financial crisis, the Vietnam’s stock market is still in a long period waiting for a signal of real prosperity

Fig.1 VNIndex during a period from December 2007 to December 2016

3.2 Measuring asymmetric information

This study measures the adverse selection component for each stock as a proxy for asymmetric information

To do so, this study uses the model of George, Kaul and Nimalendran (1991) (hereafter GKN) and Kim and Ogden (1996) (hereafter KO) to accommodate transactions data These two models are discussed briefly below

3.2.1 George, Kaul and Nimalendran (1991)

Trade-indicator GKN model assumes the transaction price and true price of the stock is determined by the following equation:

Pit = Mit + πi (Sqi/2)Qit (1)

Where Pt is the transaction price; Mt is the true price; Qt is the trade indicator variable; Sq is bid ask spread;

π is the proportion of the order processing component in spead, and (1– π) is the proportion of the adverse selection component Take the differential Equation (1) given by the new equation as follows:

∆Pit = ∆Mit + πi (Sqi/2)∆Qit (2)

Let RDTM,it = ∆Pit – ∆Mit denote the difference between the change in the transaction price and the change

in the bid price, Equation (2) becomes:

RDTM,it = πi (Sqi/2)[Qit – Qit–1] (3)

Equation (3) can be written as a regression equation as follows:

2RDTM,it = a0 + a1 (Sqi)[Qit – Qit–1] + εit (4)

The GKN model uses the regression Equation (4) to estimate the coefficient a1 = π as the order processing

cost component Therefore, the average adverse selection component of the stocks is calculated as 1 – a1

Next, let x it = (S qi )[Q it – Qit–1] and y it = 2RD TM,it correspond to each stock i, the average adverse selection component of the stock i, ASC i ,GKN is estimated according to the formula below:

1 1,

,

2 1

T

t i

it t









0

200

400

600

800

1000

VNIndex

This study measures the variables in regression equation (4) as follows: RD TM,it = ∆P it – ∆M it is the difference between the change in the closing price at the end of the day (∆P it) and the change in the mean of the bid price and ask price or change in the midpoint (∆M it ); Q it is a trading indicator variable determined by Lee and

Ready (1991), Q it = +1 if the closing price is higher than the midpoint; otherwise, Q it = –1; Sqi is the difference between ceiling price and floor price

3.2.2 Kim and Ogden (1996)

KO model adjusts and modifies the GKN model under serial covariance Accordingly, the regression equation estimated by KO model has the following form:

S i KO = β0 + β1√𝑆̅𝑞𝑖2 + ε i (6)

Where S i KO = 2  Cov RD ( TM it, , RDTM it, 1)

is the spread in the KO model, with RD TM,it = ∆P it – ∆M it is the difference between the change in transaction price (∆P it) and the change in the midpoint (∆M it); S̅ qi 2 = 2

1

1 T qit t

S

is the mean of the sum of the squared spreads, where S qit is the spread changing over time; β1 is the regression coefficient as a proxy for the order processing cost component, π

The regression Equation (6) is used by the KO model to estimate the coefficient β1 = π that is the order

processing cost component Therefore, the average adverse selection component of the stocks is calculated as

1 – β1

Next, Kim and Ogden (1996) proposed a convenient formula for estimating asymptotic average adverse

selection for each stock in the KO model Accordingly, ASC i,KO is estimated by the following formula:

,

2 1

1

1

qit t

ASC

S T







 

(7)

This study measure the variables in the regression Equation (7) as follows: S i KO = 2



where RD TM,it = ∆P it – ∆M it is the difference between the change in the closing price at the end of the day (∆P it) and the change in the midpoint (∆M it ); S qi is the difference between ceiling price and floor price

3.3 Econometric model

Based on the studies involving the factors affecting information asymmetry on the stock market according

to theory (Bagehot, 1971; Copeland and Galai, 1983; Glosten and Milgrom, 1985) and according to empirical research (Van Ness et al., 2001; Acker et al., 2002; Hegde and McDermott, 2004; Draper and Paudyal, 2008; Fosu et al., 2016), this study formulates the following regression equation:

ControlVar

Equation (8) describes explanatory variables that are likely to affect the asymmetric information corresponding to the expected mark and the control variables that improve the effectiveness of the regression model The left side of the Equation (8) is a dependent variable which serves as a proxy for the asymmetric

information measured by ASC GKN and ASC KO The right hand side of the Equation consists of the following

explanatory variables: Liquidity are the factors of the stock liquidity including frequent trading (Turover) and liquidity trading (Depth); Volatility is the volatile stock price; Growth are the factors of firm growth including growth opportunity (TobinQ) and level of growth opportunity (Opp); Debt are the factors of debt financing

including total of debt ratio (DebtRatio) and bank loan ratio (BankRatio); and Policy is the policy on adjusting price limit range from 5% to 7% and ControlVar are the control variables including the firm size (Asset) and the number of years since listing (ListYear) Measurement of research variables is detailed in Table 1 below

Table 1 Definition and measurement of variables

ASC Adverse selection component ASC GKN and ASC KO estimated from trade-indicator

GKN model and KO model

shares

ask prices divided by number of outstanding shares

by total assets

Opp Level of growth opportunity Opp = 1 if TobinQ > 1, high growth opportunity

Opp = 0 if TobinQ < 1, low growth opportunity

loan divided by total assets

Policy Government’s policy on adjusting

price limit range from 5% to 7%

Policy = 1 if years of study belong to the period

2013-2016 (the price limit range is 7%)

Policy = 0 if years of study belong to the period

2010-2012 (the price limit range is 5%)

ListYear Number of years since listing Natural logarithm of number of years since listing

4 Results

4.1 Characteristics of research sample

Table 2 presents the mean values of ASC GKN in Panel A and ASC KO in Panel B under 0 < ASC < 1 from 2010

to 2016 ASC GKN is in the range of (52.4%; 73.3%) while ASC KO is in the range of (50.1%, 68.7%) Generally,

ASC GKN and ASC KO have the same trend

Table 2 The average adverse selection component of individual firm over the years

Panel A ASC using the model of George, Kaul and Nimalendran (1991)

Panel B ASC using the model of Kim and Ogden (1996)

Note: ASC GKN and ASC KO are referred to trade-indicator GKN model and KO model

To fit sample size between two variables of ASC and variables of market and firm characteristics and arm

to estimate the regression equation, the final research sample consists of 174 firms with a total of 1102 observations for the period 2010-2016 Table 3 below presents the statistics of the study variables

Table 3 Descriptive statistics of variables

ASCGKN and ASCKO is estimated from trade-indicator GKN model and KO model; Turnover: the average

of traded shares divided by outstanding shares; Depth: the average of shares available at both the best bid and ask prices divided by number of outstanding shares; Volatility: standard deviation of the midpoint; TobinQ: the sum of market value of stock and total debt divided by total assets; DebtRatio: ratio of total debt to total assets; BankRatio: the sum of short-term bank loan and long-term bank loan divided by total assets; ListYear: number of years since listing; Asset: total assets (mil vnd)

Next, the mean value of ASC GKN and ASC KO for different levels of growth opportunity from 2010 to 2016 illustrates in Figure 2 below

Fig 2 Adverse selection component at high and low growth opportunity during 2010-2016

Figure 2 shows that ASC GKN and ASC KO at a high level of growth opportunity (Opp = 1) are almost greater than those at a low level of growth opportunity (Opp = 0) This statistical data shows that the higher level of

growth opportunity that company has the more severe asymmetric information is

Another important factor which is likely to affect asymmetric information is price limit Widening price limit range from 5% to 7% according to Regulation No 01/2013/QĐ-SGDHCM in Vietnam may affect the

private information Table 4 below presents the mean values of ASC GKN , ASC KO , Turnover and Depth between the period 2013-2016 with price limit range of 7% (Policy = 1) and the period 2010-2012 with price limit range

of 5% (Policy = 0)

0.0% 20.0% 40.0% 60.0% 80.0%

2010

2011

2012

2013

2014

2015

2016

GKN (Opp = 1) GKN (Opp = 0)

0.0% 20.0% 40.0% 60.0% 80.0%

2010 2011 2012 2013 2014 2015 2016

KO (Opp = 1) KO (Opp = 0)

Table 4 Factors of study under different price limit range

Satterthwaite-Welch t-test -18.87 *** -13.40 *** 1.15 4.72 ***

Note: *** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level

ASC GKN and ASC KO is estimated from trade-indicator GKN model and KO model; Turnover: the average of traded shares divided by outstanding shares; Depth: the average of shares available at both the best bid and ask prices divided by number of outstanding shares; Policy: policy on adjusting price limit range, Policy = 1 if years of study belong to the period 2013-2016 (price limit range of 7%), and Policy = 0 if years of study belong

to the period 2010-2012 (price limit range of 5%)

The Satterthwaite-Welch t-test statistic in Table 4 shows that the mean values of ASC GKN and ASC KO under

Policy = 0 are smaller than those under Policy = 1 In addition, the mean value of Depth under Policy = 0 is greater than that under Policy = 1 (t = 4.72, p = 0.00) while no significant changing in the mean value of Turnover between two periods is found (t = 1.15, p = 0.25)

These results show that the policy adjusting the price fluctuation range from 5% in the period 2010-2012 to 7% in the period 2013-2016 is likely to cause no changing in frequency trading noticeably, reducing the volume

of stocks at the best bid and ask prices and increasing adverse selection component significantly In other words, this adjustment policy could not only reduce the liquidity of stock but also increase asymmetric information

4.2 Regression results

Before estimating Equation (8), the study estimates the pairwise correlation coefficients between the determinants and asymmetric information The matrix of correlations between the studied variables is shown

in Table 5 below

Table 5 Correlation matrix

(6) DebtRatio -0.12 *** -0.08 ** 0.07 ** -0.09 *** -0.23 *** 1 1.11

(7) Asset 0.18 *** 0.18 *** -0.05 0.14 *** 0.20 *** 0.12 *** 1 1.10

(8) ListYear 0.29 *** 0.20 *** -0.05 * -0.06 * -0.07 ** -0.08 *** 0.09 *** 1 1.03

Note: *** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level

ASC GKN and ASC KO is estimated from trade-indicator GKN model and KO model; Turnover: the average of traded shares divided by outstanding shares; Volatility: standard deviation of the midpoint; TobinQ: the sum of market value of stock and total debt divided by total assets; DebtRatio: ratio of total debt to total assets; Asset: natural logarithm of total assets; ListYear: natural logarithm of number of years since listing

Table 5 illustrates that there is a significantly strong correlation coefficient between ASC GKN and ASC KO (r

= 0.83, p < 1%) Considering a correlation between the explanatory variable and dependent one, TobinQ reveals

a signal of positive and statistically significant correlation with ASC GKN as well as ASC KO while DebtRatio shows the opposite In addition, a negative and statistically significant correlation is found between Turnover and

Volatility and ASC GKN Moreover, the maximum VIF in Table 5 is 1.38, showing that the problem of

multicollinearity is not severe in a regression model

Table 6 below shows the results of regression from Equation (8) The results of F test, Breusch-Pagan test and Hausman test from Column [1] to [3] in Table 6 recommend FEM method for performing regression equation Although the Hausman-test result in Column [4] recommends REM method (χ2 = 12.59; p = 0.08 >

0.05), the regression results using the FEM method are not significantly different compared to those using the REM method Therefore, the FEM method is used to estimate regression equation

Table 6 Determinants of ASC estimated from GKN model

Breusch-Pagan Test 863.88 *** 789.35 *** 863.73 *** 928.00 ***

Note: *** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level The t-statistics are based on robust standard errors

ASC GKN is estimated from trade-indicator GKN model; Turnover: the average of traded shares divided by outstanding shares; Depth: the average of shares available at both the best bid and ask prices divided by number of outstanding shares; Volatility: standard deviation

of the midpoint; TobinQ: the sum of market value of stock and total debt divided by total assets; Opp: growth opportunity, Opp = 1 if TobinQ > 1 high growth opportunity; otherwise, Opp = 0; DebtRatio: ratio of total debt to total assets; BankRatio: the sum of short-term bank loan and long-short-term bank loan divided by total assets; Policy: policy on adjusting price limit range, Policy = 1 if years of study belong to the period 2013-2016 (price limit range of 7%), and Policy = 0 if years of study belong to the period 2010-2012 (price limit range of 5%); Asset: natural logarithm of total assets; ListYear: natural logarithm of number of years since listing.

Table 6 reveals the regression results in 4 Columns Columns [1] and [2] present the relationship between

Turnover and ASC GKN while Columns [3] and [4] consider the effect of Depth on ASC GKN In addition, the

regression coefficients on TobinQ and DebtRatio are presented in Columns [1] and [3] while Opp and BankRatio are shown in Columns [2] and [4] Finally, the regression results of the Volatility, Policy and control variables including Asset and ListYear affecting ASC GKN were presented in all 4 Columns It can be seen that the model

has a relatively high degree of relevance by virtue of the adjusted R2 from Column [1] to [4] with 47.5%, 46.6%, 47.6% and 46.7% respectively

The results show that two liquidity factors have a negative and significant effect on asymmetric

information Specifically, the coefficients on Turnover (Columns [1] and [2]) and Depth (Columns [3] and [4])

are negative and statistically significant at 1% significance level, accepting the hypothesis H1 Unlike the initial

expectation, the coefficient on Volatility is negative and statistically significant at 1% significance level in all 4

Columns, rejecting the hypothesis H2

In addition, the coefficients on TobinQ (Column [1] and [3]) and Opp (Column [2] and [4]) are positive and

statistically significant at 1% significance level The hypothesis H3, therefore, is accepted Considering two