Stock Liquidity, Minimum Tick Size and
Seasoned Equity Offering Underpricing
Antonie C. Kist
Master Thesis
Student Number: 6030092
Supervisor:
prof. dr. F. (Florencio) Lopez de Silanes Molina
Date:
July 1, 2017
Statement of Originality
This document is written by Antonie Kist who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.Abstract
This thesis does an attempt to extend the available knowledge of SEO underpricing by focusing on stock market liquidity. While literature suggests stock liquidity is an important determinant of SEO underpricing, this is to my knowledge the first paper examining this relationship in detail. By investigating both individual stock liquidity characteristics and market wide liquidity dynamics, relating stock liquidity to SEO underpricing gives additional explanatory findings. Using OLS regression with year fixed effects, I find pre‐issue stock liquidity being a strong predictor for SEO underpricing. A shift from the most liquid quintile to the least liquid quintile of stocks issued, explains a 1.64 percentage point increase in SEO underpricing. With an average underpricing level of 3.22%, the magnitude of liquidity as a predictive variable is economically substantial. Decimalization, a regulatory change reducing tick sizes, improved SEO underpricing circumstances for low priced stocks significantly by stepping away from offer price rounding practices.
Contents
1 Introduction ... 5 2 Literature Review ... 8 2.1 Introduction to Seasoned Equity Offerings ... 8 2.2 Underpricing of Seasoned Equity Offerings ... 9 2.2.1 Theoretical Research ... 9 2.2.2 Empirical Research ... 10 2.3 Stock Liquidity ... 12 2.3.1 Introduction to Stock Liquidity... 12 2.3.2 Stock Liquidity and Seasoned Equity Offering Underpricing ... 12 2.3.3 Decimalization as an Exogenous Liquidity Shock ... 13 3 Hypotheses and Methodology ... 14 3.1 Stock Liquidity and SEO Underpricing ... 14 3.1.1 Hypothesis ... 14 3.1.2 Methodology ... 14 3.2 Minimum Tick Size and SEO Underpricing ... 17 3.2.1 Hypothesis ... 17 3.2.2 Methodology ... 17 4 Data and Summary Statistics ... 19 4.1 Data Collection ... 19 4.1.1 Seasoned Equity Offerings ... 19 4.1.2 Stock Data ... 19 4.1.3 Balance Sheet Items ... 20 4.1.4 Underwriter Reputation ... 20 4.1.5 Credit Ratings ... 20 4.2 Summary Statistics ... 21 5 Results ... 22 5.1 Stock Liquidity and SEO Underpricing ... 22 5.2 Minimum Tick Size and SEO Underpricing ... 24 6 Robustness ... 27 6.1 SEO Underpricing Determinants Over Time ... 27 6.2 Multicollinearity ... 30 6.3 Propensity Score Matching ... 32 7 Conclusion and Discussion ... 34 Appendix A ‐ Variable Descriptions ... 39 Appendix B ‐ Correlation Matrix ... 411 Introduction
Underpricing coming with initial public offerings (IPOs) is widely discussed in current available academic literature. The phenomenon of underpricing coming with seasoned equity offerings (SEOs) is relatively underexposed. SEO underpricing occurs for several reasons. In theory SEOs are underpriced to attract both informed and uninformed investors (Rock, 1986). Another information asymmetry related theory states that SEO are underpriced as a result of manipulative arbitrage by informed traders (Gerard & Nanda, 1993). Corwin (2003) uses empirical research to find determinants of SEO underpricing. In his paper, he puts an alternative view of an SEO being a temporary liquidity shock that needs to be absorbed by the market and is compensated by a discount on the issue. Butler, Grullon & Weston (2005) find lower SEO underwriter fees for liquid stock because they are easier absorbed by the market and therefore less costly to issue. Having publicly traded stock prior an offering as a unique characteristic of an SEO compared to an IPO, the alternative view of Corwin (2003) and the findings of Butler et al. (2005) are the motivation to focus on the relation between liquidity and SEO underpricing.
The view of SEO underpricing being the compensation for a temporary liquidity shock is followed by the idea that issues of stock that are able to absorb this shock relatively well, need less compensation in the form of underpricing. As liquid stock has better abilities to absorb liquidity shocks, the first hypothesis I examine is: firms with a more pre‐issue liquid stock experience lower levels of SEO underpricing.
To test the first hypothesis, I run OLS regressions using year fixed effects with SEO underpricing as dependent variable and liquidity and controls as independent variables. For robustness purposes, I use seven different measurements of liquidity. Regarding the topic of this thesis, I consider the Amihud illiquidity measurement to be the most important liquidity variable since it measures change in price relative to the dollar volume traded. Control variables are based on Corwin’s (2003) findings on SEO determinants.
I collect a sample of 5,419 SEOs dated from 1990 to 2016 from the Thomson ONE equity issue database and I collect 180‐day pre‐issue stock and firm data from CRSP. The SEOs in the sample are on average significantly underpriced with 3.22%. The results of the regression show a significant negative relation between pre‐issue stock liquidity and SEO underpricing during the sample period at six of the seven liquidity variables, indicating that more liquid pre‐issue stock comes with lower levels of SEO underpricing. Controlling for other factors, the least liquid quintile is on average 1.64 percentage point more underpriced compared to the most liquid quintile of the sample. This represents a substantial part of SEO underpricing as it is 50.9% of the average underpricing in the
measurement shows more pronounced effects. Shifting from the most liquid to the least liquid quintile shows a 1.92 percentage point increase in SEO underpricing.
To further explorer the interactions between liquidity conditions and SEO underpricing, I conduct a test with an exogenous liquidity shock in the form of changes in minimum tick size. The minimum tick size changed from 1/8 dollar to 1/16 dollar in 1997 to 1/100 dollar in 2001. These rule changes are often called decimalization. The regulatory change allows underwriters to deviate from price rounding at 1/8 dollar increments and offer closer to a stock’s consensus value. As can be seen in Figure A, rounding at certain decimal fractions is clustered less after reductions in tick size. The table shows the decimal fractions of offer prices across time. Corwin (2003) documents that offer price rounding affects SEO underpricing in a sample from 1980 to 1998. Offer price rounding is most pronounced with low priced stock, since with low priced stock the difference in decimal portions as a result of price rounding is relatively large. Corwin (2003) finds a significant negative relation between stock price and SEO underpricing. By combining Corwin’s findings and decimalization, my second hypothesis states that the effects of offer price rounding on SEO underpricing is less pronounced after decreases in tick size. Figure A ‐ Decimal Portions of Offer Prices This figure shows the decimals portions of 5,419 SEO offer prices from 1990 to 2016. Decimal portions are on the vertical axis and issue dates on the horizontal axis. Interacting a decimalization dummy with price in an OLS regression with year fixed effects results in a confirmation of the second hypothesis. I find less dependency on stock pricing as explanatory variable for SEO underpricing after the decimalization of 1997 and 2001 by finding a significant
0 .2 5 .5 .7 5 1 De c im a l P o rt io n Of fe r P ri c e 1990 1995 2000 2005 2010 2015 Issue Date
positive signed interaction term in the regression. This means that reducing the tick size leads to reduced underpricing for firms with low priced stock. The magnitude in witch SEO underpricing depends on the stock price has on average reduced with 77.2% as a result of the 1997 decimalization. Similar results are found as a consequence of the 2001 decimalization. In line with these findings, a diff‐in‐diff analysis with propensity score matching shows that offer price rounding reduced with 19.5% for low priced stocks and 11.6% for high priced stock. This supports the theory that low priced stocks benefit the most from stepping away from offer price rounding. This thesis differentiates from the majority of current literature by analyzing SEOs in a way that is consistent with Corwin’s (2003) alternative view of an SEO being a temporary liquidity shock with a compensation in the form of underpricing. While literature suggests liquidity is an important determinant of SEO underpricing, this is to my knowledge the first paper examining this relationship in detail. I find liquidity being a powerful predictor for SEO underpricing and robust to omitting firm size as independent variable. An additional contribution to current literature is the change in the effect of offer price rounding on SEO underpricing as result of decimalization. Decimalization does not only affect regular stock trading, but also improves the equity offerings of low priced stock in terms of SEO underpricing. Besides the structural change in offer price rounding over time, I also find differences in the predictive power of other SEO determinants compared to literature that is based on less recent samples of SEOs. In the used sample, I find Corwin’s (2003) determinants come with an R2 of 0.25 using SEOs before 1997 and an R2 of 0.05 when running regressions on SEOs occurred after 2001. The theoretical framework and a review of previous academic work related to this topic will be described in Section 2. Section 3 describes the hypothesis and methodology. Data collection and results are discussed in Section 4 and 5 respectively. I conduct robustness checks in Section 6 and conclude in Section 7.
2 Literature Review
The literature review starts with an introduction to SEOs with a focus on underpricing theories and empirics. Secondly, different aspects of liquidity are discussed and how it is assumed to be related to SEO underpricing. At last, I introduce changes in tick size. These regulatory changes serve as an exogenous liquidity shock to find evidence on causal relations and to prevent endogeneity.
2.1 Introduction to Seasoned Equity Offerings
An SEO is a sale of shares of an already publicly traded company (Welch, 2011). While most research in the area of equity offerings is about IPOs, research done in the area of SEOs is relatively limited. Nonetheless, literature shows that SEOs raise more capital than IPOs and have become a more important way of raising capital for firms through time. Bortolotti, Megginson and Smart (2008) document that the global SEO issuance volume in 2004‐2005 was almost twice as large as the IPO issuance volume. They also state that the number of SEOs has increased globally from 1,099 raising $91,9041 million in 1991 to 3,223 SEOs raising $320,714 million in 2004.
Firms can have several reasons to plan an SEO. DeAngelo, DeAngelo and Stulz (2010) state that issuing shares is related to the stage of a firm’s lifecycle and cash needed for investments. Besides these firm growth driven reasons, asymmetric information dynamics play a key role in equity issues. Several studies find evidence that SEOs coincide with market timing (DeAngelo et al., 2010; Hovakimian, Opler, & Titman, 2001). Graham and Harvey (2001) asked CFOs anonymously if the under‐ or overvaluation of stock was important to consider an SEO. Two‐thirds of the respondents agreed.
Myers and Majluf (1984) develop an asymmetric information based pecking order theory. They argue that the sensitivity of equity to private information is larger than the sensitivity of debt or retained earnings to private information. The pecking order theory suggests that managers reveal less information when they use retained earnings over debt or debt over equity for financing new investments. As a consequence, a firm only chooses to use an SEO to finance investments if the use of retained earnings or debt result in lower cash flows than using equity as a financial resource when taking information costs in account. Trade‐off theory suggests that firms issue equity to keep their debt‐equity ratio constant (De Jong, Verbeek, & Verwijmeren, 2011). Using a two‐step generalized method of moments (GMM) De Jong et al. (2011) find that pecking order theory is more pronounced when firms issue equity and trade‐off theory is more pronounced when a firm repurchases equity.
The costs coming with issuing equity are called flotation costs. Flotation costs are a significant portion of the gross proceeds of an SEO and can be separated in direct flotation costs and indirect flotation costs (Eckbo, Masulis, & Norli, 2007). Direct flotation costs include underwriter fees, legal costs, registration fees and costs of management time of which underwriter fees are most discussed in the literature (Eckbo et al., 2007). While Chen and Ritter (2000) and Hansen (2001) find that fees cluster at 7% of gross proceeds at IPOs, there is no evidence on the clustering of underwriter fees at SEOs (Butler et al., 2005).
The two main indirect flotation costs involve announcement effects and underpricing. Several studies find evidence on negative stock returns after SEO announcements. Hansen and Crutchley (1990) report a negative abnormal return of 3.65% from one day prior the announcement to the day of the announcement. Altinkilic and Hansen (2003) find an abnormal return of ‐2.23% surrounding SEO announcements in their sample from 1990 to 1997. Since this thesis is focused on the underpricing of SEOs, the next paragraph is dedicated to SEO underpricing.
2.2 Underpricing of Seasoned Equity Offerings
Underpricing is considered the most important indirect flotation cost of equity offerings (Eckbo et al., 2007). It is defined as the relative difference between the offer price and the pre‐offer closing price (Altınkılıç & Hansen, 2003; Corwin, 2003) and can be interpreted as money left on the table by firms when issuing equity (Mola & Loughran, 2004). While underpricing at IPOs is investigated extensively, SEO underpricing is relatively underexposed.
Previous empirical studies show that on average the offer price is lower than the pre‐offer closing price and that SEOs are indeed underpriced. Altinkilic and Hansen (2003) find an average SEO underpricing of 2.58%. Corwin (2003) finds a close‐to‐offer return of ‐2.21% in a sample of 4,454 SEOs from 1980 to 1998. A more recent study by Jeon and Ligon (2011) report an average SEO underpricing of 3.04% during their sample from 1997 to 2007. Given the large amounts of gross proceeds and the economic importance of underpricing of SEOs have encouraged researchers to investigate this phenomenon.
2.2.1 Theoretical Research
Many underpricing theoretical frameworks are derived from IPO underpricing and are related to information asymmetry theories. A well‐known model is introduced by Rock (1986). His model assumes that investors subscribing for equity offerings consist of informed and uninformed investors. He states that informed investors only subscribe if the offer price is less than the fair value
not willing to subscribe for those equity offerings. In order to guarantee that uninformed investors purchase the issue, underwriters must price the shares at a discount. This theory is commonly known as the winner’s curse.
Parsons and Raviv (1985) model a theoretical framework for SEO underpricing where market prices and offer prices are jointly determined. In equilibrium, the price in the competitive market operating prior to the arrival of the new issue will always be higher than the price at which the new issue will be sold (Parsons & Raviv, 1985).
A model of manipulative arbitrage by informed traders explaining SEO underpricing is developed by Gerard and Nanda (1993). Informed traders anticipate on the winner’s curse theory suggested by Rock (1986) by selling shares prior to an SEO and then subscribe for the offer. Therefore, consistent with market timing theory, the price drops. According to Rock’s model, the offer price is lower than the market price and traders profit from this. This worsens the amount of indirect flotation costs both in negative abnormal returns as in underpricing (Corwin, 2003).
2.2.2 Empirical Research
Empirical studies have shown an increase in SEO underpricing over time. Corwin (2003) reports an average underpricing of 1.30% in the 1980s and underpricing levels of 2.92% in the 1990s. Consistent with these results are the findings of Autore (2011). He finds that SEO underpricing is significantly increased from 0.87% between 1982 and 1987, to 2.16% between 1988 and 1993 and finally raised to 3.20% from 2000 to 2004. Corwin (2003) and Kim and Shin (2004) evidence that this increase is a result of the on August 25, 1988 adopted SEC Rule 10b‐21, that prohibits investors from covering a short position with stock purchased in a new offering if the short position was established between the filing date and the distribution date . Restricting short sells reduced the amount of information floating into the market as a result of trading and therefore increased SEO underpricing. Mola and Loughran (2004) state that the increase in SEO underpricing is a result of an increase in investment banking power. They also find a positive relationship between underpricing and underwriter spreads.
Besides the increase in SEO underpricing, empirical research provides determinants and relevant measurements that explain underpricing of SEOs. Corwin (2003) discusses several determinants for SEO underpricing. Using a sample of 4,454 SEOs from 1980 to 1998, he contributes to the literature by specifying and testing SEO underpricing determinants. He uses the bid‐ask spread and firm size as measures of asymmetric information, but finds little evidence on its relation with SEO underpricing. Huang and Zhang (2011) find evidence that market capitalization, as a measure of firm size, is significantly positively related to SEO underpricing. Other empirical research is more consistent with the theoretical frameworks regarding information asymmetry. A working paper of
Liu and Malatesta (2006) finds evidence for the influence of information asymmetry on SEO underpricing. They state that having a credit rating reduces asymmetric information and consequently the underpricing of SEOs. Eckbo et al. (2007) state that the percentage of tangible assets on total assets has a negative effect on SEO underpricing. This is because tangible assets come with less information asymmetry and are better to valuate (Peyer, 2002).
As an uncertainty variable Corwin (2003) uses stock volatility, measured by the standard deviation of stock returns between 30 and 11 days prior the offering. He finds a statistical significant positive relation between volatility and SEO underpricing. These findings are consistent with other literature (Altınkılıç & Hansen, 2003; Bhagat, Marr, & Thompson, 1985; Kim & Shin, 2004). The mechanism assumed to increase underpricing is that underwriters want to be certain that the offer price is under the market price. When the price is fluctuating more, the offer price has to be lower relative to the market price to ensure underpricing (Corwin, 2003).
The offer size relative to firm size is significantly positively related to SEO underpricing and is suggested to be a result of price pressure (Corwin, 2003). Assuming downward sloping demand curves, an increase in supply will drop stock prices. According to Scholes (1972) and supported by Corwin (2003), only temporary price pressure will be incorporated in the underpricing. It can be seen as a temporary liquidity shock that needs to be absorbed by the market. The results of Altinkilic and Hansen (2003) are consistent with this view.
Corwin (2003) finds clustering of stock prices and offer prices at integers, $0.25 and $0.125 increments. According to Corwin (2003), offer price rounding is a significant determinant of SEO underpricing, since he finds a negative relation between underpricing and stock price. The theory behind this, is that rounding has a relatively larger effect on low priced stocks. Altinkilic and Hansen (2003) support Corwin’s findings by finding a positive relation between the inverse of stock price and SEO underpricing.
The manipulative arbitrage hypothesis derived from the model of Gerard and Nanda (1993) is tested in Corwin’s model by taking cumulative abnormal returns (CAR) prior the offer. He finds that SEO underpricing increases significantly when the CAR prior the offer was negative. In line with current literature, Corwin (2003) also finds a negative coefficient on a New York Stock Exchange (NYSE) dummy, representing firms listed on the NYSE. This indicates that SEOs of NYSE listed firms are less underpriced. Altinkilic and Hansen (2003) find similar results in their study on SEO underpricing determinants. They explain the difference of underpricing between NYSE and NASDAQ listed firms as a result of SEOs of NASDAQ listed firms being more uncertain. Mola and Loughran (2004) find a positive coefficient with a NASDAQ dummy on SEO underpricing.
Altinkilic and Hansen (2003) state that underwriters have to maintain their reputation for future offers by offering fair valued issues. Therefore, they also include underwriter reputation in their regression. They use the method of Carter and Manaster (1990). More empirical studies find higher ranked underwriters being associated with lower amounts of SEO underpricing (Kim & Shin, 2004; Mola & Loughran, 2004; Safieddine & Wilhelm, 1996).
2.3 Stock Liquidity
2.3.1 Introduction to Stock Liquidity
Stock liquidity is the ability to trade a security quickly at a price close to its consensus value (Foucault, Pagano, & Röell, 2013). Stock liquidity is important to investors, since it has impact on security prices (Amihud, Mendelson, & Pedersen, 2005). Amihud et al. (2005) define five sources of illiquidity: exogenous transaction costs, demand pressure, search frictions, inventory risk and private information. These sources of illiquidity cause a stock’s trading price to differ from its consensus value. Exogenous transaction costs affect prices directly and are considered to be brokerage fees, order‐processing costs, or transaction taxes. Demand pressure has impact on liquidity when the availability of buyers is limited. Also the ability to search for suitable buyers gives frictions and limits the demand and liquidity of the asset. Simple microeconomics tell that with low demand, prices drop (Amihud et al., 2005).
To cover the gap between buyers and sellers, market makers provide intermediary services. They buy shares from sellers, keep them in their inventory and later sell them to buyers. Several papers (Amihud et al., 2005; Brunnermeier & Pedersen, 2009) provide a model where market makers demand a compensation for the risk they are exposed to when holding securities in their inventory. Liquid assets need less inventory, so the compensation for inventory holding risks can be lower for liquid securities (Amihud et al., 2005).
Information asymmetry has effect on liquidity, since buyers (sellers) worry that the seller (buyer) they buy (sell) the stock from (to) has private information. As a result, bid prices will drop and ask prices will rise to compensate a potential loss from a trade against an informed counterparty (Amihud et al., 2005).
2.3.2 Stock Liquidity and Seasoned Equity Offering Underpricing
Little evidence exists of the effect of stock liquidity on SEO underpricing. The effect of stock liquidity on another type of floatation costs of issuing seasoned equity has been investigated by Butler, Grullon and Weston (2005). Using different measurements of stock liquidity, they show that the gross fees of underwriters decrease when the stock of the issuing firm is more liquid. The hypothesis
they test and confirm is that higher liquid stock is easier to be absorbed by the market and therefore less costly for the underwriter to issue.
He, Wang and Wei (2014) also show an interaction with liquidity, SEOs and SEO underpricing. Their research provides evidence that SEOs provide post‐SEO liquidity. They find that equity issues that increase stock liquidity the most, experience lower amounts of SEO underpricing.
Corwin (2003) finds some evidence of the effect of liquidity on underpricing by showing a positive relationship between the bid‐ask spread and SEO underpricing. He also suggests the view of an SEO being a temporary liquidity shock that needs to be absorbed by the market. This temporary liquidity shock can be compensated in the form of a discount on the stock price, resulting in underpricing (Corwin, 2003).
2.3.3 Decimalization as an Exogenous Liquidity Shock
To find evidence on the causal relation between liquidity and SEO underpricing, I use and exogenous factor to prevent endogeneity problems. An exogenous liquidity shock used in the literature, is a change in rules of the NYSE, NASDAQ and American Stock Exchange (AMEX) in the form of decimalization (Bessembinder, 2003; Bharath, Jayaraman, & Nagar, 2013; Chordia, Roll, & Subrahmanyam, 2001; Goldstein & Kavajecz, 2000). The rule change implied a change in tick size, allowing securities to trade closer to their consensus value and by definitions improve liquidity. On June 24, 1997 and on June 2, 1997 the NYSE and the NASDAQ respectively changed the tick size from 1/8 of a dollar to 1/16 of the dollar (Jones & Lipson, 2001). A similar rule change On January 29, 2001 and on April 9, 2001 implied a change in tick size on the NYSE and NASDAQ respectively from 1/16 to 1/100 of a dollar (Bharath et al., 2013).
Liquidity improvements as result of tick size reductions in security markets are often shown by empirical studies. Most evidence is found in the form of reduced bid‐ask spread. Goldstein and Kavajecz (2000) show bid‐ask spread reductions post the 1997 change in tick size. Bessembinder (2003) finds a substantial decline in quoted bid‐ask spreads on each market after the 2001 decimalization. Chrodia, Sarkar and Subrahmanyam (2005) show that the decimalization shock of 2001 reduced bid‐ask spreads with 61%. Ahn, Cao and Choe (1996) show that stocks with greater trading activity, lower prices and strong competition from the regional exchanges experienced great spread reductions after a decreased tick size.
According to Foucault et al. (2013), reducing tick size reduces profits for liquidity suppliers and trading costs for liquidity demanders. It also reduces the numbers of shares offered at each
3 Hypotheses and Methodology
To find evidence on a relation between liquidity and SEO underpricing, I conduct empirical research on SEO underpricing related to both stock liquidity and market liquidity. First I will describe my hypothesis and methodology on SEO underpricing related to stock liquidity, followed by my market liquidity related hypothesis and methodology on SEO underpricing.
3.1 Stock Liquidity and SEO Underpricing
3.1.1 HypothesisPrimarily based on Corwin’s (2003) alternative view of an SEO being a temporary liquidity shock and underpricing being the compensation for this, a stock with best liquidity shock absorption characteristics needs less compensation and therefore experiences lower amounts of SEO underpricing. Supported by Butler et al. (2005) who finds evidence of liquidity being negatively related to floatation costs in the form of underwriter spreads, the first hypothesis to be tested is:
H1: Firms with a more pre‐issue liquid stock experience lower levels of SEO underpricing.
Since stocks with a more pre‐issue liquid stock are considered to have better liquidity shock absorption characteristics, I expect higher levels of SEO underpricing coming with issues with relatively illiquid stock. An overview of expected outcomes of variables described in the next section, can be found in Table 1.
3.1.2 Methodology
In order to test the first hypothesis, I combine previous used methodologies from Corwin (2003) and Butler et al. (Butler et al., 2005). Both papers used ordinary least squares (OLS) regression models with annual fixed effects to find evidence of several determinants of SEO flotation costs. To test my hypothesis, I will run a OLS regressions with year fixed effects containing SEO underpricing as dependent variable and liquidity and controls as independent variables. The regression function is as follows:
Consistent with previous literature, I define underpricing as the percentage difference between the previous day closing price and the offer price. I multiply this by minus one hundred to have more meaningful coefficients in the output. Section 3.1.2.1 and 3.1.2.2 describe liquidity and control
variables respectively. For exact definitions of variables used including supporting literature, see Appendix A.
3.1.2.1 Liquidity Measurements
Due to limitations on data availability, I use liquidity measurements based on daily quoted stock data. I use multiple measurements of liquidity to find stronger evidence on the effect of stock liquidity on SEO underpricing. All measurements of liquidity are based on a maximum of 180 trading days before the offer date. Volume based liquidity measurements are adjusted for NASDAQ listed stock. Volumes of these stocks are divided by two. This is a commonly used adjustment because in dealer markets, trades are often immediately turned around by the market maker and thus double counted (Butler et al., 2005). In my research, I will use the following measurements of liquidity:
1. Amihud Illiquidity Ratio. Since the rationale behind the first hypothesis is to relate SEO
underpricing to a stock’s ability to absorb temporary liquidity shocks, I borrow a widely used liquidity measurement suggested by Amihud (2002). The metric is based on price changes associated with order flow or trading volume. It is measured by taking the average of absolute daily returns divided by the dollar volume of that day:
1 | |
Where Diy is the number of days for which data of a specific stock is available, Riyt is the return of a stock on a specific day and VOLDiyt is the dollar traded volume on that day. If a stock is able to process large trading volumes without price changes, the stock is more liquid. Since this metric is most consistent with my hypothesis, I considered it as the most important liquidity measurement regarding this topic. I expect a positive relation with the Amihud illiquidity ratio and SEO underpricing.
2. Turnover. Another measurement proposed by Amihud (2002) is the turnover of a stock. It
is calculated as the average volume of daily traded stock divided by the total number of shares outstanding on that specific day. It is a metric of the average proportion of a firm’s equity that is traded daily. To be in line with the first hypothesis, I expect a positive signed coefficient when putting turnover as a liquidity measurement in the regression.
3. Volume. This measure is defined as the average daily traded volume of a stock. This metric
is also used by Butler et al. (2005). A negative signed coefficient is expected when putting volume in the regression.
4. Dollar Volume. This variable is calculated by taking the average of the multiplication of the
daily closing price with the daily traded volume of a stock (Foucault et al., 2013). I expect a negative relation between dollar volume and SEO underpricing.
5. Quoted Spread. This metric is defined as the average of the difference between the daily
closing bid and closing ask prices. Butler et al. (2005) also used this measure. Corwin (2003) uses it as a measurement for asymmetric information. I expect higher levels of underpricing with larger quoted spreads.
6. Relative Spread. The relative spread is calculated by taking the average of the quoted
spread divided by the closing price on a daily basis. Consistent with my hypothesis, I expect a positive signed coefficient when running the regression. 7. Trade Size. This liquidity measurement is the average of the traded volume divided by the number of trades (Butler et al., 2005). I expect lower levels of underpricing coming with larger trade sizes. 3.1.2.2 Control Variables
To control for other explanatory variables determining SEO underpricing, I use SEO underpricing determinants found in existing literature. For comparability and consistency reasons, most control variables I use are explained by Corwin (2003). These determinants are similar to the ones used in most other literature (Altınkılıç & Hansen, 2003; Butler et al., 2005; Mola & Loughran, 2004). For robustness purposes I include additional control variables with a focus on asymmetric information measurements.
As a uncertainty control, I use stock return volatility between 30 and 11 trading days before the offer. The offer size relative to the firm size one day before the offer is used to control for price pressure. Controlling for the manipulative trading hypothesis, I use 5‐day pre‐offer cumulative abnormal returns (CAR). In line with Corwin’s (2003) method, I create a variable equal to the CAR when the CAR is positive and zero otherwise, and a variable equal to the CAR when the CAR is negative and zero otherwise. To control for offer price rounding, I include a tick dummy labeled tick<1/4 that is equal to one if the decimal portion of the pre‐offer closing price is below $0.25 (Corwin, 2003). The tick dummy is equal to zero otherwise. The offer price clustering effect must be more pronounced at lower prices, so I include an interaction term between price and the tick dummy (Corwin, 2003). A NASDAQ dummy equal to one when a stock is listed on NASDAQ and zero when a stock is listed on the NYSE is also included.
Limited evidence is found by Corwin (2003) on the effect of asymmetric information on SEO underpricing. Since theoretically both underpricing and liquidity are determined by asymmetric information, therefore I explicitly control for asymmetric information using serval metrics. I include
firm size measured in market capitalization and book value of intangible assets relatively to the book value of the total assets (Eckbo et al., 2007). I also include a dummy equaling one if a firm has a credit rating one year or less before the issue date (Liu & Malatesta, 2006). Controlling for both underwriter power and asymmetric information, I include an underwriter reputation ranking proposed by Loughran and Ritter (2004).
3.2 Minimum Tick Size and SEO Underpricing
3.2.1 HypothesisMaking intuitive assumptions based on SEO underpricing literature and micro‐market structures, I attempt to puzzle out the effect of market liquidity improvements on SEO underpricing. If comparing equity offering markets to regular stock markets, one can state that the difference between pre‐ offer closing price and offer price can be compared to a regular stock market’s bid‐ask spread. Based on existing evidence of quoted spread reductions as a result of decimalization, I use decimalization as an exogenous liquidity shock. In this way I try to find evidence of the effect of market liquidity dynamics on SEO underpricing.
The intuition is that reducing the tick size on a stock market, enables stock to be valued closer to their consensus value and by definition becoming more liquid. Valuing stocks for equity offerings with a limited number of decimal fractions can be seen as rounding of pre‐offer closing prices and offer prices. Increasing the number of decimal fractions, should enable underwriters to deviate more from price rounding. According to Corwin (2003), the effect of price rounding has most effect on low priced stock. He states stock price is an important determinant of SEO underpricing as result of offer price rounding. High priced stock suffers less from offer price rounding than low priced stock and therefore he expects and finds a significant negative effect of stock price on SEO underpricing. However, the effect of stock price on underpricing is not a qualitative effect on underpricing but a result of market limitations that limit SEOs to price at their consensus value. To test if decimalization reduces the effect of offer price clustering on SEO underpricing, I will test the following hypothesis: H2: The effect of offer price rounding on SEO underpricing is less pronounced after decimalization. 3.2.2 Methodology
To test the second hypothesis, I will focus on Corwin’s (2003) explanatory variables for offer price rounding on SEO underpricing. These are price, tick<1/4 and an interaction term between price and
portion of the pre‐offer closing price is below $0.25, is significantly positively related to SEO underpricing according to Corwin (2003). He also states and find that the effect of the tick<1/4 dummy is more pronounced at low priced stocks, since an interaction between price and tick<1/4 is significantly negatively related to SEO underpricing.
In order to find if decimalization reduced the effect of price rounding on SEO underpricing, the findings of Corwin must be less pronounced after a reduction on tick size. First I generate dummy variables representing the 1997 and 2001 changes in tick size. Decimalization1997 is equal to one if an issue takes place after June 24, 1997 (June 2, 1997) and is listed on the NYSE (NASDAQ) and zero otherwise. Decimalization2001 is equal to one if an issue takes place after January 29, 2001 (April 9, 2001) and is listed on the NYSE (NASDAQ) and zero otherwise. The effect is tested using OLS regression with year fixed effects and SEO underpricing as dependent variable. By interacting the decimalization dummies with price, tick<1/4 and the interaction price*tick<1/4, I attempt to find a reduced effect of offer price rounding on SEO underpricing after a decrease in minimum tick size. I test this effect for both regulatory changes and I will use similar controls as described in Section 3.1.2.2. The regression function is stated as: ∗ 1 4⁄ ∗ ∗ 1 4⁄ ∗ 1 4⁄ ∗ 1 4⁄
If decimalization improved conditions for low priced stock in terms of a reduced effect of price rounding during SEOs, I expect the variables price, tick<1/4 and price*tick<1/4 to be similar signed as Corwin’s (2003) findings and opposite signed when interacted with decimalization dummies. Table I ‐ Key Variable Expectations This table lists all key variables and expected signs according to the proposed OLS regression in section 3. The signs are based on the hypotheses stated in section 3. Hypothesis 1 Hypothesis 2
Variable Expected sign Variable Expected sign Amihud Illiquidity + Price*Decimalization + Turnover ‐ Tick<1/4*Decimalization ‐ Volume ‐ Price*Tick<1/4*Decimalization + Dollar Volume ‐ Decimalization ‐ Quoted Spread + Price ‐ Relative Spread + Tick<1/4 + Trade Size ‐ Price*Tick<1/4 ‐
4 Data and Summary Statistics
To be in line with existing literature, the collecting of the SEO data is done in a similar way as Corwin (2003). This part will explain the collecting and modification of the data per different type of data. Starting with SEOs followed by stock data, balance sheet items, underwriter reputation and credit ratings. Summary statistics are provided at last.
4.1 Data Collection
4.1.1 Seasoned Equity Offerings
SEO data is collected from Thomson ONE (formerly known as SDC Platinum). Excluding IPOs from common stock issues in the United States from 1990 to 2016 yields 19,395 SEOs. Following Corwin (2003) I exclude utilities2, rights, mutual conversions, closed‐end funds3 and real estate investment trusts (REITs)4. This leaves out 3,727 observations. I only keep issues with common stock, class A common shares and class B common shares5. This eliminates 3,913 observations. 141 observations were eliminated to prevent bias from multiple issues on the same date from the same firm. Finally, a firm should at least issue some primary shares. Excluding issues without primary shares drops out another 1,921 SEOs, yielding 9,693 SEOs. 4.1.2 Stock Data Stock data is downloaded from the Center for Research in Security Prices (CRSP) database using the portal from Wharton Research Data Services (WRDS). Identifying corresponding stock information is done using 8‐digit historic CUSIP codes. Searching the CRSP data base from 1989 to 2016 matched the 8‐digit historic CUSIP codes from Thomson ONE with the CRSP PERMCO and PERMNO identifiers. The matching was done successfully for 8,544 observations. Only NYSE and NASDAQ issues6 are kept, dropping out 735 observations. This results in 7,809 observed SEOs.
Daily stock data around the issues is downloaded from CRSP using the Daily Extract with Time Window tool from WRDS. Underpricing is typically calculated relative to the previous day closing stock price. However, a significant part of the SEOs occur after the stock market closes (Eckbo & Masulis, 1992; Lease, Masulis, & Page, 1991; Safieddine & Wilhelm, 1996). In these cases, the closing stock price of the issue date should be used to calculate underpricing. To control for this, I use a volume‐based approach to identify these SEOs issued after the stock market closes (Corwin, 2003; Eckbo & Masulis, 1992; Liu & Malatesta, 2006). To consider an issue to be offered after the 2 Utilities are excluded by using 3‐digit SIC codes 491 to 494. 3 Mutual conversions and closed‐end funds are excluded by using the 4‐digit SIC code 6726. 4 REITs are excluded by using the 4‐digit SIC code 6798.
stock market closes, the trading volume on the day after the issue date must be twice as large as the trading volume on the issue date and also twice as large as the average trading volume of the 250 trading days before the issue date. For SEOs holding these two conditions the day after the issue date is considered to be the effective issue date. In my sample this applies to 52% of the observations. According to Altinkiliç and Hansen (2003) and Corwin (2003) this identification is at least 98% accurate.
Furthermore, all issues with less than 30 available trading days before and less than 11 days after the effective issue date are dropped out. Also issues with a stock split7 in the 11 days surrounding the stock offer are dropped out. Consistent with Corwin (2003) I exclude SEOs with offer prices lower than $3.00 and larger than $400.00. After these selection criteria 7,208 observations are left.
4.1.3 Balance Sheet Items
The balance sheet items of firms reported before the issue date is subtracted from the CRSP/Compustat merged database using WRDS. To obtain the most complete and recent data I first get all available data from quarterly reports and subsequently complete them with data from annual reports. The most recent data compared to the issue date is kept and should not be older than one year compared to the effective issue date. Sufficient data is available for 6,221 observations.
4.1.4 Underwriter Reputation
The reputation of underwriter is downloaded from the website of prof. Jay R. Ritter8. The data is partly based on the Carter and Manaster (1990) and Carter et al. (1998) rankings. The data is updated by Ritter on his website to 2015 and ranks underwriter with a grade scaled from 0 to 9 (Loughran & Ritter, 2004). If there is more than one underwriter per SEO, I use the lead investment bank underwriting the issue (Loughran & Ritter, 2004). Observations with an underwriter that is not present in the reputation ranking of prof. Jay R. Ritter are given a rating of 0 (Liu & Malatesta, 2006). Therefore all remaining observations have an underwriter reputation ranking. 4.1.5 Credit Ratings
After obtaining Standard & Poor’s (S&P) Global Company Keys (GVKEY) through the CRSP/Compustat merged database, I downloaded the S&P Domestic Long‐Term Issuer Credit Rating Variable from the Ratings database from Compustat (An & Chan, 2008; Liu & Malatesta, 2006). Of the sample of 7,208 SEOs, 1,546 were assigned a credit rating at one year before the effective issue date. 7 Stock splits are identified with the CRSP distribution code 5523. 8 https://site.warrington.ufl.edu/ritter/files/2016/06/Underwriter‐Rank‐1980‐2015.xls
4.2 Summary Statistics
The number of SEOs with all data available is 5,419. Table II summarizes offer and firm characteristics from the selected sample. I find significant underpricing of SEOs with an average of 3.22%. This is in line with current literature and continues to reflect higher levels of underpricing when the sample shifts towards more recent observations (Jeon & Ligon, 2011). Capital raised as a result of SEOs in the sample range from 400 thousand dollar to 17 billion dollar with an average of 167.19 million dollar. The relative offer size averages 22.82%. A small portion of SEOs in the sample have a relative offer size that exceeds their pre‐issue size. It is worth mentioning that the majority of these firms are pharmaceutical companies, indicating a possible capital issue to finance new drugs. Before running regressions, I checked for nonlinearities by generating scatter plots of the independent variables with SEO underpricing. I find curved plots with the liquidity variables, firm size and price. A correlation matrix containing used variables can be found in Appendix B.
Table II ‐ Summary Statistics
The table shows summary statistics of a sample of 5,419 SEOs on the NYSE and NASDAQ from 1990 to 2016 that meet sample restrictions described in section 4. SEO Underpricing is defined as minus one times the return from the previous day's closing price to the offer price in percentages. Capital raised is the number of shares offered times the offer price in millions. Price is the pre‐offer closing price. Offer price is the price at which the issue is offered. Relative offer size is measured as the number of shares issues divided by the number of shares outstanding on the day prior to the offer in percentages. Firm size is calculated as the number of shares outstanding times the price on the day before the offer quoted in millions. Volatility is the standard deviation of return between 30 and 11 days before the offer in percentages. Relative spread is the 180 pre‐offer trading day average of closing ask prices minus closing bid prices divided by the closing price in percentages. The significance of the mean of a variable being different from zero is showed by the p‐value derived from a student t‐test.
N Minimum Median Maximum Mean Standard
Deviation p‐value SEO Underpricing (%) 5,419 ‐36.45 2.12 34.47 3.22 5.13 0.000 Capital Raised (millions $) 5,419 0.40 70.00 17,000.00 167.19 573.50 0.000 Price ($) 5,419 2.32 19.09 390.00 24.73 23.46 0.000 Offer Price ($) 5,419 3.00 18.50 389.75 24.12 23.17 0.000 Relative Offer Size (%) 5,419 0.51 16.06 540.95 22.82 39.31 0.000 Firm Size (millions $) 5,419 2.87 459.21 317,592.10 2,042.86 8,816.81 0.000 Volatility (%) 5,419 0.34 3.14 53.10 3.63 2.46 0.000 Relative Spread (%) 5,419 0.00 0.71 19.15 1.43 1.80 0.000
5 Results
Section 5.1 shows results from tests of hypothesis one. The results from the second hypothesis are discussed in Section 5.2.5.1 Stock Liquidity and SEO Underpricing
Table III ‐ Regression Results for SEO Underpricing on Pre‐Stock Liquidity The table lists coefficients (standard errors) from OLS regressions of SEO underpricing on several pre‐stock liquidity characteristics and control variables. The panel is based on a sample of 5,419 SEOs on the NYSE and NASDAQ from 1990 to 2016 that meet sample restrictions described in section 4. The dependent variable is SEO underpricing defined as negative hundred times the return from the previous day's closing price to the offer price. The Amihud illiquidity ratio is the 180 pre‐offer trading day average of absolute daily returns divided by the dollar volume of that day. The turnover is the 180 pre‐offer trading day average volume of daily traded stock divided by the total number of shares outstanding on that day. Volume is the 180 pre‐offer trading day average traded volume. Dollar volume is the 180 pre‐offer trading day average the multiplication of the closing price with traded volume. Quoted spread is the 180 pre‐offer trading day average of closing ask prices minus closing bid prices. Relative spread is the 180 pre‐offer trading day average of closing ask prices minus closing bid prices divided by the closing price. Trade size is the 180 pre‐offer trading day average of the traded volume divided by the number of trades. Trade size is only available for NASDAQ listed firms. Firm size is calculated as the number of shares outstanding times the price on the day before the offer. Volatility is the standard deviation of return between 30 and 11 days before the offer. Relative offer size is measured as the number of shares issues divided by the number of shares outstanding on the day prior to the offer. CAR positive (negative) is equal to the sum of returns minus the returns of the CRSP value‐ weighted index 5 days prior the offer when positive (negative) and zero otherwise. Price is equal to the closing price on the day prior to the offer. Tick<1/4 is a dummy variable equal to one if the decimal portion of the pre‐offer closing price is below $0.25 and zero otherwise. NASDAQ is a dummy variable equal to one if stock was issued on NASDAQ and zero when otherwise. %Intangible assets is calculated as percentage of the book value of intangible assets relative to the book value of total assets. Underwriter reputation is the reputation ranking according to J. Ritter of the lead underwriter during the time of the offer. Credit rating dummy is equal to one if the firm has a credit rating one year or less prior the offer and zero otherwise. Standard errors are in the parentheses and significant levels on the 10%, 5% and 1% are indicated by *,**,*** respectively. 1 2 3 4 5 6 7 8 Log(Amihud Illiquidity) 0.295*** (0.070) Log(Turnover) ‐0.142 (0.098) Log(Volume) ‐0.195** (0.077) Log(Dollar Volume) ‐0.284*** (0.093) Log(Quoted Spread) 0.379** (0.140) Log(Relative Spread) 0.652*** (0.155) Log(Trade Size) ‐0.690** (0.308) Volatility 20.068*** 19.037*** 21.610*** 22.149*** 22.909*** 19.761*** 17.651*** 16.604*** (3.410) (3.690) (3.697) (3.617) (3.459) (3.399) (3.713) (3.216) Relative Offer Size 0.613*** 0.808*** 0.637*** 0.750*** 0.808*** 0.694*** 0.756*** 0.409** (0.199) (0.207) (0.197) (0.198) (0.201) (0.201) (0.202) (0.195) % Intangible Assets 1.238*** 1.129*** 1.210*** 1.179*** 1.147*** 1.238*** 1.234*** 1.457*** (0.391) (0.377) (0.384) (0.380) (0.373) (0.394) (0.390) (0.467) NASDAQ Dummy 0.619*** 0.388* 0.563** 0.528** 0.477** 0.663*** 0.658*** (0.210) (0.189) (0.203) (0.207) (0.198) (0.200) (0.198) Log(Price) ‐0.694*** ‐0.616** ‐0.682*** ‐0.835*** ‐0.643** ‐0.935*** ‐0.522** ‐0.745* (0.238) (0.243) (0.234) (0.229) (0.234) (0.239) (0.252) (0.363)Table III Continued Tick<1/4 0.438 0.476 0.443 0.445 0.434 0.455 0.440 0.696 (0.737) (0.741) (0.738) (0.740) (0.743) (0.737) (0.739) (0.813) Log(Price)*Tick<1/4 ‐0.113 ‐0.121 ‐0.113 ‐0.112 ‐0.106 ‐0.115 ‐0.109 ‐0.186 (0.220) (0.221) (0.221) (0.221) (0.222) (0.220) (0.219) (0.240) Log(Firm Size) ‐0.320*** 0.041 ‐0.292*** ‐0.121 ‐0.030 ‐0.184 ‐0.076 ‐0.490*** (0.094) (0.125) (0.090) (0.112) (0.120) (0.112) (0.114) (0.121) Credit Rating Dummy ‐0.109 0.028 ‐0.090 ‐0.043 ‐0.004 ‐0.107 ‐0.082 0.003 (0.238) (0.235) (0.237) (0.232) (0.231) (0.238) (0.238) (0.344) CAR Positive 2.796 2.622 2.957 2.925 3.007 2.848 2.420 2.916 (2.051) (1.998) (2.116) (2.085) (2.067) (2.070) (2.009) (2.298) CAR Negative 1.652 1.584 1.599 1.658 1.312 1.969 1.646 2.974 (2.087) (2.108) (2.094) (2.092) (2.096) (2.043) (2.089) (1.932) Underwriter Reputation 0.002 0.001 0.001 0.001 ‐0.000 0.001 ‐0.001 ‐0.006 (0.017) (0.017) (0.017) (0.017) (0.017) (0.017) (0.017) (0.023) Constant 10.115*** 7.887*** 8.721*** 8.738*** 8.154*** 8.936*** 8.064*** 18.185*** (1.811) (2.040) (1.873) (1.865) (1.928) (1.952) (2.032) (2.794) Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes N 5,419 5,419 5,419 5,419 5,419 5,419 5,419 4,000 Adj. R2 0.073 0.078 0.073 0.074 0.075 0.075 0.079 0.063 Economic Magnitude (%) 1.98 0.39 0.98 1.64 1.41 2.51 2.57
The results of the OLS regressions to test the first hypothesis are shown in Table III. Column 1 shows the result of control variables only. Column 2 to 8 show regression results including liquidity measurements. All seven coefficients of liquidity measurements are signed in the same direction as predicted in Table I. The Amihud illiquidity ratio, dollar volume and relative spread are significantly related to SEO underpricing in the expected direction at the 1% level. Volume, quoted spread and Trade Size are related to SEO underpricing in the proposed direction at the 5% significance level. Column 3 lists an insignificant related coefficient of turnover as explanatory variable to SEO underpricing. Please note that trade size is only available for NASDAQ listed stock, which explains the absence of the NASDAQ dummy and a lower amount of observations in this regression.
The economic magnitude can be best explained by comparing the first (lowest and most liquid) with the fifth (least liquid) quintile of SEOs based on the Amihud Illiquidity ratio. Controlling for other factors, shifting from the first to fifth quintile gives a 1.98 percentage point increase in SEO underpricing. This represents a substantial part of SEO underpricing as it is 61.4% of the average underpricing in the sample. The results are consistent with the findings of Butler et al. (2005) and the alternative view of SEO underpricing by Corwin (2003). The bottom row of Table III shows economic magnitudes for the other liquidity measurements calculated in a similar way. They all show an increase in SEO underpricing when shifting from the most liquid to the least liquid quintile of the specific liquidity measurement used in that row. The economic magnitude varies from a 0.39
measurements and averages a 1.64 percentage point increase. This is 50.9% of the average SEO underpricing of the sample.
Other significant determinants I find which are consistent with current literature about SEO underpricing are volatility, relative offer size and the NASDAQ dummy (Altınkılıç & Hansen, 2003; Corwin, 2003). I find limited evidence for asymmetric information as determinant of SEO underpricing. The percentage of the book value of intangible assets compared to the total book value of assets is the only significant variable explaining information asymmetry as determinant for SEO underpricing. Other proposed independent information asymmetry variables such as a credit rating dummy are insignificant but signed in the direction that supports information asymmetry theory. Consistent with Huang and Zhang (2011), I find weak evidence of firm size being negatively related to SEO underpricing. The variables price, tick<1/4 and the interaction term between them are in line with offer price rounding theory, but insignificant. I find no support for the manipulative trading theory by Gerard and Nanda (1993), as the CAR based variables are not significantly different from zero in any of the regressions.
Based on the correlation matrix in Appendix B, there is no perfect multicollinearity between the independent variables used. The high correlation between firm size and some liquidity variables together with the alternately signed coefficient of firm size across the regressions in Table III, suggests the regression suffers from multicollinearity. The Amihud illiquidity ratio and firms size have a significant correlation of ‐0.845, therefore the model experiences difficulties to separate the explanatory magnitude of the two variables. Since most liquidity variables are significant, I assume the liquidity variables have more explanatory power than firms size. In Section 6.2, I run robustness checks to investigate the impact of this possible multicollinearity on the Amihud illiquidity ratio coefficient.
5.2 Minimum Tick Size and SEO Underpricing
Table IV shows results of the regression to test the second hypothesis. Column 1 is the same regression as the first regression in Table III to show a regression with only controls for comparison purposes. Column 2 and 3 show results with the 1997 decimalization as an exogenous shock and Column 4 and 5 show results with the 2001 decimalization interactions. I find evidence of liquidity improvements expressed in SEO underpricing as result of reductions in tick size. While offer price rounding practices are still present after the regulatory change, the magnitude in which they affect SEO underpricing have significantly decreased after the decimalization in 1997 and to a smaller extent after the 2001 decimalization.
As can be seen in column 2, the effect of offer price rounding is present as the price coefficient is significantly negative signed. This effect is compensated for SEOs that took place after
the 1997 decimalization since the interaction of price with the decimalization dummy is significantly positive signed. The magnitude in which SEO underpricing depends on the stock price has on average reduced with 77.2% as a results of decreasing minimum tick size from 1/8 dollar to 1/16 dollar. When adding interactions with decimalization and tick<1/4 variables in column 3, we see similar reduced effects after decimalization in 1997 with limited significance.
Table IV ‐ Regression Results for SEO Underpricing on Decimalization Interactions
The table lists coefficients (standard errors) from OLS regressions of SEO underpricing on several pre‐stock liquidity characteristics and control variables. The panel is based on a sample of 5,419 SEOs on the NYSE and NASDAQ from 1990 to 2016 that meet sample restrictions described in section 4. The dependent variable is SEO underpricing defined as negative hundred times the return from the previous day's closing price to the offer price. . Decimalization1997 is equal to one if an issue takes place after June 24, 1997 (June 2, 1997) and is listed on the NYSE (NASDAQ) and zero otherwise. Decimalization2001 is equal to one if an issue takes place after January 29, 2001 (April 9, 2001) and is listed on the NYSE (NASDAQ) and zero otherwise. Firm size is calculated as the number of shares outstanding times the price on the day before the offer. Volatility is the standard deviation of return between 30 and 11 days before the offer. Relative offer size is measured as the number of shares issues divided by the number of shares outstanding on the day prior to the offer. CAR positive (negative) is equal to the sum of returns minus the returns of the CRSP value‐weighted index 5 days prior the offer when positive (negative) and zero otherwise. Price is equal to the closing price on the day prior to the offer. Tick<1/4 is a dummy variable equal to one if the decimal portion of the pre‐offer closing price is below $0.25 and zero otherwise. NASDAQ is a dummy variable equal to one if stock was issued on NASDAQ and zero when otherwise. %Intangible assets is calculated as percentage of the book value of intangible assets relative to the book value of total assets. Underwriter reputation is the reputation ranking according to J. Ritter of the lead underwriter during the time of the offer. Credit rating dummy is equal to one if the firm has a credit rating one year or less prior the offer and zero otherwise. Standard errors are in the parentheses and significant levels on the 10%, 5% and 1% are indicated by *,**,*** respectively. 1 2 3 4 5 Log(Price)*Decimalization1997 1.676*** 1.769* (0.303) (0.919) Tick<1/4*Decimalization1997 0.289 (2.771) Log(Price)*Tick<1/4*Decimalization1997 ‐0.118 (0.868) Decimalization1997 ‐5.120*** ‐5.347* (0.941) (2.904) Log(Price)*Decimalization2001 1.274*** 1.310** (0.336) (0.633) Tick<1/4*Decimalization2001 0.018 (1.862) Log(Price)*Tick<1/4*Decimalization2001 ‐0.046 (0.554) Decimalization2001 ‐3.711*** ‐3.755* (1.063) (2.038) Log(Price) ‐0.694*** ‐2.170*** ‐2.250** ‐1.687*** ‐1.704*** (0.238) (0.391) (0.897) (0.379) (0.590) Tick<1/4 0.438 0.346 0.094 0.414 0.424 (0.737) (0.739) (2.663) (0.730) (1.668) Log(Price)*Tick<1/4 ‐0.113 ‐0.081 0.020 ‐0.107 ‐0.087
