U
NIVERSITY OF
A
MSTERDAM
How do Investors React to Seasoned Equity Offerings
of High-Tech Firms?
Bachelor Thesis Economics and Business Specialization Finance and Organization
Merijn Tiebie, 10560815
Supervisor: Jan Lemmen
Statement of Originality
This document is written by Student [Merijn Tiebie] 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 study investigates short term abnormal returns surrounding the announcement day of seasoned equity offerings (SEOs) of U.S. companies that are listed on the NASDAQ or NYSE in the period 2002-2014. In particular, it is researched whether there is a difference between abnormal returns of high-tech firms and non-high-tech firms. A Welch’s t-test and an event study in combination with a cross-sectional regression analysis are performed in order to obtain the results. In accordance with previous literature, a negative price effect is observed after the announcement of a SEO. Also, limited evidence is found that this price effect is more favourable for high-tech firms. In addition, some evidence is found that the mitigated price effect is mainly caused by higher returns of large high-tech firms. Although this paper does not provide any substantial evidence, the theories and findings described in this paper can be used to initiate further research.
Keywords: Seasoned equity offerings, high-tech sector, information asymmetry, capital structure, JEL Classifications: G12, G14, G32
Contents
1. Introduction ... 5
2. Literature background and hypothesis development ... 6
2.1 Capital structure and the role of information asymmetry ... 6
2.2 Empirical evidence on seasoned equity offerings ... 8
2.3 Capital structure of high-tech firms ... 10
2.4 Seasoned equity offerings of high-tech firms ... 12
3. Research design ... 13
3.1 Methodology ... 13
3.2 Variables for cross-sectional regression analysis ... 16
3.3 Sample selection ... 19
4. Results ... 20
4.1 Descriptive statistics ... 20
4.2 Welch’s unequal variances t-test ... 23
4.3 Cross-sectional regression analysis ... 25
5. Conclusions ... 30
References ... 32
Appendix ... 34
List of tables Table 1: Predicted effects of variables on CAR ... 19
Table 2: Descriptive statistics ... 21
Table 3: Correlation matrix ... 21
Table 4: Average abnormal returns and T-statistics ... 22
Table 5: Relevant statistics for Welch’s t-test ... 24
Table 6: Output Welch’s t-test ... 24
Table 7: Relevant statistics for Welch’s t-test for the small-firm sample ... 24
Table 8: Relevant statistics for Welch’s t-test for the large-firm sample ... 25
Table 9: Regression results with dependent variable CAR(0,1) and estimation window (-250,-50) .... 27
Table 10: Regression results with dependent variable CAR(-2,2) and estimation window (-730,-10) . 28 Table 11: Regression results with dependent variable CAR(-1,1) and estimation window (-250,-50) . 34 Table 12: Regression results with dependent variable CAR(-2,2) and estimation window (-250,-50) . 35 Table 13: Regression results with dependent variable CAR(0,1) and estimation window (-730,-10) .. 36 Table 14: Regression results with dependent variable CAR(-1,1) and estimation window (-730,-10) . 37
1. Introduction
In their papers, Modigliani & Miller (1958) and Modigliani & Miller (1961) describe a perfect capital market in which external and internal financing are perfect substitutes. As a result, how to finance investments should not affect firm value. Since then, literature has developed new theories in which the idea of a perfect capital market has been rejected. Ross (1977) and Myers (1984) designed theories in which information asymmetry plays an important role in capital structure. Both theories claim that debt should be preferred over equity financing as it sends a positive signal to the market. As a result, Brealey & Myers (2000) state that investors will interpret SEOs as a negative signal. Empirical studies have confirmed that abnormal stock returns following SEOs are negative.
Carpenter & Petersen (2002) and Brealey, Leland & Pyle (1977) state that adverse selection and moral hazard problems will cause external finance to be more expensive. Arrow (1962) and Carpenter & Petersen (2002) claim that this is especially true for high-tech firms as their operations are risky. Therefore, high-tech firms are mainly dependent on internal financing. Because high-tech firms generally have less access to external finance than regular firms, the negative signal of SEOs may be mitigated for high-tech firms. If investors realize that high-tech firms are mainly dependent on external finance, issuances of equity may already be reflected in stock prices. Moreover, when investors know that investments are mainly financed by external finance, the issuance of equity can also be interpreted as a sign of investment opportunities and growth.
On the other hand, in a study of Huang & Tompkins (2010), it is found that stock prices of high-tech firms decrease more than stock prices for regular firms after SEOs. This can be explained by the fact that there is more information asymmetry involved when high-tech firms issue equity. Hence, it may lead to more price discounting and a larger price decrease for high-tech firms.
From the above, it can be concluded that the price effect of SEOs on high-tech firms is difficult to predict. Although research has been done on the effect of SEOs on the returns of firms in general, the high-tech sector in particular has not been investigated. In this paper, it will be researched whether stock returns after a SEO are different for high-tech firms compared with non-high-tech firms. Besides that, it is also investigated how the size of a high-tech firm influences the stock returns after a SEO.
An event study in combination with cross-sectional regression analysis and a Welch’s t-test will be used to acquire the results. Following upon the theories of Myers (1984) and Ross (1977) on information asymmetry, this paper focuses on the relation between
information asymmetry and the short term price effect of SEOs.
The results of this research will further attribute to the literature on SEOs and the empirical evidence will add value to the theoretical framework on information asymmetry. Moreover, conclusions will be drawn about the relation of being a high-tech firm and stock returns after SEOs. Besides that, in previous studies, the abnormal returns for event studies are calculated with the use of the market model. For this study, the Fama-French 3 factor model will be used (Fama & French, 1993) to see whether the results are consistent with the results of previous studies that used the market model. Also, the dataset consists of recent data and will therefore add value to the existing empirical evidence on SEOs.
In the next section, a literature background is presented in which capital structure, information asymmetry, empirical evidence on SEOs and the capital structure of high-tech firms will be discussed. The third section contains the methodology and the data sample, which will be followed by the presentation of the results. Finally, in the last section, conclusions will be drawn and suggestions for further research will be made.
2. Literature background and hypothesis development
2.1 Capital structure and the role of information asymmetry
Modigliani and Miller (1958) claimed that in perfect capital markets, arbitrage rules out the opportunity of benefitting of differences between the cost of equity and the cost of debt. As a result, capital structure cannot affect firm value. However, in reality equity and debt financing are never perfect substitutes. Debt financing has the advantage of the interest tax shield, but it also has the disadvantage of potential financial distress costs. Subsequently, firms have to find the perfect balance between the present value of the tax shield and the distress costs. In papers such as Kraus & Litzenberger (1973), Scott (1977) and Kim (1978), the balancing of the present value of the tax shield and financial distress costs is essential. Together, they build on what is called the static trade-off theory.
Also, Jensen & Meckling (1976) find that agency costs have to be taken into
consideration when determining the perfect capital structure. These costs arise when there are conflicts of interest between stakeholders of the firm. When these conflicts occur, suboptimal
decisions may be made with respect to investment opportunities. Furthermore, debt can be used to keep the interests of the firm and the manager aligned. If a manager of a firm owns stock of the firm, equity issuing will dilute his return on equity. Subsequently, it may cause reduced effort and excessive perks.
Besides the mentioned aspects that influence capital structure, Brealey, Leland & Pyle (1977), Myers (1984) and Ross (1977) describe the role that information asymmetry plays in capital structure issues. Information asymmetry occurs when the manager of a firm has more information about the firm than investors have. According to Brealey Leland & Pyle (1977), it cannot be expected from the manager of a firm that he is completely honest to a lender when the firm borrows cash. Exaggerating the positive characteristics of the firm may lead to a more beneficial outcome for the borrower. Similar to the debt market, managers of firms want to benefit from information asymmetry on the equity market. If the manager of a firm knows the value of his firm, he knows whether the shares of the firm are under or overvalued (Myers, 1984). When shares are overvalued, the firm will receive more cash for the shares than the actual value of the shares. Consequently, Myers (1984) claims that in the case of overvalued shares, managers will always issue equity even if the firm is facing zero net present value (NPV) investment opportunities. If the shares are undervalued and the firm acts on behalf of current shareholders, the firm will delay the equity offering until the information asymmetry has been reduced and the share price has risen (Myers & Majluf, 1984). If the firm waits and the share price has risen, less shares have to be issued in order to collect the amount of cash needed. As a result, the return on equity for existing shareholders does not dilute as much. Theoretically, managers of firms will therefore only issue equity when the share price reflects current firm value or when the share price overstates the current firm value. This leads to the adverse selection problem (Akerlof, 1970).
According to the adverse selection problem, only the firms that are overvalued will issue equity, whereas undervalued firms will wait until the share price rises. When equity is issued, investors will consider the possibility that shares are overvalued and the manager of the firm has some bad news that he withholds from the market. To compensate for this possibility, investors will discount the price they are willing to pay. Hence, the amount of cash that firms receive when they issue equity decreases, which makes equity financing costly. Myers & Majluf (1984) state that the price discounting may even cause firms not to invest in positive NPV’s. If the amount of cash that is discounted is larger than the NPV itself, the firm will decide not to invest. Information asymmetry also arises when firms choose to engage in debt financing. However, Myers (1984) argues that debt is not as sensitive to
private information as equity and the price discounting will therefore be mitigated. Obviously, the costs of information asymmetry can be completely avoided when investments are financed by retained earnings.
Based on this, Myers (1984) designs the pecking order theory. According to this theory, internal financing is preferred over debt financing and debt financing is preferred over equity financing. Empirical evidence has confirmed that only a small fraction of the external financing consists of equity financing.
Information asymmetry also affects capital structure in another way. This is explained by the incentive-signalling theory of Ross (1977). Suppose that information asymmetry exists and the manager of a firm wants to maximize shareholder value. When a manager faces an investment opportunity, the manager of a firm will only engage in debt financing when he is sure that the firm will be able to repay the debt in the future. He will not issue debt if the firm risks bankruptcy. Investors will interpret debt financing of an investment as a positive signal, because it suggests that the firm is able to repay the debt. On the other hand, managers will choose equity financing when debt financing is too risky. Ross (1977) argues that investors will therefore interpret equity financing as a negative signal. As a result, they will discount the price they are willing to pay for the shares.
The incentive-signalling theory (Ross, 1977) and the pecking order theory (Myers, 1984) both suggest that debt financing is preferred over equity financing. As debt financing sends positive signals and equity financing does not, equity financing should have a negative impact on share prices (Brealey & Myers, 2000). In corporate finance, this price decline is generally referred to as the ‘‘New Issues Puzzle’’.
Another factor that affects the price decline is share dilution. Asquith & Mullins (1984) investigate the relation between SEOs and share dilution. When firms issue new equity, their earnings do not necessarily increase. As a result, the earnings per share may decrease, which will result in lower share prices. However, in previous literature, the price effect of SEOs is mainly attributed to information asymmetry.
2.2 Empirical evidence on seasoned equity offerings
Asquith & Mullins (1984) find a negative price effect after the announcement of SEOs. From their sample, 80 % of the firms endured a price decrease. In addition, they find that firms which issue equity have positive abnormal returns prior to the announcement day of the SEO. This indicates that firms delay their equity offerings until shares are overvalued, so the
proceeds are higher than the actual value of the shares. Huang & Tompkins (2010) also investigate short-term stock returns after SEOs. They find an average three-day cumulative abnormal return of -3 % which confirms the results of Asquith & Mullins (1984). Bayless & Chaplinsky (1996) find negative abnormal returns as well. Furthermore, they observe that the negative abnormal returns are larger during periods of recession. Mikkelson & Partch (1986), Masulis & Korwar (1986), and Becker-Blease & Irani (2008) also find negative abnormal returns on the days surrounding the announcement of a SEO.
Jegadeesh (2000) matches firms which issued equity with comparable firms who did not issue equity in order to compare stock returns. He finds a negative relation between SEOs and long-run firm performance. Cai & Loughran (1998) investigate the impact of SEOs on Japanese firms and they also report that SEOs negatively influence firm performance in the long run. Masulis & Korwar (1986) find a negative relation as well and they observe that the negative stock returns are larger for industrial firms than for public utilities. Loughran & Ritter (1995), Affleck-Graves & Spiess (1995) and Mikkelson & Partch (1986) also find evidence that firms which issued equity underperformed the market in the long run.
Korajczyk, Lucas & McDonald (1991) claim that firms prefer to issue equity when information asymmetry is minimal. By doing so, the price shock will be reduced and a
possible negative price reaction will be mitigated. They state that information asymmetry is at its minimum short after earnings reports. Hence, they find evidence that price declines after SEOs increase in time since earnings reports.
Liu and Malatesta (2005) investigate the relation between credit ratings and the price reaction after the announcement of SEOs. They observe a more favourable price reaction for firms with credit ratings and this price reaction is even more favourable for firms with high credit ratings. They argue that the credit rating decreases information asymmetry which decreases the price discounting by investors.
Dasilas & Leventis (2013) investigate the impact of seasoned equity offerings on share prices in the short and long run in Greece. On the contrary with other literature, they find abnormal returns surrounding and at the day of announcement to be positive. Dasilas & Leventis (2013) state that this can be explained by the unpredicted increase in the Greek stock exchange in the years 1999-2000. They claim that the Greek firms took advantage of this increase by issuing equity on the overvalued exchange. Also, Dasilas & Leventis (2013) claim that the high ownership concentration may have reduced information asymmetry. The major stockholders are often actively involved in the firm and information asymmetry is hereby reduced. Without information asymmetry, the incentive-signalling theory (Ross, 1977) and
pecking order theory (Myers, 1984) are not valid and therefore share prices are not necessarily expected to decrease. Also, Dasilas & Leventis (2013) argue that the high involvement of the major stockholders guarantees that the interests of the firms and stockholders are aligned. As a result, corporate decisions such as SEOs will be in the interests of shareholders. With respect to the long run, Dasilas & Leventis (2013) find price increases preceding the announcement of the SEO and price drops after the announcement.
2.3 Capital structure of high-tech firms
Compared with regular firms, high-tech firms have a substantial different capital structure. In this section, a summary of literature will be given to explain this deflection. According to Myers (1984), firms with risky operations have higher chance of default. These firms will therefore have higher potential financial distress costs, which results in higher cost of debt.
Besides that, Myers (1984) states that firms with tangible assets, which have an active second-hand market will be able to easily sell their assets in case of bankruptcy. On the other hand, assets of firms with specialized intangible assets will lose value when they risk
bankruptcy. This will also increase financial distress costs and cost of debt. As high-tech firms are considered to have risky operations and specialized intangible assets, debt is expensive for high-tech firms. The expensive debt will limit the accessibility to the capital market.
Apart from the mentioned aspects, Carpenter & Petersen (2002) provide four other reasons why tech firms have difficulties attracting cash on the debt market. Firstly, high-tech firms have highly volatile returns and their projects have a relatively low probability of success. As a result, the probability of repayment of debt can be uncertain.
Secondly, Carpenter & Petersen (2002) claim that debt financing may lead to moral hazard problems. High-tech firms are more tempted to increase the riskiness of their projects than regular firms. The increase in riskiness does not provide additional returns for the lender in case of success, but it decreases the returns in case of failure.
Furthermore, they state that high-tech firms often have a lack of collateral. They do not own many operating assets as their main activities entail R&D.
Finally, Carpenter & Petersen (2002) argue that there is more information asymmetry in the high-tech sector than other industries. This is because highly specialized projects make it difficult for banks to assess the prospects of high-tech firms.
As the mentioned aspects apply specifically to small high-tech firms, Himmelberg & Petersen (1994) claim that small high-tech firms have more financial constraints than large high-tech firms. Small high-tech firms will therefore have more difficulties obtaining external finance.
Brealey, Leland & Pyle (1977) explain the role of information asymmetry in more detail. In the presence of information asymmetry, banks cannot distinguish between good and bad projects. Hence, the market will reflect the average quality of projects and the cost of capital for firms will equal to the average costs. This will lead to adverse selection and only firms investing in bad projects are willing to accept the average cost of capital. The adverse selection will decrease the average quality of projects even more and this will result in an adverse selection spiral.
With screening and signalling, banks and firms try to avoid the adverse selection problem. As screening of high-tech firms is more difficult, the adverse selection problem is larger in the high-tech sector. Berger & Udell (1998) state that the most efficient way of reducing asymmetric information is creating a long-term relationship between lender and borrower, as to increase the capability of lenders to assess the prospects of firms. However, Carpenter & Petersen (2002) argue that such a relationship is difficult to create for high-tech firms, because high-tech firms are often young and operating in a highly uncertain
environment.
Carpenter & Peterson (2002) state that equity financing has several advantages over debt financing for high-tech firms. Equity financing does not require collateral and it does not affect the present value of financial distress costs.
In their paper, Jensen & Meckling (1976) claim that agency costs will arise when the fraction of equity held by the manager dilutes. For young high-tech firms though, managers are to a great extent financially dependent on the performance of the firm and issuance of external equity will therefore not really affect the effort that the manager exerts.
From the above, it can be concluded that high-tech firms have difficulties attracting debt and that equity financing has multiple advantages over debt financing. Consequently, the advantages of equity and the disadvantages of debt result in lower debt/equity ratios in the high-tech sector compared with other industries. Therefore, the pecking order theory (Myers, 1984) and the incentive-signalling theory (Ross, 1977) might be implemented differently by high-tech firms (Hogan & Hutson, 2005 and Sjögren & Zackrisson, 2005).
2.4 Seasoned equity offerings of high-tech firms
When firms issue equity, investors account for the possibility that managers withhold some important negative information and shares are overvalued (Myers, 1984). Therefore, a SEO generally leads to a price decrease. As a result, Myers (1984) claims that firms want to avoid SEOs and will only issue equity as a last resort. On the other hand, SEOs are used more often by high-tech firms. This is because other sources of financing are difficult to obtain. Because high-tech firms are often young and have highly uncertain and volatile profits, their retained earnings may be exhausted. Besides that, the access to debt is limited, so issuing equity will be done more frequently. Hogan & Hutson (2005) and Sjögren & Zackrisson (2005) found evidence that high-tech firms prefer equity over debt.
If investors know that high-tech firms issue equity regularly, it may be that SEOs are already reflected in stock prices. Issuing equity will then not result in a price decrease. When SEOs are not reflected in stock prices it may be that investors do not consider the issuance of equity as a last resort, but rather look at it more positively. If one assumes that investors are aware that investments have to be financed mainly by SEOs, a SEO may then be interpreted as a sign of growth and investment opportunities.
As for the incentive-signalling theory (Ross, 1977), it is assumed that firms are able to send a positive signal to investors by using debt instead of a SEO. However, as attracting debt is difficult for high-tech firms, sending positive signals is difficult too. Subsequently, the issuance of equity for high-tech firms cannot necessarily be considered to be a bad signal and a SEO may therefore not result in a price decrease.
Denis (1994) and Jung, Kim & Stulz (1992) report that firms with investment and growth opportunities have a more favourable price reaction to the announcement of SEOs than mature firms. As high-tech firms are often facing growth and investment opportunities, it is feasible that high-tech firms will have a more favourable price reaction as well.
However, the effect of SEOs on stock prices of high-tech firms has not been
specifically addressed before. Although, Huang & Tompkins (2010) used a high-tech dummy in their research as a control variable. The dummy equals one for high-tech firms and zero for non-high-tech firms. In contrast with the theory just described, they found an insignificant but negative coefficient. In their paper, Huang & Tompkins (2010) do not further discuss this coefficient. Nevertheless, for this research, the outcome is relevant and must be seriously taken into consideration.
It can be that investors do not take into account the limited access of high-tech firms to external finance. The pecking-order theory (Myers, 1984) and the incentive-signalling theory (Ross, 1977) are both based on information asymmetry. Due to the information asymmetry, investors discount the share prices because of the possibility that managers withhold negative information. As Carpenter & Petersen (2002) explained that information asymmetry is larger in the high-tech sector, the discounting of stock prices may be larger in the high-tech sector as well. This may explain the negative coefficient that is found in the study of Huang &
Tompkins (2010).
Also, the theory that price decreases are mitigated for high-tech firms is based on the assumption that high-tech firms have limited access to the capital market (Carpenter & Petersen, 2002). Himmelberg & Petersen (1994) stated that external finance is more difficult to obtain for small high-tech firms. In this sense, it is likely that small high-tech firms will have a more positive price reaction after a SEO than large high-tech firms. Because the sample of Huang & Tompkins (2010) contained many firms from the NYSE and the AMEX, it indicates that the sample consisted of many large firms. As a result, it might be that the assumption of limited access to capital markets was not valid for the research of Huang & Tompkins (2010).
On the other hand, information asymmetry is also larger for small high-tech firms than for large high-tech firms. As a result, it may be the case that investors discount share prices more for small high-tech firms than for large high-tech firms.
Based on the discussed literature, it will be investigated how being a high-tech firm affects the price reaction on SEOs and how the size of a high-tech firm influences this price effect.
3. Research design
3.1 Methodology
In order to investigate stock returns surrounding the day of the announcement of a SEO, the cumulative abnormal returns (CARs) are used. The set-up is like an event study. The events are selected according to the sample selection that will be described in the section 3.2. The event dates are identified as the announcement day of the SEO. Event windows of two days (Dasilas & Leventis 2013, Kim & Purnanandam, 2014), three days (Huang & Tompkins, 2010 and Dasilas & Leventis, 2013) and five days (Kim & Purnanandam, 2014) are used to
calculate the cumulative abnormal returns. The event window of two days consists of the announcement day and the day after the announcement day (0,1). The event window of three days consists of the day before the announcement day, the announcement day and the day after the announcement day (-1,1). The event window of five days depicts the five days surrounding the announcement day (-2,2). Before calculating the CARs, the normal returns must be calculated. These are calculated during the estimation period, which takes place before the event window. Based on a paper of MacKinlay (1997), an estimation period of 200 days 250,-50) is used to find the normal returns. Also, an estimation period of 720 days (-730, -10) is used to see whether the estimation period influences the results.
To calculate the normal or expected returns, the Fama-French 3-factor model is used (Fama & French, 1993). Thus, the expected returns are equal to:
(1) 𝐸(𝑅𝑖) = 𝛼 + 𝛽𝑖(𝑀𝐾𝑇𝑃𝑅) + 𝛽𝑖(𝑆𝑀𝐵) + 𝛽𝑖(𝐻𝑀𝐿) + 𝜀
where MKTPR stands for market premium, SMB is the small minus big factor and HML is the high minus low factor as is explained in Fama & French, 1993.
To acquire the abnormal returns (ARj,t,), for every event by means of the 3-factor model, the
difference between the actual return and the estimated normal return calculated in (1) is established. This is depicted in the succeeding expression:
(2) 𝐴𝑅𝑖,𝑡 = 𝑅𝑖,𝑡 − 𝐸(𝑅𝑖,𝑡)
where 𝑅𝑖,𝑡 is the return of a firm (𝑖) on day (𝑡) of the event window and 𝐸(𝑅𝑖,𝑡) is the
expected return as is formulated in (1) of a firm (𝑖) on day (𝑡). The formula for the CARs is presented below and displayed for the three different event windows:
(3) 𝐶𝐴𝑅 = ∑ 𝐴𝑅𝑖,𝑡 +1 0 𝐶𝐴𝑅 = ∑ 𝐴𝑅𝑖,𝑡 +1 −1 𝐶𝐴𝑅 = ∑ 𝐴𝑅𝑖,𝑡 +2 −2
The CARs are calculated with the Event Study Suite of WRDS. With the cumulative
abnormal returns, a cross-sectional regression can be done to see what factors other than the Fama-French model influence the cumulative abnormal returns. The cross-sectional
regression will be performed in Stata by using OLS regressions with robust standard errors. The robust standard errors are used to correct for heteroscedasticity. As three different event windows and two different estimation windows are used, six regressions are performed in Stata. The results of these regressions will be presented in section 4.3.
Besides the regression analysis, a Welch’s t-test is conducted to compare the
cumulative average abnormal return (CAAR) of the high-tech sample with the CAAR from the non-high-tech sample. The Welch’s t-test is a test that is used to compare the means of two different samples. Similar to the student’s t-test, Welch’s t-test assumes that both samples are normally distributed. Different from the student’s t-test, the Welch’s t-test accounts for unequal variances between the two samples. Therefore, this test is appropriate to compare the CAARs of high-tech and non-high-tech firms. The sample is divided into two groups and with the help of the following formula, the t-statistic is calculated.
(4) 𝑡 = 𝑋̅ 1− 𝑋̅ 2 √𝑆12
𝑁1+
𝑆22
𝑁2
𝑋̅ 1, 𝑆1and 𝑁1 refer to the average, the standard deviation and the sample size of the non-high-tech sample, whereas 𝑋̅ 2, 𝑆2and 𝑁2 represent the average, the variance and the sample size of the high-tech sample. For this test, the observations of the estimation window (-250,-50) are used. The CAARs are calculated with the use of CARs(-1,1). The degrees of freedom (𝜗) for this test statistic are determined with the help of the subsequent equation:
(5) 𝜗 ≈ (𝑆1 2 𝑁1+ 𝑆22 𝑁2) 2 𝑆14 𝑁12𝑣1+ 𝑆24 𝑁22𝑣2
where 𝑣1 is equal to 𝑁1− 1 and 𝑣2 = 𝑁2− 1.
Also, the effect of size on the CAARs is investigated with the Welch’s t-test. Two different samples are created by dividing the main sample into two groups by using the median of the variable market capitalization. Thus, the t-test will be performed for the whole sample, the small-firm sample and the large-firm sample.
3.2 Variables for cross-sectional regression analysis
To investigate the price shock of SEOs of high-tech firms, a dummy for high-tech firms is included. The dummy equals one for high-tech firms and zero for all other firms. As is shown in section 2.4, it is difficult to predict how investors respond to SEOs of high-tech firms. Hence, the beta of the dummy is expected to be unequal to zero.
H1: β(HIGHTECH) ≠ 0
Huang & Tompkins (2010) argue that market capitalization is an important indicator of information asymmetry. They state that analysts focus more on large firms. Hence, large firms have to be more transparent which will decrease information asymmetry. Less
information asymmetry will decrease the effect of a SEO on stock prices. Therefore, the variable MKTCAP is expected to have a positive influence on the CARs. The market
capitalization is obtained from the Zephyr database and is defined as the market capitalization during the first year available after the SEO. Based on Huang & Tompkins (2010) and Dasilas & Leventis (2013), the natural logarithm is taken from the market capitalization and included in the regression model.
In section 2.4, two theories are presented that can explain how the announcement of a SEO may have a different impact on small high-tech firms compared with large high-tech firms. To test whether one of these theories is in accordance with the data, an interaction term is included that multiplies HIGHTECH with LNMKTCAP.
H2: β(HIGHTECH × LNMKTCAP ≠ 0
To summarize, investors can follow two different theories. It is possible that they take into consideration the fact that high-tech firms are to a certain extent dependent on equity. This will lead to a reduced price effect when a SEO is announced. In addition, small high-tech firms have even more difficulties obtaining external finance and are therefore more dependent on equity. As a consequence, when the price effect of the announcement is more favourable for high-tech firms than for non-high-tech firms, the price effect will be even more favourable for small high-tech firms. In other words, if being a high-tech firm has a positive effect on abnormal returns surrounding the day of the announcement of a SEO, this effect is mainly caused by high returns of small high-tech firms.
On the other hand, it is possible that investors focus on information asymmetry which will result in a price decline that is worse than the price decline of non-high-tech firms. Since the information asymmetry is larger for small-high-tech firms, the price decline would be larger for small-high-tech firms. If being a high-tech firm has a negative influence on abnormal returns surrounding the day of the announcement of a SEO, this effect will be mainly caused by low returns of small high-tech firms.
H3: in the model without interaction term: β(HIGHTECH) > 0 and when the interaction term is included: β(HIGHTECH × LNMKTCAP) < 0, or in the model without interaction term: β(HIGHTECH) < 0 and when the interaction term is included: β(HIGHTECH ×
LNMKTCAP) > 0.
Based on Asquith & Mullins (1984), Mikkelson & Partch (1986), Bayless & Chaplinsky (1996), Huang & Tompkins (2010) and Dasilas & Leventis (2013), a control variable for the relative size of the offer is included. This is measured by the total deal value divided by the market capitalization before the offering and is obtained from the Zephyr database. Myers (1984) claimed that firms will issue shares when shares are overvalued. Huang & Tompkins (2010) state that overvalued firms will therefore issue more shares. A large offer is therefore expected to have a negative impact on the CAR. This is confirmed by Asquith & Mullins (1984) and Bayless & Chaplinsky (1996) who found that the size of the offer has a negative and significant impact on the returns surrounding the announcement date of a SEO.
Also a variable for the price run-up of the stock is included in the regression (Asquith & Mullins (1984), Masulis & Korwar (1986), Huang & Tomkpins (2010) and Dasilas & Leventis (2013)). The variable RUNUP is calculated with the Event Study Suite of WRDS and is equal to the cumulative total return over the three months preceding the announcement day. As is explained by Myers & Majluf (1984), firms prefer to wait with issuing shares until the share price reflects or overstates firm value. Asquith & Mullins (1984) find that for firms which issue equity after a price run-up, short term stock returns are more favourable than for firms who did not have a price run-up. On the other hand, Masulis & Korwar (1986) find a negative coefficient for price run-up. They state that SEOs of firms with large price gains preceding the SEO announcement were more of a surprise to investors. Hence, a SEO could then cause a larger negative price shock. Based on this, it is difficult to predict what effect the variable RUNUP will have on the CAR.
Furthermore, a dummy CRISIS is included for SEOs that took place during the most severe period of the financial crisis which is from 1 July 2007 until 1 July 2009. Bayless & Chaplinksy (1996) found that short term stock returns after SEOs are lower during cold periods. Consequently, the financial crisis must be taken into consideration and it is expected that the dummy CRISIS will affect the CAR negatively.
Firms who pay dividend are generally more transparent than firms that do not pay dividends (Booth & Chang, 2011). Miller & Rock (1985) developed a model in which firms are able to inform the market by paying out dividends. Consequently, the information asymmetry for dividend paying firms is smaller than for non-paying firms. As a result, dividend paying firms are expected to have smaller price declines surrounding the
announcement day of a SEO than non-paying firms. To incorporate the effect of dividends on the short term stock returns after a SEO, a dummy DIV is included. The dummy equals one for firms that paid dividend in the year preceding the SEO and zero for non-paying firms. The information concerning the payment of dividends is obtained from the COMPUSTAT
database and the NASDAQ database. High-tech firms often have different dividend policies than regular firms. To a great extent, they try to retain or reinvest their earnings and hereby choose to not pay dividends. When they do pay dividends this may be interpreted as a sign of weakness as it could mean that investment opportunities are scarce. Therefore, it would have been interesting to include an interaction term that multiplies the variables HIGHTECH and DIV. However, since the sample only contains seven high-tech firms that paid dividend in the year preceding the SEO, the variable is not included.
Finally, a dummy ISPRY is included for firms who issued equity during the year preceding the SEO (Masulis & Korwar, 1986). According to Masulis & Korwar, SEOs of companies who frequently issue equity are to a certain extent predictable. As a result, the market anticipates that the company has to issue equity regularly and therefore the announcement of the SEO will have less effect on share prices. Subsequently, ISPRY is expected to have a positive effect on the CAR. The cross-sectional regression is now equal to:
(6) 𝐶𝐴𝑅 = 𝛼 + 𝛽1(𝐿𝑁𝑀𝐾𝑇𝐶𝐴𝑃) + 𝛽2(𝑆𝐼𝑍𝐸) + 𝛽3(𝑅𝑈𝑁𝑈𝑃) + 𝛽4(𝐻𝐼𝐺𝐻𝑇𝐸𝐶𝐻)
+ 𝛽5(𝐻𝐼𝐺𝐻𝑇𝐸𝐶𝐻 × 𝐿𝑁𝑀𝐾𝑇𝐶𝐴𝑃) + 𝛽5(𝐶𝑅𝐼𝑆𝐼𝑆) + 𝛽6(𝐷𝐼𝑉) + 𝛽7(𝐼𝑆𝑃𝑅𝑌 )
The predicted effects of the variables on the CAR are summarized in table 1.
Table 1: Predicted effects of variables on CAR
MKTCAP SIZE RUNUP HIGHTECH MKTHIGH CRISIS DIV ISPRY
Prediction + - ? ? ? - + +
3.3 Sample selection
The sample of this study consists of firms from the United States that are listed on the NASDAQ National Market or the New York Stock Exchange. The time period covers the period from 1 January 2002 until 31 December 2014. The data that is used comes from the Zephyr database. In this database, 4723 equity issues are found. From this selection, IPOs, warrants exercises, convertible debt/shares, private placements, secondary offerings and rights offerings (Masulis & Korwar, 1986) are deleted.
Also all utility offerings (Loughran & Ritter, 1995) and issuances of companies in the financial sector (Cai & Loughran, 1998) are excluded from the sample. With respect to the utility companies, the companies with a SIC code between 491 and 494 are removed from the sample. For the financial companies, companies with a SIC code between 60 and 67 are excluded from the sample. These industries are excluded from the sample, because these firms tend to have completely different capital structures than regular companies.
In addition, all issuances are deleted of which the market capitalization, size of the offer or stock price information is unknown. After this, 502 SEOs are left, of which 118 offerings are from a high-tech firm. From the remaining sample, equity issuances which coincide with other corporate events such as acquisitions are excluded from the sample. Besides that, observations which have a standard deviation that exceeds three times the standard deviation of either the variable RUNUP or CAR((-2,2), (-730, -10)) are removed from the sample.
The final sample consists of 471 firms, of which 110 equity offerings are made by a high-tech firm. For high-tech firms, the same classifications are used as for the research of Carpenter & Petersen (2002). Firms from the following industries are considered to be a high-tech firm: drugs and medicinal (SIC 283), computer and office equipment (SIC 357),
industrial measuring instruments (SIC 382), and surgical instruments (384). The 471 SEOs are issued by 380 different companies.
As is explained in section 3.1, the CARs are calculated with the use of two different estimation periods. By using the (-250, -50) estimation period, 14 observations are lost when using the WRDS Event Study Suite. Hence, the sample for this estimation period consists of 457 SEOs, of which 108 offerings are made by a high-tech firm. There are no significant differences between the descriptive statistics of the two different samples. As the (-250, -50) estimation window is the most common estimation window, only the descriptive statistics of this sample will be presented.
Figure 1: Seasoned equity offerings per year
From figure 1, it can be observed that the SEOs are mainly concentrated in 2012, 2013 and 2014. Also, the amount of SEOs increased sharply in 2009 at the end of the financial crisis. Consistent with the beliefs of Myers & Majluf (1984), companies postponed their SEOs until shares prices represented the actual value of the firm or overstated the actual value of the firm.
4. Results
4.1 Descriptive statistics
From the sample, 54 firms issued equity in the year preceding the equity issuance. There were 99 equity issuances made by dividend paying companies and 53 seasoned equity offerings were made between the period 1 July 2007 until 1 July 2009. In table 2, the descriptive statistics of the dependent variables and the other control variables are tabulated. The excess
0 10 20 30 40 50 60 70 80 90 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
kurtosis is measured as the deviation from the kurtosis under a normal distribution. The skewness of the CARs indicate that the CARs are approximately symmetric distributed.
Table 2: Descriptive statistics
Average Median St. Dev. Minimum Maximum Skewness Excess kurtosis
CAR (0,1) -3.90% -3.54% 7.35% -32.73% 29.37% -0.26 2.19 CAR(-1,1) -4.19% -3.83% 7.88% -29.01% 37.69% 0.08 2.36 CAR(-2,2) -4.08% -3.96% 9.06% -32.75% 30.85% 0 1.40 MKTCAP * (€1000) 2182208 701780 4566738 7727 36054327 4.02 18.36 SIZE 15.85% 13.86% 9.38% 0.50% 61.02% 1.58 3.91 RUNUP 31.89% 22.85% 46.76% -72.64% 217.26% 1.23 1.96
A correlation matrix of the relevant variables is presented below. From the matrix, it can be concluded that there are no variables that are highly correlated. Therefore, the
correlation of the variables will not bias the results of the regression analysis. The variables SIZE and MKTCAP have the highest correlation. This is no surprise as the relative size of the offer is dependent on the size of the firm. The negative coefficient can be attributed to the fact that a SEO of a certain amount is relatively smaller for a large firm than for a small firm.
Table 3: Correlation matrix
CAR (-1,1) MKTCAP SIZE RUNUP HIGHTECH CRISIS DIV ISPRY CAR (-1,1) 1.000 MKTCAP -0.032 1.000 SIZE 0.071 -0.245 1.000 RUNUP -0.111 -0.028 0.035 1.000 HIGHTECH 0.053 -0.004 -0.021 -0.006 1.000 CRISIS -0.038 0.015 -0.122 -0.131 -0.025 1.000 DIV -0.018 0.239 -0.168 -0.125 -0.205 0.092 1.000 ISPRY -0.056 0.023 0.140 -0.031 -0.044 -0.048 0.071 1.000
The average abnormal returns (AARs) on the days surrounding the event are presented in table 4 and figure 2. To evaluate the level of statistical significance of the average abnormal returns, a parametric t-test is included. The fundamental assumption of independent and
identical distributed abnormal returns, following a normal distribution with mean 0, lies at the heart of the trustworthiness of this test. The formula of the t-statistic equals:
(7) 𝑡𝐴𝐴𝑅𝑡 = √𝑁 ∗
𝐴𝐴𝑅𝑡 𝑆𝐴𝐴𝑅𝑡
Table 4: Average abnormal returns and T-statistics
Day AAR T-Statistic Day AAR T-Statistic
-15 0.14% 0.52 1 -2.49% *** -8.42 -14 0.33% ** 1.97 2 0.02% 0.14 -13 0.02% 0.12 3 -0.11% -0.72 -12 0.30% 0.99 4 -0.23% * -1.46 -11 0.05% 0.26 5 -0.19% * -1.36 -10 0.32% 1.07 6 0.16% 1.15 -9 -0.04% -0.30 7 0.20% * 1.33 -8 0.58% ** 2.00 8 0.11% 0.52 -7 0.13% 0.76 9 -0.02% -0.15 -6 0.35% * 1.63 10 0.04% 0.30 -5 0.34% * 1.37 11 0.01% 0.07 -4 0.23% 1.12 12 -0.02% -0.19 -3 -0.32% ** -2.05 13 -0.07% -0.50 -2 0.09% 0.54 14 -0.13% -0.87 -1 -0.25% * -1.38 15 -0.27% *** -2.33 0 -1.45%*** -6.66 Total -2.17% ** -2.05
Figure 2: Average abnormal returns (-15,15)
As table 4 shows, the average abnormal returns on day 0 and day 1 are significant at a 1% significant level. This means that the SEOs had a significant negative impact on the stock returns on the announcement day and the day after the announcement day. This is consistent with previous literature such as Asquith & Mullins (1984), Mikkelson & Partch (1986), Masulis & Korwar, Bayless & Chaplinksy (1995), Becker-Blease & Irani (2008) and Huang & Tompkins (2010).
The fact that the abnormal returns on day 1 are lower than the abnormal returns on day 0 can be explained by the fact that markets need some time to adjust. Also, it can be that the announcement of a SEO occurs in non-trading hours. In this case, the market cannot react on day 0 and will therefore respond to the announcement on day 1 (Becker-Blease & Irani, 2008). The AARs before the announcement of the SEO are mainly positive. This is consistent with the average RUNUP (31.89%) that can be observed from table 2. The market seems to adjust to the SEO announcement on day 0 and day 1. Although, after these days, significant negative abnormal returns are also observed on day 4,5, and 15.
4.2 Welch’s unequal variances t-test
In table 5, the relevant statistics for the Welch’s t-test are summarized. As the CAAR from the nonhightech sample is equal to 4.42% and the CAAR from the hightech sample equals -3.44%, it will be tested whether the difference of the CAAR is smaller than zero. This is represented by the succeeding hypothesis:
-3.00% -2.50% -2.00% -1.50% -1.00% -0.50% 0.00% 0.50% 1.00% -15 -14 -13 -12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
H0: CAAR non-high-tech sample – CAAR high-tech sample = 0 H1: CAAR non-high-tech sample – CAAR high-tech sample < 0
Table 5: Relevant statistics for Welch’s t-test
Non-High-Tech High-Tech
N 349 108
X̅/CAAR -4.42% -3.44%
St. Dev. 7.44% 9.15%
In table 6, the results of the Welch’s t-test are presented. The results do not indicate that the difference is significant. Nevertheless, the t-statistic has a p-value of 0.15, which is not far from the 10% significance level.
Table 6: Output Welch’s t-test
Difference X̅/CAAR -0.98%
t -1.02
ϑ 154.20
P-value 0.15
Also the t-test will be applied to a small-firm and a large-firm sample. The median of the variable MKTCAP divides the sample into two equal groups. From table 7, it can be
concluded that the CAARs of the small non-high-tech firms and small high-tech firms do not differ that much. The difference equals 0.09 %, which results in a t-statistic of -0.05.
Obviously, this difference is not close to significance.
Table 7: Relevant statistics for Welch’s t-test for the small-firm sample
Non-High-Tech High-tech
N 168 61
CAAR -3.98% -3.89%
S 6.88% 10.87%
In table 8, the relevant statistics for Welch’s t-test for the large-firm sample are tabulated. The difference of the CAARs is substantially larger in this case. The difference between the
CAARs is equal to -2%. The t-statistic of this difference equals -1.83, which is significant at a 5% significance level.
Table 8: Relevant statistics for Welch’s t-test for the large-firm sample
Non-high-tech High-tech
N 181 47
CAAR -4.84% -2.84%
S 7.93% 6.32%
The results of the Welch’s t-test suggest that there is a moderate difference between the CAARs of the non-high-tech and the high-tech sample. The high-tech sample has less
negative abnormal returns than the non-high-tech sample. When the t-test is corrected for size effects, it becomes clear that this difference comes from the large-firm sample. The abnormal returns of large high-tech firms are significantly higher than the abnormal returns of large non-high-tech firms
4.3 Cross-sectional regression analysis
As is explained in section 3.1, three different event windows and two different estimation windows are used to obtain regression results. Hence, six different models are acquired. Regressions with the inclusion of outliers are also performed, but these lead to less significant results. Using different estimation windows resulted in substantial differences between the models. As a result, it can be concluded that the sign and significance of some coefficients were seriously influenced by the estimation window. The different event windows had an impact on the significance of the coefficients but it did not lead to any sign changes.
Therefore, of each estimation window the model with the most significant results is presented. The other models can be found in the appendix. Nevertheless, conclusions will be drawn based on the results of all models.
In table 9, it can be observed that the HIGHTECH dummy has a positive but
insignificant coefficient. However, when the interaction term LNMKTCAP * HIGHTECH is included, the coefficient of the HIGHTECH dummy becomes negative and significant. The interaction term has a positive sign and is significant as well. This tendency can also be observed for the other event windows (see appendix). Although, for these windows, the coefficients are less significant. The exact effect of HIGHTECH on CAR(0,1) observed in
table 9 is equal to -0.0961 + 0.0077LN(MKTCAP). Solving for MKTCAP yields 263160.5. This means that the effect of being a high-tech firm is negative for firms with a market capitalization smaller than this amount, but positive for firms with a market capitalization larger than this amount. In table 10, a positive coefficient for the HIGHTECH dummy is found as well and it is also significant. In this case, the introduction of the interaction term does not lead to more significant results. The positive coefficient for HIGHTECH is not significant for the other event windows, but the sign of the coefficient remains the same.
The variable LNMKTCAP has a significant and positive impact on the CARs when the estimation window is equal to (-730,-10). This is in accordance with the expectation formulated in section 3.2. The coefficient is less significant for the other event windows. On the other hand, in the case of the estimation window of (-250,-50), the coefficient becomes very insignificant. Theoretically, it is feasible that this variable has a positive impact on stock returns surrounding the announcement day of a SEO. The results of these regressions support that theory to some extent. Just as the LNMKTCAP coefficient differs across the different regressions, empirical evidence of previous literature on this matter is also inconsistent. For example, Dasilas & Leventis (2013) and Huang & Tompkins find negative coefficients for the variable LNMKTCAP, whereas Bayless & Chaplinsky (1996) find that size has a positive influence on the CAR.
Table 9: Regression results with dependent variable CAR(0,1) and estimation window (-250,-50)
Variable Model 1 Model 2 Model 3 Model 4
CONSTANT -0.0298 -0.0337 -0.0050 -0.0123 (-0.92) (-1.04) (-0.13) (-0.31) LNMKTCAP -0.0006 -0.0004 -0.0025 -0.0017 (-0.27) (-0.19) (-0.97) (-0.58) SIZE 0.0267 0.0283 0.0269 0.0288 (0.72) (0.77) (0.74) (0.77) RUNUP -0.0193*** -0.0193*** -0.0198*** -0.0216*** (-2.65) (-2.64) (-2.70) (-2.88) HIGHTECH 0.0055 -0.0961** -0.0910 (0.65) (-1.73) (-1.54) LNMKTHIGH 0.0077** 0.0073* (1.89) (1.65) CRISIS -0.0140 (-1.37) DIV -0.0030 (-0.31) ISPRY -0.0105 (-0.89) N 457 457 457 457 𝑅2 0.0161 0.0171 0.0228 0.0285 𝑅𝑎𝑑𝑗2 0.0095 0.0084 0.0120 0.0112
Note: the regression is conducted by using OLS and the use of robust standard errors. LNMKTCAP is the natural logarithm of the market capitalization, SIZE is the relative size of the offer, RUNUP is the price movement during the three months preceding the SEO announcement, HIGHTECH is a dummy which equals one for high-tech companies, LNMKTHIGH is an interaction variable that multiplies LNMKTCAP with HIGHTECH, DIV is a dummy that equals one for dividend paying companies and ISPRY is a dummy that equals one for SEOs which took place within a year of another equity offering.
The t-statistic is given in parenthesis beneath the corresponding coefficient. *, **, *** reflects significance at respectively a 10%, 5% and 1% significance level.
Table 10: Regression results with dependent variable CAR(-2,2) and estimation window (-730,-10)
Note: the regression is conducted by using OLS and the use of robust standard errors. LNMKTCAP is the natural logarithm of the market capitalization, SIZE is the relative size of the offer, RUNUP is the price movement during the three months preceding the SEO announcement, HIGHTECH is a dummy which equals one for high-tech companies, LNMKTHIGH is an interaction variable that multiplies LNMKTCAP with HIGHTECH, DIV is a dummy that equals one for dividend paying companies and ISPRY is a dummy that equals one for SEOs which took place within a year of another equity offering.
The t-statistic is given in parenthesis beneath the corresponding coefficient. *, **, *** reflects significance at respectively a 10%, 5% and 1% significance level.
Variable Model 1 Model 2 Model 3 Model 4 Model 5
CONSTANT -0.1289*** -0.1434*** -0.1270*** -0.1486*** -0.1497*** (-3.32) (-3.62) (-3.49) (-3.74) (-3.72) LNMKTCAP 0.0068*** 0.0075*** 0.0063** 0.0082*** 0.0083*** (2.62) (2.84) (2.17) (3.08) (3.01) SIZE -0.0532 -0.0476 -0.0477 -0.0517 -0.0602 (-1.11) (-0.99) (-0.99) (-1.07) (-1.23) RUNUP 0.0175 0.0174 0.0170 0.0154 0.0152 (1.6) (1.59) (1.56) (1.39) (1.36) HIGHTECH 0.0209** -0.0371 0.0206* 0.0202* (1.98) (-0.48) (1.94) (1.86) LNMKTHIGH 0.0044 (0.77) CRISIS -0.0221* -0.0215* (-1.75) (1.7) DIV -0.0047 (-0.51) ISPRY 0.0135 (0.96) N 471 471 471 471 471 𝑅2 0.0327 0.0422 0.0435 0.0482 0.0507 𝑅𝐴𝑑𝑗2 0.0265 0.0340 0.0332 0.0380 0.0363
When the estimation window is (-250,-50), the variable RUNUP has a negative coefficient in all regressions and it is statistically significant for event windows (0,1) and 1,1). On the other hand, the coefficient is positive when the estimation window is equal to (-730,-10). As the average price run-up (31.89 %) is strongly positive, it suggests that firms that issued equity outperformed the market in the three months preceding the announcement of the SEO. As the Carhart momentum factor (Carhart, 1997) is not included in the calculations of the expected returns, the expected returns calculated in (1) are actually too low, resulting in abnormal returns (2) that are too high. Hence, the variable RUNUP has a more positive influence on the CAR when the estimation window is equal to (-730,-10) as this time frame has more overlap with the variable RUNUP than the estimation window (-250, -50). This explains how the sign of the coefficient RUNUP can differ between the two estimation windows. The estimation window with the smallest overlap is thus the most trustworthy. Overall, it seems that the negative coefficient observed in table 9 is the most reliable and that the outcome is in line with the coefficient observed in Masulis & Korwar (1986).
The coefficient of SIZE also changes when the estimation window changes. It once more emphasizes the extent to which the estimation window influences the results. However, because of the insignificance of the SIZE coefficients, it can be concluded that SIZE does not affect the CARs.
The variable CRISIS is negative in all regression models and in some models it has a significant coefficient. This is in accordance with previous literature and it supports the theory of Bayless & Chaplinsky (1996). The coefficients DIV and ISPRY are neither significant, nor jointly significant in any of the models. Consequently, they do not add any value to the models.
Overall, the estimation window (-730,-10) yields a higher adjusted 𝑅2 than the estimation window (-250,-10). Model 4 from table 10 is the model that fits the data best. Hence, the coefficient for HIGHTECH in this model seems to be the most reliable. However, the size effect for high-tech firms that can be observed from the Welch’s t-test and table 10 is not observed in this model.
5. Conclusions
The predominant view about the price effect of SEO announcements is that the announcement negatively influences stock returns. In this study, it is investigated whether this price effect is different for high-tech firms. Besides that, the Fama-French model is used to calculate the normal returns in order to find out whether this model provides different results than the commonly used market model.
As is observed in table 4, the abnormal returns that are found on day 0 and day 1 are significantly below zero. Subsequently, it can be concluded that the negative price effect of SEO announcements is also observed in the sample of this study.
In addition, there is some evidence that shows that the price effect of the
announcement of SEOs is mitigated for high-tech firms. Welch’s t-test shows that the CAAR for non-high-tech firms is almost 1 % lower than for high-tech firms. However, with a p-value of approximately 0.15, this difference is not significant. The regression results confirm the results of the t-test. Without the interaction term, the high-tech dummy is positive in all models and significant in one model. Based on the regression results and Welch’s t-test, it can be concluded that there is limited evidence that investors react more positively to
announcements of SEOs of high-tech firms.
Welch’s t-test and the inclusion of the interaction term HIGHTECH * LNMKTCAP show that the positive effect of HIGHTECH is mainly caused by high abnormal returns of large high-tech firms. Though, the interaction term is only significant in two models. To summarize, some evidence is found that shows how the positive effect of HIGHTECH can be mainly attributed to the high abnormal returns of large high-tech firms. This is inconsistent with the prediction that a possible positive effect of HIGHTECH on the CARs will be primarily caused by high returns of small high-tech firms.
This study shows that the negative price effect of SEOs, which is observed in many previous papers, is also found in recent data. Furthermore, whereas previous literature calculates normal returns with the market model, this study shows that the negative price effect also exists when normal returns are calculated with the Fama-French 3 factor model. The findings concerning the price effect of SEOs in the high-tech sector has not created any substantial evidence. Nonetheless, the weak evidence that is found is in line with Denis (1994) and Jung, Kim & Stulz (1992).
It would be interesting to use a dataset consisting of less mature firms. The theories described in this paper are based on information asymmetry and limited access to debt markets for high-tech firms. The firms that are used in this study are listed on the NASDAQ or the NYSE. Logically, these firms are all mature whereas the theories described are more appropriate for young and small high-tech firms.
In addition, the overlap between the estimation window and the RUNUP coefficient lead to inconsistent results on the coefficient RUNUP. This problem can be overcome by changing the estimation window or the calculation of the variable RUNUP. Another possibility is to use the Carhart (1997) momentum factor to calculate normal returns. By doing this, the variable RUNUP could be eliminated from the regression analysis.
Also, the adjusted 𝑅2 of the regressions could be increased by including more
corporate governance variables that can be found in Huang & Tompkins (2010), Dasilas & Leventis (2013) and Kim & Purnanandam (2014). By including these variables, it is likely that more of the variation of the dependent variable can be explained, which will increase the reliability of the results.
Besides that, as high-tech firms often have high R&D expenses and low earnings. Including these variables in the regression models could help explaining the relation between high-tech firms and the price effect of SEOs. Furthermore, it would be interesting to add a variable to the regression model that can be used as a proxy for information asymmetry. For example, the time between the SEO and IPO can be an indication of information asymmetry, as firms that are listed on an exchange for a longer period are better known by the market.
Also, from table 4 it can be concluded that 4, 5 and 15 days after the announcement of the SEO the AARs were negative and significant. This suggests that the used event windows might not have incorporated all effects of the announcement of the SEO. Hence, an event window of (-5,5) or (-15, 15) may have resulted in different results. Due to time constraints, it was not possible to implement any of the mentioned solutions. Nonetheless, further research could focus on these matters to find out whether these suggestions improve the existing research.
Finally, it is remarkable that the use of different estimation windows and different event windows resulted in substantial differences across the models. However, in many papers, the estimation window is not reported or not actively discussed. As this paper shows that the estimation window is an important factor when calculating abnormal returns, further research should be more transparent on this matter. Moreover, experimenting with different estimation windows would increase the reliability of regression results.
References
Affleck-Graves, J. & Spiess, K. (1994). Underperformance in Long-Run Stock Returns Following Seasoned Equity Offerings. Journal of Financial Economics.
38(3), 243-267.
Akerlof, G. A. (1970). The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism. The Quarterly Journal of Economics. 84(3), 488-500.
Arrow, K. J. (1962). Economic Welfare and the Allocation of Resources for Invention. in R. R. Nelson (ed.), The Rate and Direction of Inventive Activity: Economic and Social
Factors. Princeton: Princeton University Press, 609-625.
Asquith, P., & Mullins, D. J. (1986). Equity Issues and Offering Dilution. Journal of
Financial Economics. 15(1), 61-89.
Bayless, M., & Chaplinsky, S. (1996). Is there a Window Opportunity for Seasoned Equity Issuance? The Journal of Finance. 51(1), 253-278.
Becker-Blease, J. R., & Irani, A. J. (2008). Do Corporate Governance Attributes affect Adverse Selection Costs? Evidence from Seasoned Equity Offerings. Review of
Quantitative Finance and Accounting. 30(3), 281-296.
Berger, A. N., & Udell, G. F. (1998). The Economics of Small Business Finance: The Roles of Private Equity and Debt Markets in the Financial Growth Cycle. Journal of
Banking & Finance. 22(6), 613-673.
Booth, L., & Chang, B. (2011). Information Asymmetry, Dividend Status, and SEO Announcement Day Returns. Journal of Financial Research. 34(1), 155-177. Brealey, R., Leland, H. E., & Pyle, D. H. (1977). Informational Asymmetries, Financial
Structure, and Financial Intermediation. The Journal of Finance. 32(2), 371-387. Brealey, R., & Myers, S. (2000). Principles of Corporate Finance. Boston:
Irwin/McGraw Hill.
Cai, J., & Loughran, T. (1998). The Performance of Japanese Seasoned Equity Offerings, 1971-1992. Pacific-Basin Finance Journal. 6(5), 395-425.
Carhart, M. M. (1997). On Persistence in Mutual Fund Performance. The Journal of Finance.
52(1), 57-82.
Carpenter, R. E., & Petersen, B. C. (2002). Capital Market Imperfections, High-Tech Investment and New Equity Financing. The Economic Journal. 112(477), 54-72. Dasilas, A., & Leventis, S. (2013). Corporate Governance, Dividend Status, Ownership
Structure, and the Performance of Greek Seasoned Equity Offerings. International
Journal of the Economics of Business. 20(3), 387-419.
Denis, D. J. (1994). Investment Opportunities and the Market Reaction to Equity Offerings.
The Journal of Financial and Quantitative Analysis. 29(2), 159-177.
Fama, E. F., & French, K. R. (1993). Common Risk Factors in The Returns on Stocks and Bonds. Journal of Financial Economics. 33(1), 3-56.
Hogan, T., & Hutson, E. (2005). Capital Structure in New Technology-Based Firms: Evidence from the Irish Software Sector. Global Finance Journal. 15(3), 369-387. Huang, R., & Tompkins, J. G. (2010). Corporate Governance and Investor Reactions to
Seasoned Equity Offerings. Managerial Finance. 36(7), 603-628.
Himmelberg, C. P., & Petersen, B. C. (1994). R&D and Internal Finance: A Panel Study of Small Firms in High-Tech Industries. The Review of Economics and Statistics. 76(1), 38-51.
Jegadeesh, N. (2000). Long-Term Performance of Seasoned Equity Offerings: Benchmark Errors and Biases in Expectations. Financial Management. 29(3), 5-30.
Costs and Ownership Structure. Journal of Financial Economics. 3(4), 305-360. Jung, K., Kim, Y.C. & Stulz, R.M. (1996). Timing, Investment Opportunities, Managerial
Discretion, and the security issue decision. Journal of Financial Economics. 42(2), 159-186.
Kim, E. H. (1978). A Mean-Variance Theory of Optimal Capital Structure and Corporate Debt Capacity. The Journal of Finance. 33(1), 45-63.
Kim, E. H., & Purnanandam, A. (2014). Seasoned Equity Offerings, Corporate Governance, and Investments. Review of Finance. 18(3), 1023-1057.
Korajczyk, R. A., Lucas, D. J., & McDonald, R. L. (1991). The Effect of Information
Releases on the Pricing and Timing of Equity Issues. The Review of Financial Studies.
4(4), 685-708.
Kraus, A., & Litzenberger, R. H. (1973). A State-Preference Model of Optimal Financial Leverage. The Journal of Finance. 28(4), 911-922.
Loughran, T., & Ritter, J. R. (1995). The New Issues Puzzle. The Journal of Finance. 50(1), 23-51.
MacKinlay, A. C. (1997). Event Studies in Economics and Finance. Journal of Economic
Literature. 35(1), 13-39.
Masulis, R. W., & Korwar, A. N. (1986). Seasoned Equity Offerings: An Empirical Investigation. Journal of Financial Economics. 15(1), 91-118.
Mikkelson, W. H., & Partch, M. M. (1986). Valuation Effects of Security Offerings and the Issuance Process. Journal of Financial Economics 15(1), 31-60.
Miller, M. H., & Modigliani, F. (1958). The Cost of Capital, Corporation Finance, and The Theory of Investment. The American Economic Review. 49(4), 655-669.
Miller, M. H. & Modigliani, F. (1961). Dividend Policy, Growth and the Valuation of Shares. Journal of Business. 34, 411-33.
Miller, M. H. & Rock, K. (1985). Dividend Policy Under Asymmetric Information. The
Journal of Finance. 40(4), 1031-1051.
Myers, S. C. (1984). The Capital Structure Puzzle. The Journal of Finance. 39(3), 574-592. Ross, S. A. (1977). The Determination of Financial Structure: The
Incentive-Signalling Approach. The Bell Journal of Economics. 8(1), 23-40.
Myers, S. C. & Majluf, N. C. (1984). Corporate Financing and Investment Decisions when Firms have Information that Investors do not have. Journal of Financial Economics.
13(2), 187-221.
NASDAQ database. Information retrieved from http://www.nasdaq.com/
Scott, J. H. (1977). Bankruptcy, Secured Debt, and Optimal Capital Structure. The Journal of
Finance. 32(1), 1-19.
Sjögren, H., & Zackrisson, M. (2005). The Search for Competent Capital: Financing of High Technology Small Firms in Sweden and USA. Venture Capital: An International
Journal of Entrepreneurial Finance. 7(1), 75-97.
Yang, L. & Malatesta, P. H. (2005). Credit Ratings and the Pricing of Seasoned Equity Offerings. Unpublished working paper.
Appendix
Table 11: Regression results with dependent variable CAR(-1,1) and estimation window (-250,-50)
Variable Model 1 Model 2 Model 3 Model 4
CONSTANT -0.0351 -0.0421 -0.0200 -0.0218 (-1.03) (-1.25) (-0.52) (-0.57) LNMKTCAP -0.0007 -0.0004 -0.0021 -0.0018 (-0.34) (-0.20) (-0.80) (-0.69) SIZE 0.0580 0.0609 0.0598 0.0575 (1.26) (1.32) (1.30) (1.24) RUNUP -0.0190** -0.0190** -0.0196** -0.0206** (-2.18) (-2.18) (-2.22) (-2.33) HIGHTECH 0.0099 -0.0684 -0.0714 (1.04) (-1.22) (-1.27) LNMKTHIGH 0.0060 0.0062 (1.44) (1.49) CRISIS -0.0119 (-1.19) DIV 0.0019 (0.19) ISPRY -0.0163 (-1.30) N 457 457 457 457 𝑅2 0.0181 0.0209 0.0239 0.0277 𝑅𝐴𝑑𝑗2 0.0116 0.0122 0.0131 0.0103
