Initial Public Offerings and the stock effect of industry rivals : an empirical research on rival valuation effects caused by underpriced IPOs listed on the NASDAQ Stock Market and the New York Stock Exchange

(1)

1

Initial Public Offerings and the stock effect of industry rivals

An empirical research on rival valuation effects caused by underpriced IPOs listed

on the NASDAQ Stock Market and the New York Stock Exchange.

Name Abraham Jacobus Vendel Student number 10829555

Supervisor dr. E. Zhivotova Coordinator dr. P.J.P.M. Versijp

Study programme BSc Economics and Business Specialization Finance & Organisation Amount of ECs 12

Date June 2018

The industry effect that occurs when underpriced IPOs are completed on the New York Stock Exchange and the Nasdaq Stock Exchange is being researched by analyzing stock data on daily returns from 2011-2017. According to the effect, the daily abnormal returns of seated firms are negatively affected around the completion date of the Initial Public Offering. The cumulative abnormal returns for different time intervals around the IPO are used as the main explanatory variable. Data shows that a significant negative effect is found for seated firms reacting to an IPO. With this effect in mind, there is indication that this negative effect could be explained by industry specifics. The data shows that the exchange is significantly determinative for the level of the cumulative abnormal return. To further extend this, the data shows that the industrial sector the seated firm operates in is highly determinative for the cumulative abnormal return.

(2)

2

Statement of originality

This document is written by student Bram Vendel who declares to take full

responsibility for the contents of this document.

I declare that the text and the work presented in this document are 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.

(3)

3

For many decades companies and investors worldwide are well familiar with the term: IPO. An Initial Public Offering or as it is briefly described by Draho (2014): “An IPO is the first time that shares in a company are sold to public investors and subsequently traded on the stock market”. As IPOs are well-known, worldwide several studies have done research regarding IPOs and their specific characteristics. Throughout the years IPOs became notoriously famous because of their first day average returns. These first day positive average initial returns result primarily from deliberate underpricing (Ruud, 1993). Underpricing is usually estimated as the percentage difference between the price at which shares were sold to investors during the offering period and the price at which they are being traded in the secondary market (Engelsen & Essen, 2010). Several aspects such as the IPO firm performance over the short and long term are often researched.

This worldwide phenomenon deviates widely in terms of the level of underpricing. Throughout the years the deliberate underpricing has seen lots of ups and downs. Especially in the years 2000-2001 the level of underpricing met an astonishing level of 65% (Loughran & Ritter, 2004). These enormous returns on the first day of trading are an extraordinary investment opportunity. CFOs base IPO timing on all market conditions, they are well informed regarding expected underpricing and feel that underpricing is a compensation for investors who are taking risk (Brau & Fawcett, 2006). These high variances in the level of underpricing are not only different in periods of time, but more specifically different for each and every IPO.

Despite the fact that there has been lots of research specifically on this phenomenon and that here have been several articles on the effect of different variables that do might have influence on this underpricing, there is very limited research to be found on the effect that IPOs have on their competitors. Research on the country an IPO was listed in or firm-specific variables such as the level of investments in research and development is previously done by (Engelen & Essen, 2010). However, in a world where markets are becoming more globalized, companies are growing at astonishing rates and investors seek to be all informed about everything, it is likely to assume that all companies are connected to each other in an interesting way. Almost solely

(5)

5 Akhibge, Borde & Whyte (2003) and Hsu, Reed & Rocholl (2010) have addressed this rival topic partly and have put emphasis on the effect of IPOs on industry rivals.

Therefore an opportunity is created to even further investigate this topic with new variables and a different timeframe. The latter is one of the most important aspects as the level of underpricing is changing tremendously over time (Loughan & Ritter, 2004). Solely Akhibge et al. (2003) did include the Exchange variable in the regression model. However, they didn’t find any significance for this variable. Hsu et al. (2010) argued that Akhibge et al. (2003) used a wrong dataset which biased their results, which raises a good assumption to look into their variables again as Hsu et al. (2010) didn’t do it. Further, several industry sector characteristics have been examined by the two fore mentioned researches, but neither one of them put direct emphasis on a particular industry sector. Since the existing literature is lacking results regarding the above mentioned industry-wide characteristics, there is an opportunity to find interesting results. Because of this opportunity the following research question is being answered in this thesis:

Do underpriced IPOs listed on the New York Stock Exchange (NYSE) or Nasdaq Stock Market (NASDAQ) in the timeframe of 2011-2017 have a significant negative effect on the short-run performance of industry rivals? And to what extend do different exchanges and industries have an effect on it.

In this empirical research the effect of underpricing on industry rivals is evaluated by the use of a newly created database which consists of daily abnormal returns for all the industry rivals for each specific Initial Public Offering. At first, a t-test can be done to see whether the cumulative abnormal returns differentiate significantly from zero. The intervals for the CAR that are being tested are the following: (-10, 10), (-5, 5), (-1, +1), (-3, 0), (-1, +3), (-1, 5) and (-1, +10). Secondly several regressions will be performed for different time intervals. Each regression has multiple interaction variables and dummies to gather information on what actually causes the effect.

The data that has been used for this research is derived from Wharton Research Data Services and Zephyr. By filtering a selection out of 1243 IPOs, which is explained more briefly later

(6)

6 on, 622 IPOs are held over. These IPOs have occurred in the timeframe of 2011-2017. All the IPOs are listed on the NYSE or the NASDAQ.

For the remainder of this research the structure of the thesis is as follows. In paragraph 2 the existing literature is being reviewed in the Sections: 2.1 Firm decisions to go public, 2.2 Post IPO Performance, 2.3 Competitiveness of IPOs. Afterwards in Section 3 the methodology is explained. In Section: 3.1 Sample Construction, 3.2 Variable Description and 3.3 Research Method. The empirical data analysis and results are discussed in Section 4, in specific: 4.1 The Industry Rival Effect on the NYSE & Nasdaq, 4.2 Exchange Effect, 4.3 Industry Sector Effect and in 4.4 The Mining Sector Effect. Subsequently, all these results are being discussed and formed into a conclusion in Section 5.

(7)

7 2. Literature Review

In this paragraph the existing literature surrounding IPOs is being explained and considered. In the first Section 2.1 is explained why companies go public. After that, Section 2.2 builds further on this base and explains what happens with the IPO performance wise. Thirdly for a connection to the thesis subject, in Section 2.3 the competitive effects of IPOs are being discussed.

2.1 Firm decisions to go public

A well-considered decision must it be for the management team of a company to finally agree whether to go public or not. A lot of advantages and disadvantages bring in their own weight to the final decision. Pagano, Panetta & Zingales (1998) stated: “The conventional wisdom is that going public is simply a stage in the growth of a company”. By arguing that the choice to go public is simply another stage in the growth of a company is rashly said there is enough reason to believe why it shouldn’t. There are numerous examples of United States companies, which are very large, that stay private and don’t need to go public in order to continue and increase their growth.

The first and probably biggest advantage for a company to go public is the fact that a new source of finance becomes available (Röell, 1984). This injection of cash is particularly used for growth by acquisition. Else the injection could be used to refinance current borrowings or the repayment of loans. Pagano et al. (1998) find that IPOs in Italy are mostly used to rebalance their balance sheet after large investments and high growth. Striking is that the size of the subsidiary of a publicly traded company doesn’t matter. Independent companies are even more likely to go public after periods of huge investment and growth.

The second most important advantage that comes to place is that the company has an enhanced company image and stronger publicity (Röell, 1984). To use the IPO as an marketing device sounds a little odd, but the public listing is providing for an initial certification for financial market professionals. It shows a longer term price signal to suppliers, the workforce and their customers. This price signal makes the company more reliable for the parties named above. They

(8)

8

can clearly see where the company is heading and this increases their confidence towards the company.

Diversification also tends to be an important reason for companies to go public. Especially riskier companies are more likely to go public. If a company faces a lack of liquidity or has borrowing constraints the main way out is to go public (Pagano, 1993). Diversification is also a manner to spread risks because of the risk averseness of owners and financial backers.

The last and fourth reason for companies to decide whether they go public is that an IPO motivates its management and employees. It decreases its agency costs as shareholders have a well-informed stock price that can be used into managerial performance-linked compensation (Röell, 1984). Having more transparency between shareholders and its management reduces conflicts and boosts unity. This transparency also motivates its employees because they have a clear view on where the company is heading.

Despite all the positive advantages there is also a negative side that contains the direct and indirect costs of going public. Ritter (1987) states that the direct costs include underwriter commission, legal, printing and auditing expenses. As far of a calculation approximately $250.000 plus 7% of the gross proceeds are estimated to be the direct costs for a firm to go public. Indirect costs can be directed to the indirect underwriter costs. The level of underpricing changes throughout the years but comes to 15% on average (Ibbotson & Loughran, 2004). This level of underpricing is even higher for best effort offers, in which the underwriter tries to sell as much shares as possible.

Another negative aspect that companies face is adverse selection and moral hazard. Investors are less informed than the issuers and cause for information asymmetry. This adversely affects the quality of the companies and the price of their shares (Leland and Pyle, 1977). This adverse selection is especially bad for small and young companies that have a low visibility and track record and are therefore aggrieved compared to larger and older companies (Chemmanur and Fulghieri, 1995).

The third negative advantage for companies to go public can be accounted to the loss of confidentiality. Stock exchanges have disclosure rules for companies that oblige them to unveil secretive information that can be crucial for their competitive advantage (Pagano et al., 1998).

(9)

9

Data about ongoing Research and Development (R&D) and future marketing strategies are crucial for their competitiveness. Yosha (1995) has shown that firms with sensitive information are less likely to go public since their costs of sharing this information are too high.

Sometimes the reason to go public is somehow different to what has been discussed previously in Section 2.1. Companies might have the urge to conduct profitable acquisitions, but do not have the ability to do this. Celikyurt, Sevilir and Shivdasani (2010) investigated this topic and found several reasons why IPOs are useful for mergers and acquisitions (M&As). The cash that has been raised by the IPO can be used to fund cash-financed acquisitions. Further, this publicly traded stock can also be used as an acquisition currency that can be used to pay for future M&As.

2.2 Post IPO performance

Companies that are going public through an Initial Public Offering place their shares on the market for a discount in order to guarantee that all the shares will be sold to the uninformed investors (Rock, 1986). This discount is the amount of underpricing, which on its turn, is the difference between the opening price and the closing price of the newly issued stock. The underwriter sets the price for new stock usually lower than the expected price to attract investors. Ibbotson & Jaffe (1975) showed that during the decade of the 1960s stock prices rose with approximately 11.4% in the first month. Being less informed than issuers, investors are being compensated for the risk they take.

Over time the initial first day returns have made tremendous jumps and falls and changed through several eloquent periods in time. During the 1980s the average return on the first day was 7%. This doubled to more than 15% in the 1990s before going to an astonishing 65% during the years of the internet bubble 1999-2000 (Ibbotson & Loughran, 2004). Both Ibbotson & Jaffe (1975) and Ritter (1984) found that for a number of periods the initial first day returns are excessively high. Ritter (1984) has examined the “Hot issue” market during the 1980s and found out that those excessive returns are probably caused by the steep rise in oil and gas prices. He also finds that periods of high volume tend to follow periods of high average initial return. These periods of underpricing are highly cyclical and last many months at a time in some periods. During a financial

(10)

10

crisis, volume on the market is lower than average. Following the reasoning of Ritter (1984), the period after the financial crisis from 2011 onwards, which is examined in this thesis, tends to have lower average initial returns that increase throughout the years.

On the contrary, the IPOs appear to be overpriced and underperforming in the long run (Ritter, 1991). He finds that after at least three years of launching the IPO, the firm is significantly underperforming a set of firms that is comparable in market and size. In his empirical research he shows that the average performance of a sample IPOs in the 3 years after going public is 34.47% while the performance of comparable firms in terms of industry and market size have an average performance of 61.86%. Several possible explanations for this underperformance are the following: risk mismeasurement, bad luck or fads and over optimism. This underperformance is mostly found at young growth companies, especially the ones that went public in high-volume years.

It’s clear to state that firms do have significant initial returns after the launch of an IPO. Underwriters set the stock price below the expected price to attract investors. For the risk they take investors are being compensated by high initial returns. The IPO stocks are performing above average in the short run, but unfortunately they are underperforming in the long run. This suggests that new issues are “lousy” investments in the long run and that the short run is more important for investors. This short term investment opportunity makes it suspicious to expect an effect on industry rivals in the short run and is therefore being researched in this research.

2.3 Competitiveness of IPOs

Several papers in the existing literature have focused on the rival valuation effect of transactions within capital markets. Research on the effect of an IPO on industry rivals has been done by Akhigbe, Borde and Whyte (2003). However insignificant results have been shown for a rival effect in the latter research, they do show significant effects for some of the variables used in a multivariate regression model. This research was done for IPOs launched between 1989-2000. Since Ibbotson & Loughran (2004) found that IPOs change significantly in average performance throughout the years, this raises the assumption to expect different results in this thesis compared to the insignificant results from Akhigbe et al. (2003).

(11)

11

However, other research has shown that rival firms do indeed react to proceeds from a firm within the same industry. Akhibge, Madura & White (1997) have found that bond rating downgrades have negative valuation effects on industry rivals. Further, Lang & Stulz (1992) show that bankruptcy announcements also do have a significant effect on industry rivals.

The type of information that is injected into a specific industry is highly important for it to cause any intra-industry valuation effects. Slovin, Sushka & Poloncheck (1992) make a distinction between two different types of information. On the one hand there is firm-specific information and on the other hand there is industry-wide information. Solvin et al. (1992) find that for industrial equity issues no such thing as an intra-industry effect is found. Therefore this type of information is accounted to firm-specific information. On the contrary, commercial bank issues do have significant valuation effects on rival commercial banking firms. The latter type of information is accounted to industry-wide information. The research proves that industry-wide information has an effect on industry rivals and thereby arouses the expectation that information such as the exchange or type of industry do have a significant effect too. Moreover, firms from different industrial sectors have divergent reasons to go public and are therefore expected to have different valuation effects on industry rivals.

This difference in types of information can be redirected to the types of IPO information. As mentioned in Section 2.1 firms go public because there are several advantages. Some of these advantages have implications on the industry as a whole and some of them are purely firm specific. Akhibge et al. (2003) say that an IPO could cause investors to reevaluate the competitive position of the IPO firm to its rivals. Accordingly, information could signal the industry or actually change the competitive balance in an industry. Although Akhibge et al. (2003) did not find any significance for the industry effect, they do find significance in their cross-sectional model for competitiveness and information effects. This insignificant effect on rivals is plausible and shown by research from Laux, Starks and Yoon (1998). They find that the information effect and competitive effects offset and result in insignificant returns on average for rivals.

However the research of Akhibge et al. (2003) did not show any significance for industry rivals, Hsu et al. (2010) reexamined this research and do find significant abnormal returns for industry rivals. They expanded this research even further by researching the complete opposite.

(12)

12

Instead of IPOs that are successfully launched they investigate the industry effect of IPOs that are withdrawn after announcement. As expected for this effect to cause positive abnormal returns for industry rivals, they find significance for the positive abnormal returns too. In addition to this latter research, a similar research from Slovin, Sushka & Bendeck (1991) shows significant positive abnormal returns too. However, they investigate the opposite to what Akhibge et al. (2003) have researched. Slovin et al. (1991) examine the industry effect of public firms going private. The result indicates that the industry reacts positively to the industry-wide information of a firm leaving the publicly traded market. The latter two researches support the correlation between firms going public and causing negative valuation effects and vice versa.

Besides firms going public in the first place and causing an industry-wide effect, the level of the effect is expected to be dependent upon the exchange a firm is listed on. There is a difference in the amount of information asymmetry. Firms on the Nasdaq Stock Market face higher information asymmetry than firms traded on the NYSE (Kohers, 1999). NYSE firms may have a higher level of institutional ownership, they are generally more well-known and they are traded more often than Nasdaq stocks. Based upon the above stated facts and assumptions, the NYSE firms are expected to react more intensively to the news which implies that their results are more significant than if they were traded on the Nasdaq.

2.4 Hypotheses

Judging from the researched literature on the subject, the resignation and join of firms from and to the public market leads to valuation effects for industry rivals. The type of industry, size, timing and exchange all tend to have a varying relation with the cumulative abnormal return. In this thesis it is therefore expected that:

1. A negative valuation effect is to be found for the cumulative abnormal returns of industry rivals around the IPO completion date.

2. An effect is to be found for IPOs that were listed on the New York Stock Exchange. 3. An effect is to be found for the industrial sector in which an IPO is operating.

(13)

13 3. Methodology

In this Section the methodology that is applied in this research is established. In the first Section the sample construction is described. In the follow up Section 3.2 the format of the sample data and different variables are being discussed. In Section 3.3 the research method is explained and in the last Section 3.4 the model set up is being performed for the different hypotheses.

3.1 Sample construction

In this empirical research the effect is measured that an underpriced IPO is causing to the valuation of its industry rivals. For the construction of the sample only IPOs are taking into account that are actually completed. As Ibbotson & Loughran (2004) calculate that IPOs are 15% underpriced on average, all IPOs that were gathered are taken into account.

A combined sample of two databases is used for this empirical research. All data on each specific IPO is gathered from the Zephyr database whereas the stock data on industry rivals for each specific IPO is gathered from Wharton Research Data Services database. Only companies that are traded on the NYSE or Nasdaq are taken into account in this research. The data gathered from both databases was in the time interval 01/01/2011 – 31/12/2017. The amount of industry rivals is dependent on each specific IPO as it varies from 1 to 196 industry rivals.

Variables gathered from Zephyr are: (i) Exchange, (ii) Deal value, (iii) Industrial Code SIC. Each one of these variables is characteristic for each and every IPO. At first 1243 IPOs were retrieved from Zephyr that had a deal value greater and equal to $50.000.000. This specific amount was set because rational thinking assumes that small IPOs do not influence the industry. All IPOs were included by Akhibge et al. (2003) and they did not find a significant amount.

Through a sample selection 621 IPOs are deleted from the sample. IPOs that had no 4-digit SIC code, were unlisted or had no stock data for any industry rivals were deleted. Following Hsu et al. (2003) all financial firms are deleted. This is supported by Solvin et al. (1992) since they only find an intra-industry effect instead of an effect on industry rivals. Therefore all IPOs containing SIC codes between 6000-6999 are deleted. Finally a sample of 622 IPOs is left.

(14)

14 Daily S&P 500 Returns, (iii) Volume, (iv) Shares outstanding, (v) Price of the stock. Industry rivals

without sufficient data around the IPO completion date are deleted from the sample.

Based on variables from Wharton Research Data Services and Zephyr several new variables are created: (i) Market Capitalization, (ii) Abnormal Return, (iii) Cumulative Abnormal Return, (iv)

Manfacturing, (v) Mining and (vi) Services. A summary on how these variables are created can be

found in Section 7.2 of the Appendix.

3.2 Variable Description

The Dependent Variable in this thesis is the Cumulative Abnormal Return (CAR). By the use of the

CAR, the industry rival effect can be evaluated. The CAR is explanatory for the return the industry

rival within a given timeframe. Before this cumulative return can be calculated several steps have to be satisfied in order to end up with the CAR. Firstly the Abnormal Return has to be calculated. The Abnormal Return is calculated by subtracting the actual return of the industry rival by the return on the S&P500.

The S&P500 is chosen as a benchmark since all the firms in each IPO portfolio are traded on the NASDAQ or NYSE. Some of the industry rivals in the IPO portfolios are part of the S&P500. Therefore the S&P500 is the perfect benchmark to use for the Northern American IPO portfolios. To proceed on the valuation the CAR is calculated by taking the sum of the Abnormal Returns for each industry rival in a portfolio.

The Independent Variables in this thesis are the Exchange Dummy (NYSE) and the Industry

Dummies for each broad industry. The Exchange Dummy (NYSE) is chosen to be one of the

dependent variables because Kohers (1999) states that there is significant difference in the information asymmetry between stocks traded on the NYSE and the NASDAQ. The variable has the

(15)

15

value 1 given to it if the IPO is listed on the New York Stock Exchange and value 0 if it is listed on the NASDAQ. To avoid multicollinearity no variable for NASDAQ is created. The industry sector variable is chosen to be an independent variable because previous literature shows that industry-wide information affects the CAR.

The Industry Dummies are several dummy variables for an overarching segment for a type of industry. Dummy variables are only created for industries that are over represented in the dataset. IPOs that were listed were mainly operating in the following industries: Mining, Manufacturing and Services. For each one of these industries an overarching dummy variable is created. The dummy variable has the value 1 given to it if the IPO is operating in the concerned industry. The variable has the value 0 if the IPO is not operating in the concerned industry. As several industry codes are not taken into account or linked to a specific industry, the problem of multicollinearity is not present.

In order to find a significant relation between the dependent and the independent variables, control variables are added to the regression equation. The first control variable is the natural logarithm of the Deal Value of the IPO firm. The deal value is the total amount of $ that the IPO raised. Similar to Akhibge et al. (2003) the value of the IPO firm is chosen since they showed evidence that the size of an IPO has a negative impact on rival performance. They argue that IPOs can be a bigger threat to an industry if they are significantly large in size.

The second control variable that is added to the regression equation is the market size of the industry rival. The reason this variable is chosen is the same as the one described above and applies here too. Contrarily, the larger the size of the industry rival less impact on the price is expected. The market size (market capitalization) is calculated by multiplying the total amount of shares outstanding by the price per share.

The third control variable is the volume of the industry rival. The volume is the amount of shares traded per day. This variable is chosen to be a good control variable since Kohers (1999) states that stocks which are traded more actively are expected to have less information asymmetry. Stocks with a higher share volume are therefore related to the CAR as they generate more significant results.

(16)

16 3.3 Research method

To find evidence on the effect that underpriced IPOs are causing to industry rivals, a similar approach to Akhibge et al. (2003) is used. To firstly see whether there is an actual effect on industry rivals , their CARs can be tested by the use of a t-test. A t-test is very useful in the sense that by using a t-test, robust standard errors can be applied. For all the pre-specified CAR intervals the same t-test is applied repeatedly. The first hypothesis that can be tested for all CAR intervals:

Hypothesis 1:

The following t-test will be used to check whether the CARs do significantly differentiate negatively from zero:

Where:

= t-test score

= Standard Deviation = Total amount of CARs

= Cumulative Abnormal Return (for a specific time period)

In case of finding statistical evidence for the valuation effect to be true it is expected that: H1: CARi < 0

To further extend the research on the stock performance of the portfolios, different time intervals are chosen for the CARs. Since this research is focused on the short term stock effect of rivals, different time intervals within the short period of time benchmark (-10, 10) are created. Similar to Akhibge et al. (2003) different CAR intervals are created around the completion date.

(17)

17 The different time periods, the price valuation effects can differ significantly on a short notice. Therefore the following CAR intervals are created for the estimation and are shown in Table 1.

Table 1: Summary Statistics On The Cumulative Abnormal Return

(1) (2) (3) (4) (5)

VARIABLES N(obs) Mean SD Min Max

CAR (-10, 10) 23,562 -0.004738 0.164 -1.289 2.339 CAR (-5, 5) 23,562 -0.003014 0.122 -1.267 2.043 CAR (-1, 1) 23,562 -0.000344 0.0666 -1.179 2.118 CAR (-3, 0) 23,562 -0.0021759 0.0738 -0.963 1.548 CAR(-1, 3) 23,562 -0.0006057 0.0862 -1.031 2.312 CAR(-1, 5) 23,562 -0.0006288 0.0989 -1.462 2.085 CAR(-1, 10) 23,562 -0.0027333 0.125 -1.652 2.130

Table 2: Summary Statistics

(1) (2) (3) (4) (5)

VARIABLES N(obs) Mean SD Min Max

Deal Value** 23,562 325,253 1.640e+06 50,000 4.700e+07

Volume 23,562 1.305e+06 3.555e+06 12 3.120e+08

Market Cap.*** 23,562 5.021e+06 1.882e+07 345.2 6.323e+08

ln(Deal Value)* 23,562 11.94 0.921 10.82 17.67 ln(Volume)* 23,562 12.64 1.944 2.485 19.56 ln(Market Cap.)* 23,562 13.34 2.114 5.844 20.26 NYSE 23,562 0.397 0.489 0 1 Services 23,562 0.416 0.493 0 1 Manufacturing 23,562 0.258 0.438 0 1 Mining 23,562 0.257 0.437 0 1

* The ln is the natural logarithm of the variable between brackets ** The value is the amount of USD ($) in thousands

(18)

18 If statistical evidence is found for the rival valuation effect, it is expected that investors trade rationally and based on the information they have. As mentioned in the previous literature there is a difference between firm specific information and industry wide information. On the New York Stock Exchange investors attribute a high level of significance to firm specific information of industry related firms (Kohers, 1999). The IPOs launched on the New York Stock Exchange are expected to cause a more statistically significant effect on the CARs as stocks are more actively traded and information is more valuable to investors. The second hypothesis can be tested to see if the dummy variable is significantly lower than zero.

Hypothesis 2:

To find evidence on the Exchange Effect, the following regression is tested:

In case of an Exchange Effect, investors are more closely tied and focussed on industry rival information. The Nasdaq Stock Market has bigger information asymmetries and companies are less known than on the New York Stock Exchange. Therefore it is expected that:

H2: β2 > 0 or β2 < 0

To further examine this research on specific firm characteristics the industry an IPO is listed in is also taken into account as an explanatory variable to test the third hypothesis. It is likely to expect firms to react differently across industries to the information of a firm going public. Following Hsu et al. (2010) all the financial firms are not taken into account for the sample. Slovin et al. (1992) found evidence on an intra-industry effect for financial firms after an announcement of the issuance of common stock. However no effect is found for industry rivals in the sample. The three industries that are overly represented in the sample are chosen to be the explanatory variables for the third hypothesis. Since almost 80% of the IPOs are listed in these three industrial

(19)

19 areas, a significantly large amount of data is available for the regression. The third hypothesis can be tested to see whether any of the three dummy variables is significantly different from zero.

Hypothesis 3:

To find evidence on the industrial area effect, the following regression is tested:

In case of an Industry Effect, firms from different industries react differently to news. Some may overreact and some may underreact to the news. An Industry Effect takes place if at least one of the three variables is significantly different from zero. Therefore it is expected that:

H3: βk < 0 for k (1, 2 & 3) and βk > 0 for k (1, 2 & 3)

(20)

20 4. Results and Analysis

In this Section the results of the regressions and models that are set forth in Section 3 are being discussed and reviewed. Following the approach of Akhibge et al. (2003) & Hsu et al. (2010) and using matching models, results are generated. Each different sub-paragraph is specifically for each hypothesis that is being tested. In Section 4.1 the first hypothesis being tested. Subsequently in Section 4.2 and 4.3 the second and the third hypotheses are being tested and reviewed respectively. In Section 4.4 the first and second hypotheses are tested again due to the findings in Section 4.3.

4.1(a) Industry Rival effect on the NYSE and the NASDAQ

In this first Section of the results, the statistical evidence is shown for the effect that underpriced IPOs cause to industry rivals. By the use of the t-tests mentioned in the first Section of the

Methodology, the effect is shown.

In Table 3 (page 20) the output of all the t-tests are summarized. The table contains the 7 t-tests that are tested by the model used from Section 3.1. From top to bottom the table shows the CAR interval, mean, value and the amount of observations (N). For the ease of use each t-test is given its own number. The reference number for each t-t-test is easy to look up and interpret in the follow up paragraph (b) of Section 4.1.

Table 3: T-test table on Cumulative Abnormal Return

Test (1) (2) (3) (4) (5) (6) (7)

CAR(-10, 10)

CAR(-5, 5) CAR(-1, 1) CAR(-3, 1) CAR(-1, 3) CAR(-1, 5) CAR(-1, 10) Mean -0.0047 *** _-0.0030 *** - 0.0003 -0.0022 *** -0.0007 -0.0006 -0.0027 *** (-4.42) (-3.78) (-0.79) (-4.52) (-1.16) (-0.97) (-3.35) N 23526 23526 23526 23526 23526 23526 23526 t statistics in parentheses *_{p < 0.05,}**_{p < 0.01,}***_{p < 0.001}

(21)

21 4.1(b) Industry Rival effect on the NYSE and the NASDAQ

To find statistical evidence for the industry rival effect to occur, the CAR has to deviate significantly from zero. The CAR is either above 0 or below 0.

To start with the t-tests in Table 2 (page 20), we can see that all the different intervals have a mean below zero. However not ever CAR interval is statistically significant. We can easily interpret that the t-tests (1), (2), (4) and (7) are statistically significant and different from zero. Surprisingly all four t-tests are significant at the 1% level.

To further examine the different t-tests we start with the first two. Test (1) has a mean of -0.474% and has a t-score of -4.42. Test (2) has a mean of -0.301% and a t-score of -3.78. This research follows a same likewise approach as Akhibge et al. (2003) and Hsu et al. (2010) in the way the CAR intervals are determined. Contrarily to Akhibge et al. (2003) significant results have been found where they did not. Similarly to Hsu et al. (2010) for the same intervals significance is shown and at the (-5,5) interval even a higher significance at the 1% level. As expected the significance is shown for the overall time interval around the IPO. This is in line with the expectations and content of previous research.

To further examine the results we continue with test (4). This test measures the time interval three days in advance of the completion of the IPO. The mean is -0.218% and the t-score is -4.52. The exact opposite interval is performed in test (5) and no statistical significance is found for this test. We find that prior to the announcement, industry rivals are performing worse than after the announcement in the same time interval. This could be explained by a logical explanation. A larger share of investors might sell the stocks of industry rivals prior to the completion of the IPO in order to have cash ready for the purchase of the newly listed stocks. To enjoy maximum profits they could buy the stocks right as they are listed.

For the last significant test (7) the mean is -0.273% and the t-score is -3.35. In contrary to tests (5) and (6) we see that for a longer time interval after the completion of the IPO the CAR become significantly negative.

We find that there is a clear difference between the different CAR intervals. Industry rivals are losing significant value prior to the completion of the IPO as their share price decreases in value. However for the time after completion we see that industry rivals are not significantly

(22)

22

negative affected by the IPO. The reason for this could be that investors already react prior to the completion date to be all set and ready to intervene on the market.

4.2(a) Exchange Effect

In the second Section of the results statistical significance is shown for the Exchange to have effect on the CAR. The CAR is estimated by the use of the multivariate regressions which are explained in the second paragraph of the Methodology Section 3.2.

In Table 4 (page 22), the output of the four different regressions are displayed. The table contains three different models. The first two models CAR(-10, 10) and (-5, 5) are chosen because of their significance in the t-tests in Section 4.1. The third model which consists the CAR(-1, 1) is added for the overall clarity of the analysis. The skewness that is caused by the Deal Value, Volume and Market Capitalization is solved by transforming the three variables with the normal logarithmic scale. In order to have a clear construction in this thesis and for the ease of reference, numbers are given to each model.

Table 4: Exchange Effect on CARs (-10, 10), (-5, 5), (-1, 1)

Model (1) (2) (3)

CAR(-10, 10) CAR(-5, 5) CAR(-1, 1)

NYSE(Dummy) -0.0186*** _-0.0126*** _-0.00255** (-8.04) (-7.29) (-2.67) Ln(Deal Value) -0.0018 -0.0012 0.0002 (-1.52) (-1.28) (0.40) Ln(Market Cap.) 0.0032** _0.0021** _0.0001 (3.23) (2.59) (0.07) Ln(Volume) -0.000700 -0.0011 0.0007 (-0.68) (-1.28) (1.35) Constant -0.0093 0.0021 -0.0116 (-0.61) (0.18) (-1.87) N R-Squared 23526 0.05 23526 0.03 23526 0.01 t statistics in parentheses *_{p < 0.05,}**_{p < 0.01,}***_{p < 0.001}

(23)

23 4.2(b) Exchange Effect

In order to find statistical significance for the Exchange Effect to occur, the dummy variable has to be: NYSE > 0 or NYSE < 0.

In the first row of Table 4 the independent variable: NYSE(dummy) is shown. We can clearly see that for all the different models, the coefficient is significantly different from zero. We can also state that all the coefficients are negatively related with the dependent variable. However for some models more variables are significant but this will follow in a later paragraph of this Section.

For the first model the coefficient of the independent variable is -0.0186 and has a t-score of -8.04. The Exchange Dummy is significant at the 0.1% level. The explanatory variable is negatively related to the dependent variable. In other words, if the IPO is listed on the New York Stock Exchange the Cumulative Abnormal Return of industry rivals is 1.86% lower than if the IPO was listed on the Nasdaq Stock Exchange. This result falls in line with the expectation of Kohers (1999), which expects a higher significant effect for firms traded on the NYSE. We see that information is more valuable to investors since the CAR is significantly lower.

This significant effect is also present for the other models (2) & (3). The coefficients and t-scores are: -0.0126 & -7.29 and -0.00255 & -2.67 respectively. The significance that is present for these models follows the same reasoning as above. However, the natural logarithm of the Market

Capitalization is significant at the 1% level for model (1) and (2). Both coefficients from Market Capitalization variable positive. This follows the expectations of the existing literature. The larger

the size of the industry rival, the less impact on the industry rival is expected. As the Market

Capitalization is higher the CAR becomes less negative.

(24)

24 4.3(a) Industry Sector Effect

In this Section similar models are used for the regression to test the third hypothesis. The three new explanatory variables are added instead of the exchange explanatory variable. This explanation of the model can also be found in Methodology part in Section 3.3. In Table 5, the output of the three different models are shown. For the ease of use and construction of this thesis consecutive numbers are chosen for these models. The models start from number (4) up to and including (6).

Table 5: Specific Industry Effect on CARs (-10, 10), (-5, 5) & (-1, 1)

Model (4) (5) (6)

CAR(-10, 10) CAR(-5, 5) CAR(-1,1)

Mining -0.0137*** _-0.0082*** _-0.0033* (-4.22) (-3.32) (-2.52) Manufacturing 0.0093** _0.0042 _-0.0018 (2.58) (1.56) (-1.21) Services 0.0066* _0.0052* _0.0001 (2.09) (2.21) (0.05) Ln(Deal Value) -0.0021 -0.0018 -0.0001 (-1.72) (-1.94) (-0.13) Ln(Market Cap.) 0.0030** _0.0019* _-0.0000 (3.01) (2.38) (-0.08) Ln(Volume) -0.0005 -0.0009 0.0008 (-0.49) (-1.10) (1.47) Constant -0.0157 0.0033 -0.0080 (-0.96) (0.27) (-1.14) N R-Squared 23526 0.05 23526 0.03 23526 0.01 t statistics in parentheses *_{p < 0.05,}**_{p < 0.01,}***_{p < 0.001}

(25)

25 4.3(b) Industry Sector Effect

For a specific industry effect to occur, at least one of the explanatory dummy variables has to deviate significantly from zero. We have to find an effect that holds: Mining = Manufacturing =

Services ≠ 0, where at least one differs from zero. On the first three rows of Table 5 (page 24)

these dummy variables are displayed.

What stands out is that the Mining dummy variable is significant for model (4) and (5) at the 0.1% level. It is significant at the 5% level for model (6). All the coefficients for the Mining dummy are negative. This implies that firms, which are operating in the mining industry, are more vulnerable to underpriced IPOs. Mining firms are performing worse to an underpriced IPO than firms operating in a different industry. The coefficients and t-scores for model (4), (5) and (6) are -0.0137 & -4.22, -0.00816 & -3.32 and -0.00325 & -2.25 respectively.

Further we see that the Services sector is significant at the 5% level for model (4) and (5). The coefficient is significantly positive in those models which implies that the Service sector is less impacted by an IPO. For instance, if a firm is operating in the Services sector the CAR, in model (4), the CAR is 0.663% higher. The coefficients and t-scores for the Services sector from model (4) & (5) are 0.00663 & 2.09 and 0.00524 & 2.21 respectively.

The Manufacturing sector is only significant for model (4) at the 1% level. The coefficient is 0.00933 and the t-score is -4.22. The coefficient is positive which implies that the Manufacturing sector is also less impacted by an IPO. However, this only applies for the CAR(-10, 10) interval. Again for these models the control variable Market Capitalization is significant at the 1% level.

First of all, the results indicate that there is a clear difference in stock valuation between the several industrial sectors. The Mining sector reacts more significantly to the news of a company going public than the Manufacturing sector and the Services sector. Secondly, the Mining sector reacts negatively to the news, while the Manufacturing and the Services sector react positively. An explanation for the difference in CARs could be that a joining firm in the Mining Sector could increase the overall competitiveness in the industry which devaluates the stocks of incumbent firms. On the exact opposite the other sectors may have an industry that consists out

(26)

26

of many firms where the competitiveness is already at a large scale so joining firms don’t make a difference in stock valuation anymore.

4.4(a) The Mining Sector Effect

In Section 4.3 a clear difference is shown between the different industrial sectors that are overly represented in the data. However, rational thinking assumes that the highly significant Mining sector has some influence on the overall outcome of the regression estimations. The Mining sector is negatively related to the CAR, but the other sectors are positively related to the CAR. While the overall estimated results for the different CAR intervals are negative Table (1), the assumption arouses that the Mining sector is influencing these results.

To see whether this industrial sector effect is actually influencing the results, we run the regressions performed in Section 4.1 again on two new datasets. One data set containing solely the Mining sector and one data set only containing all the other industrial sectors. All the results on the same t-tests performed in Section 4.1 without the Mining Sector are shown in Table 6. All the results with only the Mining Sector are shown in Table 7.

Table 6: T-test table on Cumulative Abnormal Return (No Mining Sector)

Test (1) (2) (3) (4) (5) (6) (7) CAR (-10, 10) CAR (-5, 5) CAR (-1, 1) CAR (-3, 0) CAR (-1, 3) CAR (-1, 5) CAR (-1, 10) Mean 0.0008 0.0005 0.0003 -0.0012* _0.0012 _0.0018* _0.0018 (0.66) (0.57) (0.59) (-2.04) (1.90) (2.33) (1.88) N 17487 17487 17487 17487 17487 17487 17487 t statistics in parentheses *_{p < 0.05,}**_{p < 0.01,}***_{p < 0.001}

Table 7: T-test table on Cumulative Abnormal Returns (Only Mining Sector)

Test (1) (2) (3) (4) (5) (6) (7) CAR (-10, 10) CAR (-5, 5) CAR (-1, 1) CAR (-3, 0) CAR (-1, 3) CAR (-1, 5) CAR (-1, 10) Mean -0.0209***_-0.0133***_-0.0022**_-0.0051***_-0.0063***_-0.0077***_-0.0159*** (-10.81) (-9.45) (-2.93) (-6.06) (-6.48) (-6.65) (-10.70) N 6039 6039 6039 6039 6039 6039 6039 t statistics in parentheses *_{p < 0.05,}**_{p < 0.01,}***_{p < 0.001}

(27)

27 4.4(b) The Mining Sector Effect

First of all, in Table (1) (page 20) we can see that the total amount of CARs is 23.526. The total amount of CARs that come from the Mining Sector is 6039 and can be found in Table (7) on page (26). The amount of CARs from the Mining Sector account for roughly 25% of the dataset. The amount of CARs that are not from the Mining Sector is 17487 and is about 75% of the dataset.

We can clearly see that almost all the t-test results in Table (7) are significant at the 0.1% level except the (-1, 1) interval. These results show that the Mining Sector is reacting significantly negative for all the different time intervals. This is the opposite of the results we found in Section 4.1 as it showed that the stock valuation of industry rivals was affected at a higher level prior to the completion date. In Table (6) the only CARs intervals that are significant are the (-3, 0) & (-1, 5) intervals at the 5% level. Where the (-3, 0) is negatively significant and the (-1, 5) is positively significant .

If we compare the results from Table (7) with the results from Table (6), we can immediately see the difference in significance levels. Besides this difference we can also clearly see that the CARs for the data set without the Mining industry are not negative on average. This results falls in line with the results found in Section 4.3. Section 4.3 shows that he two other overly represented sectors Manufacturing and Services have positive coefficients.

However, by analyzing these results we come to the conclusion that the Mining Sector which accounts for only 25% of the data is responsible for the overall statistical significance of the models used in Section 4.1 and 4.2. Since the industrial Sector effect has not been taken into account in previous researches, the overall verity of results found might be biased. Since it seems a little odd to end up with a result that is contradict to the results found in Section 4.2 a new regression is ran with industrial fixed effects to not completely refute the findings in Section 4.2. Therefore the same regression model is used while taking differences for the industrial sector as a fixed effect.

The results of this regression are displayed in Table (8) on page(28). The same models are used as in Section 4.2, except the “+” indicates that the industrial fixed effect is taken into account.

(28)

28 Table 8: Exchange Effect on CARs (-10, 10), (-5, 5), (-1, 1) (Industry Fixed Effects)

Model (1+) (2+) (3+)

CAR(-10, 10) CAR(-5, 5) CAR(-1, 1)

NYSE -0.0186* _-0.0126 _-0.0026 (-2.08) (-1.38) (-1.42) Ln(Deal Value) -0.0018 -0.0012 0.0002 (-0.54) (-0.32) (0.14) Ln(Market Cap.) 0.0032 0.0021 0.0000 (1.94) (1.17) (0.04) Ln(Volume) -0.0007 -0.0011 0.0007 (-0.23) (-0.42) (0.77) Constant -0.0094 0.0021 -0.0116 (-0.33) (0.06) (-0.68) N 23526 23526 23526 t statistics in parentheses *_{p < 0.05,}**_{p < 0.01,}***_{p < 0.001}

In the first row of Table 8 we see that the coefficient of the NYSE dummy variable is only significant for the (-10, 10) CAR interval at the 5% level. The coefficient is -0.0186 and the t-score is -2.08 in model (1+). As expected we see that for Table (4) and Table (8) the coefficients are exactly the same. Including the industrial Sector as a fixed effect only affects the standard errors as it makes the group means non-random.

Despite the fact that the level of significance is lower than in Section 4.2 and that the significance only applies for one time interval than three, the result still falls in line with the previous research. Again we can say that the Exchange Effect is present and statistically significant for the CAR of industry rivals.

(29)

29 5. Conclusion and Decision

This thesis examines the effect that underpriced initial public offerings are causing to the valuation of industry rival firms. In further detail, it is examined which different industry characteristics do have an influence on the valuation of industry rivals. Through the investigation of relevant previous literature, the expectation is that the Industry Effect is linked to firm specific characteristics like the IPO firm size, industry rival size and the volume (amount of shares traded). Therefore, the industry-wide characteristics are assumed to be very characteristic too and have a lot of significance for the effect to take place. The following research question sums this up:

Do underpriced IPOs listed on the New York Stock Exchange (NYSE) or Nasdaq Stock Market (NASDAQ) in the timeframe of 2011-2017 have a significant negative effect on the short-run performance of industry rivals? And to what extend do different exchanges and industries have an effect on it.

Through the examining and consideration of different literature on the Industry Effect, the American exchanges: the New York Stock Exchange and the Nasdaq Stock Market are expected to have a significant effect on the stock performance of industry rivals. To go even deeper into industry specifics, the industry sector a firm is operating in is expected to have a significant effect on the stock performance of industry rivals. The researched literature leads to the expectation of a relationship of these variables with stock performance of industry rivals. Therefore the following hypotheses are formed:

1. A negative valuation effect is to be found for the cumulative abnormal returns of industry rivals around the IPO completion date.

2. An effect is to be found for the IPOs that were listed on the New York Stock Exchange. 3. An effect is to be found for the industrial sector in which an IPO is operating.

(30)

30 The first results found in Section 4.1, 4,2 and 4,3 indicate that the Industry Effect is present for industry rivals listed on the NYSE and the Nasdaq. This effect takes place especially in the days prior to the completion of the IPO. Therefore Hypothesis 1 can be confirmed. Research on the Exchange Effect turns out to be present which confirms Hypothesis 2. The latter Hypothesis 3 can be confirmed as well since the CAR is significantly related to industry sectors.

To answer the research question: evidence is found on the Industry Effect to be significantly present in industries in which IPOs are launched. This effect is caused by industry characteristics that are significantly explanatory. This conclusion is drawn because if companies were not affected by the listing of a firm the results found would not show up.

As the high level of significance for the Mining sector in Section 4.3 was causing a lot of suspicion, further research shows that this sector is influencing the overall results significantly. However, final results show that the result on the Exchange Effect is not biased as it is still present in the regression with industrial sector fixed effects.

A suggestion for further research is to examine the specific characteristics of the Mining industry and see why this sector is reacting more intensively to IPO news than other industries. A second suggestion is to examine the difference between industry sectors reacting positively or negatively to the news of an IPO.

(31)

31 6. Bibliography

Akhigbe, A., Borde, S.F., & Whyte, A.M. (2003). Does an industry effect exist for initial public offerings? The Financial Review, 38(4), 531-551.

Akhigbe, A., Madura, J., & Whyte, A.M. (1997). Intra-industry effects of bond rating adjustments.

Journal of Financial Research. pp. 545-561.

Boveniste, L.M. & Spindt, P.A. (1989). How investment bankers determine the offer price and allocation of new issues. 24(2). pp. 343-361.

Brau, J.C. & Fawcett, S.E. (2006). Initial Public Offerings: An Analysis of Theory and Practice.

Journal of Finance, 71(1), pp. 399-436.

Brealey, R., Leland, H. E., & Pyle, D. H. (1977). Informational asymmetries, financial structure, and financial intermediation. The journal of Finance, 32(2), 371-387.

Celikyurt, U., Sevilir, M., & Shivdasani, A. (2010). Going public to acquire? The acquisition motive in IPOs. Journal of Financial Economics, 96(3), 345-363.

Chemmanur, T. J., & Fulghieri, P. (1995). Information production, venture-capital financing, and the decision to go public. In Winter Meeting of the Econometric Society, Washington, DC, USA. Datta, S. et al. (2015). On post-IPO stock price performance: A comparative analysis of RLBOs and IPOs. Journal of Banking and Finance. 55, pp. 187-203.

Engelen, P. J., & Van Essen, M. (2010). Underpricing of IPOs: Firm-, issue-and country-specific characteristics. Journal of Banking & Finance, 34(8), 1958-1969.

Hsu, H.-C., Reed, A.V., & Rocholl, J. (2010). The new game in town: competitive effects of IPOs. The Journal of Finance, 65(2), 495-528.

Ibbotson, R.G. and Jaffe, J.F. (1975). Hot Issue” markets. The Journal of Finance, 4, pp. 1027-1042.

Kohers, N. (1999). The industry‐wide implications of dividend omission and initiation

announcements and the determinants of information transfer. Financial Review, 34(1), 137-158. Lang, L.H.P., & Stulz, R.M. (1992). Contagion and competitive intra-industry effects of bankruptcy announcements: an empirical analysis. Journal of Financial Economics, 32(1), 45–60.

Laux, P., Starks, L. T., & Yoon, P. S. (1998). The relative importance of competition and contagion in intra-industry information transfers: An investigation of dividend announcements. Financial

(32)

32

Loughran, T., & Ritter, J. (2004). Why has IPO underpricing changed over time? Financial Management, 33(3), 5–37.

Mauer, D. et al. (2015). Global diversification and IPO returns. Journal of Banking and Finance. 58. pp. 436-456. of Financial Research, 20(4), 545-561.

Pagano, M., Panetta, F., & Zingales, L. (1998). Why do companies go public? An empirical analysis. The Journal of Finance, 53(1), 27-64.

Ritter, J. R. (1987). The costs of going public. Journal of Financial Economics, 19(2), 269-281. Röell, A. (1996). The decision to go public: An overview. European Economic Review, 40(3-5), 1071-1081.

Ruud, J.S. (1993). Underwriter price support and the IPO underpricing puzzle. Journal of Financial

Economics, 34(2), pp. 135-151.

Sharpe, W. F. (1970). Portfolio theory and capital markets. McGraw-Hill College.

Slovin, M.B., Sushka, M.E., & Bendeck, Y.M. (1991). The intra‐industry effects of going‐private transactions. Journal of Finance 46(4), 1537–1550

Slovin, M.B., Sushka, M.E., & Polonchek, J.A. (1992). Informational externalities of seasoned equity issues: differences between banks and industrial firms. Journal of Financial Economics, 32(1), 87-101.

Yosha, O. (1995). Information disclosure costs and the choice of financing source. Journal of

(33)

33 7. Appendix

7.1 List of Variables

Variables Type Notes Database

Dependent Variable

CAR(-10, 10) Daily Cumulative of Daily Abnormal Returns WRDS (CRSP) CAR(-5, 5) Daily Cumulative of Daily Abnormal Returns WRDS (CRSP) CAR(-1, 1) Daily Cumulative of Daily Abnormal Returns WRDS (CRSP) CAR(-3, 1) Daily Cumulative of Daily Abnormal Returns WRDS (CRSP) CAR(-1, 3) Daily Cumulative of Daily Abnormal Returns WRDS (CRSP) CAR(-1, 5) Daily Cumulative of Daily Abnormal Returns WRDS (CRSP) CAR(-1, 10) Daily Cumulative of Daily Abnormal Returns WRDS (CRSP)

Independent Variable

NYSE Binary [0, 1] 1 if the IPO was listed on the NYSE, Zephyr 0 if otherwise

Mining Binary [0, 1] 1 if the Industry Rival is operating in the industry, Zephyr 0 if otherwise

Manufacturing Binary [0, 1] 1 if the Industry Rival is operating in the industry, Zephyr 0 if otherwise

Services Binary [0, 1] 1 if the Industry Rival is operating in the industry, Zephyr 0 if otherwise

Control Variable

Deal Value Daily The amount of USD $ raised by an IPO Zephyr in thousands

Market Cap. Daily The amount of USD $ each Industry Rival WRDS (CRSP) is worth. (shares outstanding * price)

(34)

34 7.2 Variable Creation

7.3 Regression Outputs

7.3(a) Cumulative Abnormal Return t-tests

T-tests On All Data

CAR(-10, 10)

Variable Caclulation/Explanation

Market Cap. Total Shares Outstanding * Share Price = Market Capitalization Abnormal Return Normal Return - Return S&P500 = Abnormal Return

Cumulative Abnormal Return Sum(Abnormal Return)/Number of Abnormal Returns = Cumulative Abnormal Return

NYSE 1 if the IPO is listed on the NYSE 0 if the IPO is listed on the Nasdaq Mining 1 if the SIC code is >999 or <1500

0 if otherwise

Manufacturing 1 if the SIC code is >1999 or <4000 0 if otherwise

Services 1 if the SIC code is >6999 or <9000 0 if otherwise

Pr(T < t) = 0.0000 Pr(|T| > |t|) = 0.0000 Pr(T > t) = 1.0000 Ha: mean < 0 Ha: mean != 0 Ha: mean > 0 Ho: mean = 0 degrees of freedom = 23525 mean = mean(cum_abn_1010) t = -4.4164 cum~1010 23,526 -.0047378 .0010728 .1645412 -.0068404 -.0026351 Variable Obs Mean Std. Err. Std. Dev. [95% Conf. Interval] One-sample t test

(35)

35 CAR(-5, 5)

CAR(-1, 1)

CAR(-3, 0)

Pr(T < t) = 0.0001 Pr(|T| > |t|) = 0.0002 Pr(T > t) = 0.9999 Ha: mean < 0 Ha: mean != 0 Ha: mean > 0 Ho: mean = 0 degrees of freedom = 23525 mean = mean(cum_abn_55) t = -3.7778 cum_a~55 23,526 -.0030143 .0007979 .1223826 -.0045782 -.0014503 Variable Obs Mean Std. Err. Std. Dev. [95% Conf. Interval] One-sample t test

. ttest cum_abn_55 == 0 if dif==0

Pr(T < t) = 0.2147 Pr(|T| > |t|) = 0.4294 Pr(T > t) = 0.7853 Ha: mean < 0 Ha: mean != 0 Ha: mean > 0 Ho: mean = 0 degrees of freedom = 23525 mean = mean(cum_abn_11) t = -0.7903 cum_a~11 23,526 -.000344 .0004353 .0667612 -.0011971 .0005092 Variable Obs Mean Std. Err. Std. Dev. [95% Conf. Interval] One-sample t test

(36)

36 CAR(-1, 3)

CAR(-1, 5)

CAR(-1, 10)

Pr(T < t) = 0.0004 Pr(|T| > |t|) = 0.0008 Pr(T > t) = 0.9996 Ha: mean < 0 Ha: mean != 0 Ha: mean > 0 Ho: mean = 0 degrees of freedom = 23525 mean = mean(cum_abn_110) t = -3.3484 cum_~110 23,526 -.0027333 .0008163 .1252054 -.0043333 -.0011333 Variable Obs Mean Std. Err. Std. Dev. [95% Conf. Interval] One-sample t test

(37)

37

T-tests On All Data (Except Mining Sector)

CAR(-10,10)

CAR(-5, 5)

CAR(-1, 1)

Pr(T < t) = 0.7448 Pr(|T| > |t|) = 0.5104 Pr(T > t) = 0.2552 Ha: mean < 0 Ha: mean != 0 Ha: mean > 0 Ho: mean = 0 degrees of freedom = 17486 mean = mean(cum_abn_1010) t = 0.6582 cum~1010 17,487 .0008406 .0012771 .1688832 -.0016626 .0033439 Variable Obs Mean Std. Err. Std. Dev. [95% Conf. Interval] One-sample t test

. ttest cum_abn_1010 == 0 if dif==0 & mining==0

Pr(T < t) = 0.7173 Pr(|T| > |t|) = 0.5655 Pr(T > t) = 0.2827 Ha: mean < 0 Ha: mean != 0 Ha: mean > 0 Ho: mean = 0 degrees of freedom = 17486 mean = mean(cum_abn_55) t = 0.5747 cum_a~55 17,487 .000549 .0009551 .1263035 -.0013232 .0024211 Variable Obs Mean Std. Err. Std. Dev. [95% Conf. Interval] One-sample t test

(38)

38 CAR(-3, 0)

CAR(-1, 3)

CAR(-1, 5)

Pr(T < t) = 0.9900 Pr(|T| > |t|) = 0.0199 Pr(T > t) = 0.0100 Ha: mean < 0 Ha: mean != 0 Ha: mean > 0 Ho: mean = 0 degrees of freedom = 17486 mean = mean(cum_abn_15) t = 2.3278 cum_a~15 17,487 .0017997 .0007731 .1022382 .0002843 .0033151 Variable Obs Mean Std. Err. Std. Dev. [95% Conf. Interval] One-sample t test

(39)

39 CAR(-1, 10)

T-tests On All Data (Only Mining Sector)

CAR(-10,10)

CAR(-5, 5)

Pr(T < t) = 0.9702 Pr(|T| > |t|) = 0.0596 Pr(T > t) = 0.0298 Ha: mean < 0 Ha: mean != 0 Ha: mean > 0 Ho: mean = 0 degrees of freedom = 17486 mean = mean(cum_abn_110) t = 1.8839 cum_~110 17,487 .0018239 .0009681 .1280223 -.0000737 .0037215 Variable Obs Mean Std. Err. Std. Dev. [95% Conf. Interval] One-sample t test

Pr(T < t) = 0.0000 Pr(|T| > |t|) = 0.0000 Pr(T > t) = 1.0000 Ha: mean < 0 Ha: mean != 0 Ha: mean > 0 Ho: mean = 0 degrees of freedom = 6038 mean = mean(cum_abn_1010) t = -10.8148 cum~1010 6,039 -.0208911 .0019317 .1501156 -.0246779 -.0171042 Variable Obs Mean Std. Err. Std. Dev. [95% Conf. Interval] One-sample t test

Initial Public Offerings and the stock effect of industry rivals : an empirical research on rival valuation effects caused by underpriced IPOs listed on the NASDAQ Stock Market and the New York Stock Exchange