IPO Underpricing and R&D Activity
Evidence from the Swedish Market
Master’s Thesis 30 credits
Programme: Master’s Programme in Accounting
Specialization: Financial Management, Department of Business Studies Uppsala University
Spring Semester of 2021
Date of Submission: 2021-06-02
Andreas Arktedius Viktor Preiman
Supervisor: Katarzyna Cieslak
(this page is intentionally left blank)
ABSTRACT
Historical research on initial public offerings (IPOs) presents strong evidence of underpricing.
This study investigates if there is a relationship between underpricing of IPOs and pre-IPO research and development (R&D) activities within a company. According to the literature, R&D activities have characteristics of information asymmetry and uncertainty, which can increase underpricing. This study’s sample consists of 231 Swedish companies listed on Nasdaq Stockholm and Nasdaq First North between January 2010 and December 2020. Sweden has a strong association with innovation activities such as R&D, and the country’s IPO market has snowballed in recent years, making it a suitable context for the study. To investigate the relationship between underpricing and R&D activities, the study uses an OLS regression. The findings indicate that R&D positively affects underpricing, which is in line with previous studies on other markets. In addition, the study finds evidence of Firm Size, Offer Size, Shares Offered, VCPE backed, and Firm Leverage related to underpricing.
KEYWORDS: IPO, Underpricing, Innovation, R&D Activity, Information Asymmetry, Swedish Stock Market
ACKNOWLEDGEMENT
This is a master's thesis written in the spring of 2021 at Uppsala University. The authors would like to thank each other for a fantastic collaboration during the dissertation. Also, direct a thanks to the other students who have opposed and given valuable feedback. Finally, a big thanks to Katarzyna Cieslak who contributed with valuable time and feedback during the dissertation.
___________________________ ___________________________
Andreas Arktedius Viktor Preiman
TABLE OF CONTENTS
1. INTRODUCTION ... 1
1.1 BACKGROUND ... 1
1.2 PROBLEMATIZATION ... 2
1.3 PURPOSE ... 3
2. LITERATURE REVIEW ... 4
2.1 INITIAL PRICE OFFERING (IPO) ... 4
2.1.1 IPO UNDERPRICING ... 5
2.2 INFORMATION ASYMMETRY ... 6
2.3 INNOVATION ... 7
2.3.1 R&D ACTIVITIES ... 8
2.4 OTHER FACTORS AFFECTING UNDERPRICING ... 10
2.4.1 FIRM SIZE ... 10
2.4.2 SHARES OFFERED & OFFER SIZE ... 10
2.4.3 FIRM MATURITY ... 11
2.4.4 FIRM LEVERAGE ... 11
2.4.5 OWNERSHIP CONCENTRATION ... 12
2.4.6 VENTURE / PRIVATE EQUITY BACKED ... 13
2.5 HYPOTHESIS DEVELOPMENT ... 14
3. METHOD ... 16
3.1 RESEARCH DESIGN ... 16
3.2 STUDY OBJECT ... 16
3.3 COLLECTION OF DATA ... 17
3.4 REGRESSION MODEL ... 19
3.4.1 THE DEPENDENT VARIABLE ... 19
3.4.2 THE INDEPENDENT VARIABLE ... 20
3.4.3 CONTROL VARIABLES ... 20
3.5 OPERATIONALIZATION ... 21
3.6 RELIABILITY AND VALIDITY ... 23
3.7 ISSUES IN A REGRESSION ANALYSIS ... 23
4. RESULT ... 25
4.1 DESCRIPTIVE STATISTIC ... 25
4.1.1 PEARSON CORRELATION MATRIX & VIF ... 27
4.2 REGRESSION RESULT ... 30
4.3 CLASSIFICATION ANALYSIS ... 33
5. DISCUSSION & ANALYSIS ... 38
5.1 UNDERPRICING ON THE SWEDISH STOCK MARKET? ... 38
5.2 DOES R&D ACTIVITY AFFECT UNDERPRICING? ... 39
5.3 HOW DO OTHER VARIABLES AFFECT UNDERPRICING? ... 42
5.4 RESEARCH LIMITATIONS ... 46
6. CONCLUSION ... 47
6.1 FUTURE RESEARCH AREA ... 48
7. REFERENCES ... 49
APPENDIX I - FULL SAMPLE SIZE ... 59
APPENDIX II - SCATTER PLOT OF RESIDUAL AND PREDICTED ... 63
APPENDIX III - RESIDUALS UNDERPRICING ... 64
1
1. INTRODUCTION 1.1 BACKGROUND
In 2020, the IPO market of Nasdaq Stockholm and Nasdaq First North surpassed expectations with a 26 percent increase in IPOs compared to 2019 (EY, 2021). In the latest years, further evidence of the solid Swedish IPO market performance is listing the venture capital firm EQT in 2019, the largest IPO on the Swedish Stock Market in 20 years (Nyemissioner, 2020; SEB, 2019). The trend continued in 2020 and 2021 with highly anticipated IPOs of the digital bank Nordnet and the online marketplace for housing, Hemnet Group (SVD, 2021; DN, 2021). An IPO is a way of receiving equity financing by issuing new securities to a publicly-traded stock exchange for the first time (Bodie, Kane & Marcus, 2018). According to Ritter (1991), it refers to a private company’s transitioning process to become a public corporation.
A central aspect of any IPO is to what price the company’s shares will be issued. Several studies have demonstrated that IPOs are, on average, priced below their intrinsic value, moreover, referred to as underpricing (e.g., Ritter & Welch, 2002; Lowry, Officer & Schwert, 2010;
Loughran, Ritter & Rydqvist, 1994). Regarding the more recent examples of EQT, Nordnet, and Hemnet Group, the level of underpricing amounted to 26, 9, and 44 percent, respectively (EQT, 2019; Nasdaq, 2021c; Nasdaq, 2021d; Nasdaq, 2021e; Cision, 2021; SVD, 2021).
According to Abrahamson and De Ridder (2015), underpricing is observed as a positive first- day (initial) return and is one of the most identified phenomena in corporate finance. Loughran and Ritter (2004) argued that the foundation behind IPO underpricing is information asymmetry enhanced by riskier investments reflected by technological or valuation uncertainty.
In the IPO process, information asymmetry exists between the following parties: inside managers of the company and the outside investors, the underwriter (i.e., the bank) and the issuer of shares, and between investors where certain investors have more information than others (Brau & Fawcett, 2006; Rock, 1986; Berk & DeMarzo, 2017). To attract significant interest in the offer, the investor requires compensation for the information asymmetry by receiving a discount, resulting in a lower IPO price (Baron & Holmstrom, 1980; Baron, 1982;
Loughran & Ritter, 2004). Since the IPOs of EQT, Nordnet, and Hemnet Group resulted in positive initial returns, the authors believe that underpricing is still a relevant research area.
2 1.2 PROBLEMATIZATION
One of the main motives behind an IPO is to create a foundation to increase financial growth and competitiveness (Ritter & Welch, 2002; Zhao-Jansson & Andersson, 2021). An essential factor in achieving this is innovation, where firms raise capital to increase their research and development (R&D) activities (Zhou & Sadeghi, 2019; Zhao-Jansson & Andersson, 2021).
However, due to the unique characteristics of innovation projects, it could be challenging to establish how R&D activities will affect the firm's valuation (Zhou & Sadeghi, 2019).
A definition of innovation is investments and outcomes in new ideas, products, or services (Berk & DeMarzo, 2017). The ambition is that these investments will increase sales and profitability and thus create a higher market value of the firm (Wies & Moorman, 2015; Fang, Tian & Tice, 2014). The importance of being innovative has increased R&D activities over the last decade due to technological development and hence increased requirements for capital (Gharbi, Sahut, & Teulon, 2013; Zhou & Sadeghi, 2019). However, according to Hall (2010), financing innovation is challenging for firms due to the unique characteristics of innovation projects. Innovation projects usually have large capital requirements, which could be hard to manage through internal financing. The consequence is that innovative firms tend to utilize external capital sources (Wies & Moorman, 2015; Wang et al., 2016). In utilizing external capital sources, making an IPO could be more strategic for companies to fund their innovation projects. The stock market provides a more strategic and easily accessible form of financing compared to debt, making it easier for companies to overcome financial difficulties (Bernstein, 2015; Dittmar, 2008).
According to Zhou and Sadeghi (2019), R&D activities are a standard and reliable measure of innovation activities within a company and are used as a proxy to measure companies' input attempts to increase innovative output. However, R&D activities do not have any guarantees to succeed or maintain the same innovation level over several years. As a result, R&D activities are uncertain by nature, which could cause problems such as information asymmetry (ibid.).
The information asymmetry could result in higher costs and tentative returns for the investor since they might have deficient information concerning the new products or services invented by the company. Further, it could affect investors' willingness to provide capital for innovative
3 companies and hence affect the pricing of the IPO (Loughran & Ritter 2004; Alam & Walton, 1995; Lanjouw, Pakes, & Putnam, 1998; Hall, 2010; Lehmann-Hasemeyer & Streb, 2016).
This study focuses on the Swedish Stock Market to contribute to the existing research on how firm innovation measured by R&D activities affects IPO underpricing. Sweden is well associated with innovation and is referred to as the start-up capital in Europe, with Stockholm as the second most prolific tech hub globally (Davidsson, 2015). Furthermore, Sweden spends almost 3.3 percent of its gross domestic product on R&D investments (OECD, 2019;
Sweden.se, 2018). Combining a solid association with innovation and an increasing number of issued IPOs makes the Swedish Stock Market an appropriate focus.
Previous research has studied the relationship between underpricing and innovation in other markets than the Swedish Stock Market. Lehman-Hasemyer and Streb (2016) demonstrated that innovative companies, measured by patents, were associated with low underpricing on the German Stock Exchange compared to previous research. Zhou & Sadeghi (2019) concluded that patents on the Chinese Stock Market resulted in lower underpricing, while higher R&D activities resulted in higher underpricing. Previous studies have found underpricing on the Swedish IPO market, however, not with the focus on innovation and R&D activity (e.g., Abrahamson & De Ridder, 2015; Abrahamson, DeRidder, & Råsbrant, 2011; Loughran, Ritter,
& Rydqvist, 1994; Rydqvist, 1997; Thorsell & Isaksson, 2014).
1.3 PURPOSE
This study aims to provide new evidence and contribute to existing research on IPO underpricing. This study will examine how R&D activities affect underpricing of IPOs on the Swedish Stock Market.
Is there an association between R&D activity and IPO underpricing on the Swedish Stock market?
4
2. LITERATURE REVIEW 2.1 INITIAL PRICE OFFERING (IPO)
The definition of the term IPO is the first time the company sells its shares to institutional and public investors and lists on the stock market with an established offering price. It describes the process of transforming a privately held firm into a public company where anyone can buy and sell the company's stock at the stock exchange (Ritter, 1991). In Sweden, the IPO process consists of different steps, from hiring an underwriter (i.e., the bank) to the first trading day.
The process starts by conducting a so-called "book-building process," where the underwriter contacts more prominent investors to explore the demand for the share at the different offer prices (Lowry, Officer & Schwert, 2010; Dalton, Certo & Daily, 2003). The underwriter offers the remaining shares to the public investors who can participate through different investment banks. This distribution is referred to as the IPO lottery, as not all participants "win" and receive shares (Loughran, Ritter & Rydqvist, 1994). The IPO company and the underwriter determine the decision to allocate shares to investors.
There are several motives behind making an IPO: publicity, expansion, liquidity, and reputation; however, one of the fundamental reasons is to raise capital to enable financial growth (Ritter & Welch, 2002). On rare occasions, the IPO company conducts a direct listing, providing no new capital. Companies like Spotify, Hemnet Group, Palantir Technologies, Coinbase, Slack, and Roblox have recently made it to the stock market through a direct listing.
However, multiple companies utilize the stock market as a source of equity funding to raise capital (ibid.). Companies pursue equity capital since traditional debt could be harder to access as it involves regular costs in the form of interest payments. In addition, the banks that issue debt usually desire collateral in the company in return. However, innovative companies have more intangible assets defined as non-tradable property, making it more challenging for outsiders to evaluate and making it more challenging to receive debt financing. Hence, this makes equity capital more appropriate than debt (Ritter & Welch, 2002; Brau & Fawcett, 2006;
Berk & DeMarzo, 2017; Lowery, 2002). The IPO process has disadvantages that the company requires to consider before going public. Researchers have found that the IPO process is costly, complex, and time-consuming, with an average cost of 7 - 14 percent, based on the IPO proceeds (Dalton, Certo & Daily, 2003; Berk & DeMarzo, 2017). Another problem with IPOs
5 is the disclosure of proprietary information, company strategy, and cost of disclosing financial information, which could affect the competitiveness and innovation of the company (Draho, 2004). Bernstein (2015) argues that there is evidence that private companies might be more innovative than those that go public, as private companies do not have to disclose sensitive information to the shareholders. However, Draho (2004) argues that access to capital compensates for the cost of going public.
Companies will be affected by factors they cannot control by going public, such as market conditions and investor sentiment. Lucas and McDonald (1990) distinguish between the optimistic and pessimistic market as when companies are more willing to make an IPO, referred to as the cyclicality of IPOs. In addition, another market condition is the market timing, which is when the market overvalues the IPO and allows owners to raise more capital by selling off less equity or shares. However, even though market timing could affect the value of the IPO, multiple researchers have found that IPOs are, on average, underpriced over time (e.g., Ibbotson & Jaffe, 1975; Loughran, Ritter & Rydqvist, 1994).
2.1.1 IPO UNDERPRICING
The definition of underpricing is the difference between the IPO offer price and the closing price of the first trading day (Dietrich, 2012; Ritter, 1991). A general explanation of IPO underpricing is asymmetric information about the value or the absolute risk of the company.
For the IPO to attract significant interest, the issuer needs to compensate the investor for the uncertainty regarding the security's value. To increase the demand, the issuer provides a discounted offer price and leaves "money on the table" (Loughran & Ritter, 2004). The issue behind leaving "money on the table" is that the issuer could have raised the same amount of money by giving up less equity if a higher initial price was received (Loughran & Ritter, 2004;
Dietrich, 2012; Brau & Fawcett, 2006). In the Swedish Stock Market, Abrahamson, DeRidder, and Råsbrant (2011) have provided documentation regarding this phenomenon and found that 130 million SEK had been "left on the table" by selling shareholders between 2000-2010.
In Swedish Stock Market, researchers have demonstrated that the first-day return on average is positive. Table I summarizes different researchers that have studied underpricing on the Swedish Stock Market and during which period.
6 TABLE I. FIRST DAY RETURN IN THE SWEDISH STOCK MARKET
Researchers Period Average Underpricing (%)
Abrahamson & DeRidder (2015) 1996-2011 7.68
Abrahamson (2018) 2004-2010 4.0
Abrahamson, DeRidder & Råsbrant (2011) 2000-2010 6.35
Thorsell & Isaksson (2014) 1996-2006 15.0
Rydqvist (1997) 1990-1994 8.0
Loughran, Ritter & Rydqvist (1994) 1970-1991 39.0
Rydqvist (1997) 1980-1990 40.7
Table I. Different researchers that have studied underpricing and the Swedish Stock Market and the average underpricing.
2.2 INFORMATION ASYMMETRY
According to various researchers, information asymmetry is essential when discussing IPOs, underpricing, and innovation (e.g., Berk & DeMarzo, 2017; Aboody & Lev, 2002; Brau &
Fawcett, 2006). Asymmetric information refers to the differences in information about a company between inside managers and outside investors (Berk & DeMarzo, 2017). Welch (1989) argues that managers know more about the company's value and future cash flow than investors. Furthermore, Aboody and Lev (2002) state that managers possess more information regarding the individual asset and that investors had inaccurate information regarding those.
Several studies have established the relationship between information asymmetry and underpricing (e.g., Brau & Fawcett, 2006; Beatty & Ritter, 1986; Chae, 2005; Ritter, 1984).
Rock (1986) developed the winner's curse model to elaborate on this relationship where he subcategorizes investors into two groups: uninformed and informed investors. The informed investors have perfect information about the IPO value, implying that information asymmetry exists between buyers and sellers and among buyers. As a result, the informed investors tend to crowd out uninformed investors from underpriced securities who often manage to bid overprice for new shares. This crowding-out effect implies that if new issues were not, on average underpriced, uninformed investors would realize negative returns and choose to withdraw from the IPO market. Hence underpricing could be used to introduce uninformed investors to the company (ibid.).
7 Brau and Fawcett (2006) argue that a reason for underpricing is the information asymmetry between the underwriter and the issuer of shares. The underwriter has deeper insight and superior market comprehension and exploits this to underprice the offer to create a buy-side of investors, resulting in selling all the shares offered. This would result in a lower IPO price to increase the demand (Baron & Holmstrom, 1980; Baron, 1982). However, Lowery, Officer, and Schwert (2010) argued that younger companies are challenging for underwriters to evaluate since these companies do not fully disclose all information regarding innovation projects. This implies that the underwriter might not underprice the issue to receive benefits for themselves, however, simply because it could be challenging to determine a price that reflects the company's actual value (ibid.).
According to proprietary cost, companies have little incentive to disclose all information due to high competition and preparation costs (Dye, 1986). However, the uninformed investors must be compensated for the risk in the IPO, i.e., by a higher first-day return, due to information asymmetry (Rock, 1986; Brau & Fawcett, 2006; Beatty & Ritter, 1986). The lack of incentive for companies to provide information about their innovation projects and the unique characteristics might cause business and financial uncertainty. This uncertainty would imply that higher levels of innovation cause more underpricing due to more information asymmetry.
However, research has demonstrated a difference in the outcome depending on how to measure innovation (Zhou & Sadeghi, 2019; Dye, 1986; Lev, 2001).
2.3 INNOVATION
The definition of innovation is the ability to implement new ideas, products, or services, including the possibility of successfully creating, recombining, and utilizing new and existing knowledge (Fang, Tian & Tice, 2014). Today a significant part of the innovation investments are focused on technological development. Porter (1992) states that innovation comes from continuously investing in tangible and intangible assets. The term innovation is a broad concept; however, a central part of the process involves searching for new ideas to benefit from commercially. This is a process where firms invest a considerable amount of resources in time, money, and labor capital. The process is long and complex, which involves a high degree of uncertainty and probability of failure (Guo & Zhou, 2016).
8 Companies have a significant incentive to invest in innovation since it is vital for long-term financial growth and competitiveness. Innovation might work as an indicator for investors to help determine the firm’s level of competitiveness and equity value (Fang, Tian & Tice, 2014).
Today, the market is more globalized, and the competition has increased, making innovations more critical than ever since companies must continuously upgrade their competitive advantage (Lafay & Maximin, 2017; Porter, 1992). If innovative companies are investing in the core renewal process of their organization, this could work as protection towards technological shifts and puts the company in an improved position to seize the opportunity of new markets.
Further, if a company manages to convert investments in innovation activities into higher sales or earnings, it should result in a higher market value (Griliches, 1981: Wies & Moorman, 2015).
2.3.1 R&D ACTIVITIES
Several different measurements can interpret innovation. One way is to focus on knowledge creation and intellectual property, measured by R&D activities or patents (Sandner & Block, 2011). Other ways to measure innovation are looking at the number of scientific publications, new product introductions, or industry-specific products (Wies & Moorman, 2015). According to Zhou and Sadeghi (2019), R&D activities are a reliable sign of innovation within a company.
The main difference between R&D activity and patents is that R&D activity should be considered innovation input while patents as innovation output. Consequently, the conception is that R&D activities have a higher level of information asymmetry than patents due to higher risk-taking and uncertainty (Lanjouw, Pakes, & Putnam, 1998; Fang, Tian & Tice, 2014).
Research on R&D activities reveals that it is one of the most used measures of innovation and the attempt to increase innovation output (Lanjouw, Pakes, & Putnam,1998). R&D activity has the advantage of assigning a monetary value to the company's engagement in innovative activities, which is easier to interpret than patents (ibid.). In Sweden, the Annual Account Act (on Swedish: Årsredovisningslagen K3 (BFN 2012:1)) applies to larger companies. Companies that fulfill more than two of the following criteria are required to utilize K3: above 50 employees, more than 50 million in assets, or 80 million in net sales. Companies in Sweden often utilize the K3 Annual Account Act before the IPO and apply it to the IFRS Accounting Standards after the IPO. According to K3 (BFN (2012:1), a company can either expense the cost of R&D activities on the income statement or capitalize it on the balance sheet. If the R&D
9 activity expects to generate economic benefits and the cost can be measured with certainty, capitalization is often used (ibid.). By using capitalization, the company instead does write-offs each year according to write-offs plans. The reason Swedish companies do capitalizations is that it increases profitability. Swedish companies commonly utilize the capitalization approach before the IPO (82 percent of the observations with R&D activity in this study). Further advantages of R&D activities are that it could decrease market share turbulence. This decrease in market turbulence is essential since the competition is increasing and the product life cycle is decreasing. Finally, studies have revealed that R&D activities could lead to a higher market value of the firm as it correlates positively with stock return (Sandner & Block, 2011; Artz et al., 2010; Matraves & Rondi, 2007; KPMG, 2017).
R&D activities have several characteristics that make them different from ordinary investments. A significant part of the R&D activity consists of investing in the researchers or engineers of the company. Their knowledge creates an intangible asset that contributes to the firm's knowledge base and hopefully generates future profits. However, the main issue with R&D activities is that higher R&D activities do not automatically result in higher and more successful innovation output (Hall & Lerner, 2010). Hence, according to Lev (2001), R&D activities could be defined as high-risk non-tradable property that could be hard for the market to evaluate. Consequently, a company that engages in extensive R&D activities should, therefore, receive a higher underpricing to increase demand and compensate for the investor's risk. Ultimately the innovation success depends on how effectively the managers and researchers of the company can transform the innovation into future profits (Zhou & Sadeghi, 2019; Brau & Fawcett, 2006).
To overcome the problems with R&D activities, the market urges companies to disclose more information about them to decrease the information asymmetry. However, due to proprietary cost, companies limit disclosing relevant financial information to the stock market as the preparation and competitive costs are high (Dye, 1986). Studies have found that companies that provide more voluntary information about their R&D activities receive more attention and a higher value for their shares due to lower underpricing (Zhou & Sadeghi, 2019). This additional information can increase the investor's comprehension regarding the R&D activity, and the asymmetry between the investor and the company would decrease, making it easier for investors to evaluate the company. Even though the investor's ability to react to new
10 information can explain the effect on returns originating from R&D activities, it could moreover result from the above-market risk-taking (Lev, 2001).
2.4 OTHER FACTORS AFFECTING UNDERPRICING
The following section will present seven different variables, which according to previous research affect underpricing. The study includes the following variables: Firm Size, Shares Offered, Offer Size, Firm Maturity, Firm Leverage, Ownership Concentration, and Venture / Private Equity backed. The reason behind the chosen variables is that they occur on several occasions in previous research regarding underpricing.
2.4.1 FIRM SIZE
Firm size is considered an essential factor affecting underpricing. Rogers (2004) discussed that large companies have an advantage over smaller companies because of higher cash flow, human capital, and easier access to capital. The problem with information asymmetry moreover appears to be lower regarding larger companies. A possible explanation is that larger companies are exposed to the public for a longer time before the IPO. As a result, they are more known by the investors, and hence, the information asymmetry and underpricing should be lower.
However, the exciting aspect is that more small firms have begun increasing their innovative activity in recent years, making it easier to adapt to new opportunities. However, researchers have divided opinions regarding firm size and underpricing since positive and negative relationships have been found (Protogerou, Calaghiou, & Vonortas, 2017; Ritter, 1991).
However, on the Swedish Stock Market, Abrahamson and DeRidder (2015) found a positive relationship between underpricing and Firm Size between 1996-2011.
2.4.2 SHARES OFFERED & OFFER SIZE
The number of shares offered to the market can affect the underpricing function. Rock (1986) discusses the demand for shares in IPOs to be critical in the IPO process, as the underwriter desires to sell all shares offered. Companies issue different amounts of shares, and this affects the demand for them (ibid.). Lowery, Officer, and Schwert (2010) demonstrated that companies with a lower number of shares offered to the market had a higher underpricing. Furthermore,
11 researchers have found that smaller share offerings will be more challenging for underwriters to evaluate (Baron & Holmstrom, 1980; Baron, 1982). Baron and Holmstrom (1980) argued that underwriters have limited information regarding smaller share offerings, and hence they cannot exploit the underpricing function. The underwriter's information is consistent with uninformed investors information, and thus the underpricing process is less valuable. Baron and Holmstrom (1980) further argued that to increase the demand and sell these amounts of shares offered, the underwriters issue a lower offer price when they have superior information regarding the company's valuation. Consistent and related to shares offered to the market, the IPO size additionally affects underpricing (Ritter, 1991; Zhou & Sadeghi, 2019). Abrahamnson and DeRidder (2015) demonstrated that the IPO size, defined as the number of shares sold multiplied by the offer price, had a negative effect on underpricing in the Swedish Stock Market between 1996-2011.
2.4.3 FIRM MATURITY
Firm maturity is another factor that has proven to be essential for underpricing (Protogerou, Calaghirou, & Vonortas, 2017). Dietrich (2012) argues that the company's maturity is a business risk, which discloses higher underpricing at less mature companies. The underlying fundamental is that mature companies have received a more extensive customer base and business knowledge. Therefore, the more mature companies have an easier way of receiving capital and hence be less underpriced because of minor information asymmetry. Less information asymmetry is because the public has more information regarding the company and the products/services that they have been offering to the market for a more extended period.
Furthermore, since companies are in varying steps in the life cycle, they are assumed to have different risk profiles. Growth companies are in an earlier phase in the cycle and are associated with higher risk, while established companies are more mature and associated with less risk, which will result in higher versus lower underpricing (Damodaran, 2009).
2.4.4 FIRM LEVERAGE
There are suggestions that the company's leverage can affect the underpricing of the IPO since it might indicate financial quality (Habib & Ljungqvist, 2001). Since debt financing is a riskier option, managers could utilize this to indicate higher firm quality, resulting in lower
12 underpricing (ibid.). According to James and Wier (1990), higher leverage increases managers' requirements to be more responsible for the companies' finances due to higher budget constraints. This is a way to reduce the risk of managers being distortive of the company's cash flow since they have to meet their debt obligations (ibid.). The utilizing of leverage could also help mitigate one of the significant drivers of underpricing: information asymmetry (Cai, Ramchand & Warga, 2004). Companies that issue debt before the IPO tend to be more known by capital markets, and information about them is more easily accessible (ibid.). Further, companies with higher pre-IPO leverage are usually older with more historical data on finances, products, and operations available, reducing investors' risk (ibid.). Based on these assumptions Cai, Ramchand, and Warga (2004) compared firms with and without pre-IPO leverage and found a negative relationship between leverage and underpricing. These results and arguments highlight the evolution of capital structure theory since the Irrelevance theory by Miller and Modigliani (1958), which argues that the capital structure does not affect the firm's value;
instead, the return of the investments is in focus.
In an attempt to question the generalizability of the argumentation by James and Wier (1990) that higher leverage could be a positive signal due to budget constraints, Kim, Pukthuanthong, and Walker (2008) focused specifically on differences between high-tech and low-tech firms.
In contradiction to prior research, the results indicate that it could be misleading to state that higher leverage will lower underpricing regarding all firms (ibid.). By comparing high-tech with low tech-firms, the results indicate that higher leverage only serves as a positive signal and results in lower underpricing for low-tech firms. The relationship between leverage and underpricing reverses for high-tech firms since leverage might imply increased uncertainty and risk for these types of firms (ibid.). Therefore, more innovative firms should utilize equity to reduce the risk of financial distress and hence have lower firm leverage and underpricing. On the other hand, less innovative firms should utilize more debt to lower underpricing (ibid.).
2.4.5 OWNERSHIP CONCENTRATION
The ownership concentration has been studied and argued to affect the underpricing of IPOs (Fernando, Krishnamurthy & Spindt, 2004; Stoughton & Zechner, 1998). The ownership structure refers to the dilution of control among several shareholders. A possible explanation of the phenomenon of ownership structure is the Agency Theory, which declares that the shareholders and the agent do not have the same interest and incentive (Jensen & Meckling,
13 1976). However, as agency cost exists, the ownership structure will be underpriced from inside managers to attract more prominent investors. The reason is that larger investors will act as the monitor for outside investors to reduce information asymmetry. Stoughton and Zechner (1998) argued that shareholders favor more significant ownership concentration in the IPO, as larger concentration can decrease the information asymmetry between inside managers and outside investors. However, as investors seek investments with a broad ownership structure and a liquid secondary market, the investors will account for this in the IPO price (Booth & Chua, 1996).
The reason is that the improved liquidity would result in a lower rate of return (i.e., underpricing) and hence a higher equilibrium of the share price (ibid.). Therefore the percentage concentration of capital is more appropriate to investigate as companies tend to hold higher voting rights after the IPO.
Sweden is one of the most common countries with dual-class shares. Hence, the investors who participate in IPOs are assumed not to take a leading position in the ownership structure and are more interested in liquidity than buying more substantial voter rights shares (Cronqvist &
Nilsson 2003). Furthermore, researchers have argued that a dilution of shareholders should decrease the information asymmetry between the investors, as it creates noise, and that this could mitigate the agency conflict (Holmström & Tirole, 1993; Shleifer & Vishny; 1997).
Furthermore, a more extensive ownership structure can reduce their interest and have an information advantage over the outsiders. However, the ownership concentration tends to negatively correlate with underpricing as investors perceive that a higher ownership concentration results in less liquidity, increased agency conflicts, and less information asymmetry between the inside managers and outside investors (Dietrich, 2012).
2.4.6 VENTURE / PRIVATE EQUITY BACKED
Research suggests that corporate governance characteristics can reduce the underpricing of an IPO. One governance characteristic is having a certifying investor such as a private equity (PE) or venture capital (VC) firm backing the IPO. VC firms focus on smaller companies where they obtain significant equity positions to influence management in the early stages of the life- cycle. PE firms instead utilize a leveraged buyout strategy and target more mature companies with stable cash flows (Bruton, Chahine & Filatotchev, 2009; Megginson and Weiss, 1991;
Barry et al., 1990). Megginson and Weiss (1991) compared VC-backed to non-VC-backed
14 IPOs and found significantly lower first-day returns of VC-backed IPOs. The interpretation should be that VC firms can certify the firm's actual value and reduce underpricing (ibid.).
Barry et al. (1990) found that ownership and the number of VC invested before the IPO correlated negatively with underpricing since this reduces information asymmetry. VC and PE firms have a strong incentive to reduce information asymmetry as it will establish a credible reputation for the IPO to access the market on favorable terms. Since the market might perceive that the VC fund has more insight and a strong incentive to monitor the activities within the IPO, investors might find VC-backed IPOs to be an improved investment (Megginson & Weiss, 1991).
Recent research suggests that VC-backed IPOs suffer from higher degrees of underpricing (Lee
& Wahal, 2004). A study found that VC-backed IPOs experience significantly higher first-day returns than non-backed IPOs. The average difference is estimated to be five to ten percentage points and even more pronounced in the bubble period between 1999-2000. The underpricing effect is less significant regarding younger VC firms and those that have conducted fewer IPOs (ibid.).
2.5 HYPOTHESIS DEVELOPMENT
There has been a significant amount of research regarding the underpricing of IPOs (e.g., Ritter
& Welch, 2002; Lowry, Officer & Schwert, 2010; Loughran, Ritter & Rydqvist, 1994). Even though researchers argued that underpricing is one of the most well-identified phenomena in financial theory, further research is required to increase the comprehension of the subject behind the drivers of underpricing (Abrahamson & De Ridder 2015; Ljungqvist, 2007).
An essential aspect of underpricing is information asymmetry. The investor receives a discounted offer price as compensation for the increased risk due to uncertainty of the company's fundamental value (Loughran & Ritter, 2004; Brau & Fawcett, 2006; Chae, 2005).
Previous research has stated that information asymmetry appears to be more severe regarding investments reflected by technological or innovation uncertainty. A standard proxy used to interpret innovation is R&D activities, defined as a high-risk non-tradable property that could be hard for the market to evaluate. As a result, the previous research finds higher underpricing
15 in IPO companies with higher R&D activities (Loughran & Ritter 2004; Lev, 2001; Zhou &
Sadeghi, 2019).
As described in Table I, underpricing on the Swedish Stock Market is well documented;
however, not focusing on innovation and R&D activity. Since Sweden is well associated with innovation and technological development, the authors believe this should be an appropriate focus (Davidsson, 2015; OECD, 2019). Hence, to explore the underpricing of IPOs regarding innovative companies on the Swedish Stock Market, the following null and alternative hypothesis is developed:
H0 = There is no relationship between IPO underpricing and R&D activity on the Swedish Stock Market.
HA = There is a positive relationship between IPO underpricing and R&D activity on the Swedish Stock Market.
16
3. METHOD 3.1 RESEARCH DESIGN
This study applied a quantitative research design characterized by a deductive approach to investigate if R&D activities impact the underpricing of IPOs on the Swedish Stock Market.
Quantitative studies are characterized by numerical data connected between different theories or concepts by using a deductive approach. By applying a deductive approach, the study’s hypothesis, which builds on previous research, can be tested to explore the relationship between the theoretical framework and the results. Furthermore, the advantages of conducting a quantitative study are that the information is straightforward, unbiased, testable, and a large amount of data can be processed. The disadvantages of a quantitative study are that the data can be misleading, no follow-up of the data, and it can generate a relatively simple explanation to a complex phenomenon (Bryman & Bell, 2015). Previous researchers that have studied underpricing on the Swedish Stock Market have used similar approaches with a quantitative method characterized by a deductive approach (e.g., Abrahamson & De Ridder, 2015;
Abrahamson, DeRidder, & Råsbrant, 2011).
3.2 STUDY OBJECT
To study the relationship between underpricing and R&D activity, the study focused on companies listed on the Nasdaq Stockholm between January 2010 and December 2020. Nasdaq Stockholm is divided into three segments, Large, Mid, and Small Cap, depending on their market capitalization. The study also included the Nasdaq First North, which focuses more on growth companies with lower market capitalization than the other segments. It is, moreover, less regulated and has fewer requirements than the more extensive lists of Nasdaq. Hence, a listing on Nasdaq First North is more accessible and less costly, which might attract more growth companies that engage more in innovative activities and was therefore appropriate to include (Damodaran, 2009; Nasdaq, 2020a; Nasdaq, 2020b).
17 3.3 COLLECTION OF DATA
The data collection was conducted by using Refinitive Eikon ("Eikon") and was analyzed through IBM SPSS ("SPSS"), which is a standard program from statistical methods within social science (Pallant, 2016). The data has been stored in Microsoft Excel during the collection time and is referred to as panel data as it involves measurements over time. Using Eikon, the authors applied filters to only collect data from before the IPO occasion. The data collected from Eikon consisted of the issue date, name, offer price, first-day percentual change, year founded, VCPE backed companies, and the shares offered to the market. Data regarding the company's R&D activities defined either by capitalization or cost expenses, assets, debt, equity, and ownership concentration has been manually collected from each IPO prospectus. The authors have read all the prospectus related to the IPO to ensure that the IPO is a "pure IPO"
and no list exchange. By reading all the prospectus, no direct listing was included in the sample.
Since data from before the IPO occasion of some companies in the presented period was not available in Eikon, a manual data collection of those companies was conducted instead of decreasing the number of observations. These data sets were collected from either the prospectus or Nasdaq Stockholm's official website. The financial data from the prospectus are categorized as primary data. The advantages of primary data are that the information is "first- hand" and is assessed to be less unbiased compared to secondary data (Bryman & Bell, 2015).
Hence, the researchers are aware that the financial information in each prospectus could be incorrect or misleading. However, as certified auditors have partly or fully audited the prospectus, the assessment is that the data is representable for this study. The authors have moreover reviewed the financial data set individually to reduce the risk of human measurement and interpretation errors. Both authors moreover examined the secondary data collected from Eikon individually to prevent any data to be incorrect.
18 TABLE II. SAMPLE SIZE
Sorted by Number of Observations
IPOs in Swedish Stock Market Jan 2010 - Dec 2020 307
(Less) IPOs; canceled, announced, rumor -4
(Less) Duplicates -11
(Less) Not on Nasdaq Stockholm and Nasdaq First North -61
Sum Observations 231
Table II. Demonstrates the final sample size of the study, and how many observations that has been excluded.
In Eikon, the following filters were applied to retrieve IPOs in the Swedish Stock Market;
deals, equity, target market (Sweden), issue date (1 Jan 2010 - 31 Dec 2020), and issue type (IPO). The first sample size that was retrieved from Eikon included 307 observations. The 307 included all pure IPOs from the Swedish Stock Market that Eikon had data about before the IPO, including Spotlight and Aktietorget. The authors are aware that the total number of IPOs in the presented period exceeds the sample size. Based on the study object of this study, sorting of the dataset was conducted: IPOs that have not been conducted (e.g,. announced, rumor, unknown, canceled) were removed. IPOs that were reported twice in the dataset were removed.
IPOs on market lists that were not on the Nasdaq Stockholm or Nasdaq First North were removed. After the sorting, the final dataset resulted in 231 observations presented in Table II.
The final dataset of all companies included in this study will be presented in Appendix I.
Tabachnick and Fidell (2013) argue that a sample size of N > 50 + 8 * V is required to conduct a regression analysis (V = Variables). With 11 different variables in the study (incl. dummies), the sample size requires to be at least 138 observations, and with this study’s 231, the assessment is that the sample size is sufficient. The sample size of previous research on underpricing in the Swedish Stock Markethase between 105 and 172 observations on a 10 to 15-year time period (e.g. Abrahamson & DeRidder, 2015; Abrahamson, DeRidder & Råsbrant, 2011; Thorsell & Isaksson, 2014). These studies focused on either fewer market segments or more, e.g. only the Nasdaq Stockholm, or extended with Nasdaq First North, Aktietorget, and Spotlight. However, the authors assess that the sample size regarding underpricing on the Swedish Stock Market is above the comparative studies, and hence are sufficient to generate generalizability.
19 3.4 REGRESSION MODEL
The quantitative study has been conducted through linear regression analysis. Linear regression is a technique used to explore the relationship between one dependent and one independent variable (Pallant, 2016). The goal of the regression model is to reject or accept hypotheses on a significant level one, five, or ten percent (ibid.). More specifically, the linear regression used was an Ordinary Least Square (OLS) regression, a linear least method used to estimate the coefficients to the minimum sum of residual squared (Newbold, Carlson & Thorne, 2013). The advantages of using OLS are that it is easy to utilize and applicable, can process a large data set, and can be used to assume the maximum likelihood. The disadvantages are instead that OLS can simplify the complex “real world” phenomenon, sensitive to outliers, heteroscedasticity, and can have multicollinearity (Pallant, 2016). To overcome the disadvantages of an OLS regression, appropriate approaches was used, presented in Section 3.7. Further, the result was analyzed on different levels through a classification analysis with a similar approach as Zhou and Sadeghi (2019). The reason to include an additional test was to more thoroughly examine different groups of RDA and their effect on underpricing.
The study additionally included t-tests, which is a method to test hypotheses (Bryman & Bell, 2015). As this study assumed that both positive and negative relationships could occur on the variable, a two-tailed test was more appropriate than a one-tailed t-test. However, the authors expected a one-tailed result from the independent variable RDA, however, a two-tailed test was included to ensure consistency throughout the study. The goals of the t-test are to accept/reject or the hypothesis on a significance level one, five, or ten percent, which equals critical values of +/- 1.645, 1.960, and 2.576 (Newbold, Carlson & Thorne, 2013). Related to the regression model, the unstandardized beta coefficient is presented.
3.4.1 THE DEPENDENT VARIABLE
The purpose of this study was to examine if R&D activity affects underpricing of IPOs and hence the dependent variable is underpricing. The level of underpricing was measured by the initial first-day return, which is defined as the difference between the closing price and the offer price on the first day of trading (Ritter, 1991). The mathematical equation is:
20 𝐼𝑅 = ( 𝑝!"# − 𝑝$%% ) / 𝑝$%%
where:
IR= the percentual underpricing from the IPO.
pEnd = the end price of the first trading day from the IPO.
pOffer = the offer price from the IPO.
3.4.2 THE INDEPENDENT VARIABLE
The independent variable used to capture underpricing of innovative companies on the Swedish Stock Market was R&D activity. The level of innovation was measured through capitalization (RDCap) or expense (RDExp) and scaled by total assets (Zhou & Sadeghi, 2019). In Sweden companies typically follow the K3 Annual Acts before the IPO and have the possibility to choose between the methods of capitalization and expensing R&D costs (BFN 2012:1). The reason to scale by total assets on both methods is to preserve a consistent measurement and because the majority (82 percent) of the observation in this study originated from the capitalization method. Based on the adjustment to Swedish Account Act and formulas from Zhou and Sadeghi (2019), the following mathematical equation is used.
RDA = ( 𝐶𝑎𝑝𝑖𝑡𝑎𝑙𝑖𝑧𝑒𝑑 𝑅&𝐷 + 𝐸𝑥𝑝𝑒𝑛𝑠𝑒 𝑅&𝐷 ) / Total Assets
3.4.3 CONTROL VARIABLES
To increase the reliability of the test, the authors added control variables to the regression model (Newbold, Carlson & Thorne, 2013). The reason is that other variables might influence the dependent and independent variables and result in spurious results if these were excluded. As relevant variables were included, this reduced the omitted variable bias, which is that multiple relevant variables are excluded from the regression model (Bryman & Bell, 2015). This study included ten control variables, and hence the omitted variable bias is assessed to be limited by the number of relevant variables. The added control variables originate from previous research, where other researchers have found a relationship or effect between underpricing and that particular variable (e.g Abrahamson & DeRidder, 2015; Zhou & Sadeghi, 2019). The variables included as control variables are: Firm Size, Shares Offered, IPO-Size, Firm Maturity, Firm
21 Leverage, Ownership Concentration, Venture Capital, or Private Equity backed IPO. The study included year dummy and industry dummy in line with research from Zhou and Sadeghi (2019). To eliminate the possibility of omitted dummy variables or dummy traps, this study excluded the year “2010” and “Materials” (Bryman & Bell, 2015). The year dummy captured time-related effects and was helpful as IPOs are assumed to have a cyclicality and hence differences in underpricing. The industry dummy was instead used to capture the differences in firms’ specific characteristics, as different companies are assumed to have differences in RDA. A further reason to include an industry dummy is to capture incremental effects as there might be differences in IPO underpricing between industries, driven by industry-specific characteristics. The study adopted the classification from Global Industry Classification Standard (GICS) as it is one commonly used industry classification and is additionally used in the Nasdaq Stockholm (Chung, Hrazdil & Li, 2017). The GICS included the following eleven industries: Energy, Materials, Industries, Consumer Discretionary, Consumer Staples, Health Care, Financials, Information Technology, Communication Services, Utilities, and Real Estate.
With all variables the following mathematical equations is used.
𝐼𝑅& = 𝛽'+ 𝛽(𝑅𝐷𝐴& + 𝛽*𝐿𝑛(𝐴𝑠𝑠𝑒𝑡)& + 𝛽+𝐿𝑛(𝑆ℎ𝑎𝑟𝑒𝑠)& + 𝛽,𝐿𝑛(𝑆𝑖𝑧𝑒)& + 𝛽-𝐿𝑛(𝐴𝑔𝑒)&
+ + 𝛽.𝐿𝐸𝑉& + 𝛽/𝑂𝑊𝑁𝐶& + 𝛽0𝑉𝐶𝑃𝐸& + 𝛽1𝑌𝐸𝐴𝑅& + 𝛽('𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌&
+ 𝜀23
3.5 OPERATIONALIZATION
Concepts are the foundation of theory and essential in business research. A quantitative study is to estimate the degree of relationship between different concepts. Operationalization is how researchers measure different concepts where different indicators can be used. These indicators can derive from different sources, where previous research is a common approach (Bryman &
Bell, 2015). The concepts in this study were based on the previous research and theory within the area of underpricing. To view how this study measured different variables, a table with the concept and their formula is provided (Table III below).
22 TABLE III. VARIABLE DEFINITION
CONCEPT VARIABLE FORMULA
UNDERPRICING IR ( Stock Pricet+1- Offer Pricet ) / Offer Pricet
Equals the percentage change between the offer price and first-day closing price.
(Ritter, 1991)
R&D ACTIVITY RDA ( Capitalized R&D + Expense R&D ) / Total Assets
Equals the sum of innovation through capitalization and expensing, dividend by firm's total asset (Zhou & Sadeghi, 2019)
FIRM SIZE Ln(Asset) LN (ASSET)
Equals the natural logarithm of the firm total assets in the IPO.
(Fernando, Krishnamurthy & Spindy, 2004)
SHARES OFFERED Ln(Shares) LN (SHARES)
Equals the natural logarithm of the total shares offered in the IPO.
(Lowry, Officer & Schwert, 2010)
OFFER SIZE Ln(Size) LN (OFFER PRICE * SHARES OFFERED)
Equals the natural logarithm of the IPO price multiplied by the shares offered.
(Abrahamson & DeRidder, 2015)
FIRM MATURITY Ln(Age) LN (1+ (YEAR IPO - YEAR FOUNDED))
Equals the natural logarithm of the (one plus) year the company makes IPO minus the year the company was founded. (Ritter, 1991)
FIRM LEVERAGE LEV DEBT / ( EQUITY + DEBT )
Equals the leverage through debt dividend with equity plus debt before the IPO.
OWNERSHIP CONCENTRATION
OWNC Total Shares of Largest Owner / Total Shares
Equals the largest owners percentual shares divided by the total shares outstanding before the IPO.
VC/PE VCPE 1 = VCPE BACKED, 0 = OTHERWISE
A dummy variable equals one if the company is VCPE backed, otherwise zero.
(Zhou & Sadeghi, 2019)
YEAR DUMMY YEAR 1 = IPO YEAR, 0 = OTHERWISE
A dummy variable that will equal one for the IPO year otherwise zero. The year 2010 is excluded to create the dummy. (Zhou & Sadeghi, 2019)
INDUSTRY DUMMY INDUSTRY 1 = SPECIFIC INDUSTRY, 0 = OTHERWISE
A dummy variable that will equal one for the industry of the IPO company, otherwise zero.
Industry “Materials” are excluded to create the dummy (Zhou & Sadeghi, 2019).
Table III. The operationalization of the study, the concept, variable, and the formula to calculate the indicator.
23 3.6 RELIABILITY AND VALIDITY
Reliability refers to measuring a concept, where a quantitative study focused on consistency, reproducibility, and accuracy (Bryman & Bell, 2015). As this study aimed to examine the relationship between underpricing and RDA, the authors considered these concepts related to the part of reliability. Consistency is how the concepts are assessed to be stable over time, and related to underpricing and RDA, the variance over time is assessed to be low as underpricing and RDA are not affected by time. Hence, the assessment is that consistency of the study is high and that the result will be consistent even in the future. The authors investigated the test results and then discussed if a similar perception of the data has been received. This personal investigation increased the reproducibility of the study. The accuracy in the study of underpricing is a well-established research area, where the authors have adopted similar approaches as others in the area to capture underpricing (e.g., Ritter, 1991; Abrahamson &
DeRidder, 2015; Zhou & Sadeghi, 2019)
Validity refers to how accurate the measurement of a concept is. To increase the validity, the authors required to fit the formula for each concept, consequently that it captures the phenomenon. This study adopted multiple formulas for public articles assessed to be well- established proxies to capture each concept. Furthermore, there are five different types of validity, however, the “construct validity” is assessed to be the essential one, as it refers to measuring the intended concepts in the study. As this quantitative study is deductive, the hypothesis was drawn from the previous research to view relevant concepts to investigate. This study drew one hypothesis from the theory section, where the previous research determined the concepts that are being measured. There is typically a possibility that the hypothesis is misguided by previous research, however, as there has been extensive research regarding underpricing, the conclusion drawn from is assessed to be valid for this study. Furthermore, the hypothesis determined the data collection, which reduced the researcher’s involvement in the process. This exclusion will increase the validity of the test (Bryman & Bell, 2015).
3.7 ISSUES IN A REGRESSION ANALYSIS
According to Newbold, Carlson, and Thorne (2013), a regression analysis is sensitive to outliers, linearity, multicollinearity, and heteroscedasticity. Therefore, researchers are required
24 to address these issues before conducting a regression model to assure more accurate results.
A normal distribution of the data set can be used to overcome the issue of outliers. To assume a normal distribution, the researchers can look at the skewness and kurtosis. Skewness refers to if the data set is skewed to either the right or left, and kurtosis refers to the peakiness of the data set (Bryman & Bell, 2015). Another way of testing normal distribution is through a Kolmogorov-Smirnov and Shapiro-Wilk test where a significance below p-value 0.05 indicates the critical value for a normal distribution (ibid.). Researchers can remove outliers if examined and judge why they are removed or adjusted in the data set (Newbold, Carlson & Thorne, 2013). A winsorizing of 2.5 and 97.5 percentiles were applied to overcome the problem of outliers. Winsorizing is used to reduce outliers and receive a more normally distributed data set instead of trimming and removing the observations (Tukey, 1962).
Linearity is the assumption that a linear relationship exists between the dependent and independent variables. To test this, a scatterplot can be used to assume a linear relationship (ibid.). Multicollinearity refers to the correlation between the independent variables. The reason is that a correlation between independent variables affects the outcome of the regression and hence can be misleading or be predicted by the same variable. A Variance Inflation Factor (VIF) and a Pearson Correlation Matrix can be used to examine multicollinearity.
Multicollinearity occurs when the Pearson correlation is +/- 0.9 and the VIF > 10 with a desirable value of VIF < 5 (ibid.). The authors removed variables from the regression model if it correlates +/- 0.9 or has VIF values above 10. Heteroscedasticity is additionally an assumption when making a regression analysis. In a regression analysis, the researchers assume that the variance of all independent variables is similar. To examine heteroscedasticity, a scatterplot between the residuals from the independent variables and the predicted values from the regression was used (Appendix II). To increase the regression accuracy and reduce the disadvantages of the selected method, tests regarding the normal distribution, outlier, linearity, multicollinearity, and heteroscedasticity were conducted. The test is presented in the following section.
25
4. RESULT
4.1 DESCRIPTIVE STATISTIC
Table IV presents a summary of the data regarding the underpricing. Table IV includes the IPO year, number of IPOs each year, the mean when underpricing is winsorized each year (Mean 2), the median, minimum, and maximum underpricing. Median, minimum, and maximum underpricing is without the winsorized sample.
TABLE IV. UNDERPRICING 2010-2020
Year n Mean Mean (2) Median Min Max
2010 5 0.0255 0.0255 -0.0091 -0.1176 0.1714
2011 9 0.0147 0.0147 -0.0276 -0.2615 0.7176
2012 2 0.4167 0.4167 0.4167 0.1667 0.6667
2013 3 -0.2684 -0.2189 -0.3333 -0.4906 0.0188
2014 26 0.0256 0.0215 -0.0226 -0.3688 1.3333
2015 48 0.1434 0.1357 0.0533 -0.4056 1.6353
2016 40 0.1410 0.1297 0.0225 -0.4077 1.7133
2017 37 0.1329 0.1343 0.0967 -0.3913 1.1031
2018 22 0.4265 0.0696 0.0013 -0.3126 6.6000
2019 18 0.1436 0.1331 0.0894 -0.3144 1.3864
2020 21 0.0796 0.0953 0.0968 -0.6727 0.6911
Sum 231 0.1387 0.1021 0.0283 -0.6727 6.6000
Table IV. Demonstrates the underpricing in the observation sample during 2010-2020, the number of IPO each year in the sample, and the mean, the mean of the winsorized sample at 2.5 and 97.5 percentile (Mean 2), median, the minimum, and maximum value for each year. The last row represents a sum of the mean or median, or minimum or maximum value.
Table IV demonstrates that during the years 2010-2020, there are 231 IPOs in the sample. The year with the highest number of observations originates from 2015 with 48 IPOs and the lowest amount in 2012 with 2 IPOs. The arithmetical mean during the complete sample is 13.87 percent, with the highest in 2018 (42.65 percent) and the lowest in 2013 (-26.84 percent). The winsorized mean for the complete sample is 10.21 percent, with the highest in 2012 (41.67
26 percent) and the lowest in 2013 (-21.89 percent). The median in the complete sample is 2.83 percent, with the highest value in 2012 (41.67 percent) and lowest in 2013 (-33.33 percent).
The minimum underpricing is -67.27 percent in 2020, and the highest minimum was 16.67 percent in 2012. The maximum underpricing was 660.00 percent in 2018, and the lowest maximum return was 1.88 percent in 2013.
TABLE V. DESCRIPTION OF VARIABLES
Variable n Mean Median Min Max Std Dev
IR 231 0.1021 0.0283 -0.3421 1.1987 0.3116
RDA 231 0.2076 0.0201 0.0000 1.9413 0.3998
Ln(Shares) 231 15.688 15.679 13.259 18.173 1.2501
Ln(SIZE) 231 18.935 18.834 16.327 22.140 1.6635
VCPE 231 0.0692 0.0000 0.0000 1.0000 0.544
Ln(Age) 231 2.4054 2.3978 0.0000 4.4461 0.9508
Ln(Asset) 231 19.067 18.5703 15.639 23.588 2.2466
LEV 231 0.5299 0.5451 0.0362 1.0880 0.2887
OWNC 231 0.4560 0.3710 0.0970 1.0000 0.2806
Table V. Demonstrates all the variables used in the study. Including the number of observations, mean, median, minimum, maximum, and standard deviation of each variable. Control variables year and industry dummy excluding from the table. Winsorizing is used in Microsoft Excel using the function “Percentile”.
In Table V, IR and RDA variables are winsorized as their kurtosis and skewness were outside the normal distribution values. After winsorizing the data, the IR decreases from 13.87 to 10.21 percent. The median is the same as winsorizing only affects the outliers. Hence, the minimum increases from - 67.27 percent to - 34.21 percent. The maximum underpricing instead decreases from 660.00 percent to 119.87 percent. Furthermore, the mean RDA is 20.76 percent, indicating that the Swedish Stock Market's IPO companies, on average, spend 20.76 percent on R&D activities either through capitalization or expensing. The maximum RDA is 194.13 percent, and the minimum is zero. The VCPE is on average 6.92 percent in the sample, indicating a low amount of Venture Capital or Private Equity-backed IPOs in the Swedish Stock Market. Companies had an LEV average of 52.99 percent, with a maximum value of
