Msc. Thesis Business administration: Entrepreneurship & Innovation ‘Information asymmetry, venture capital and the start-up at and after the time of
the Initial Public Offering’
Student: Rick Waldron Student number: 10373446 Supervisor: Dr. A.S. Alexiev Date: June 23rd 2017 Number of words: 13498
2
Statement of originality:
This document is written by Student Rick Waldron who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
List of contents
Abstract 4
1. Introduction 5-8
2. Literature review 9-20
2.1 The investor-start-up relationship 9-13
2.2 IPO pricing and performance 13-14
2.3 Information asymmetry 14
2.4 Information asymmetry between the start-up and the investor 14-15 2.5 Information asymmetry during an IPO 15-16
2.6 Research questions 16
2.7 Hypotheses 16-19
2.8 Conceptual model 19-20
3. Methodology 21-33
3.1 Measures 21-24
3.2 Sample and data collection 24-26
3.3 Summary statistics 27-29
3.4 Method and model specification 29-33
4. Results 34-47
5. Discussion, limitations, and suggestions 48-53
5.1 Discussion 48-51
5.2 Limitations and suggestions 51-53
6. Conclusion 54-55
7. Reference list 56-60
4
Abstract
Does information asymmetry affect stock returns after an IPO? How do venture capitalists play a role in information asymmetry after an IPO? This study investigates 246 start-up companies in the high-technology industry that went public between 2011 and 2015, where 123 companies where venture capital backed and the other 123 companies were not. The results of this research show that non-venture capital backed companies outperform venture capital backed companies and that the total sample of IPOs underperformed compared to different benchmarks. Although one effect from information asymmetry turns out to have the expected effect on stock performance, a real conclusive answer on the effects of information asymmetry on stock performance cannot be drawn. However the evidence is found that venture capitalist investment in a start-up does decrease the effects of information asymmetry. Although this influence of venture capital on information asymmetry depended on the influence of information asymmetry on stock performance.
5
1. Introduction
Entrepreneurs usually start their adventure with a problem they encounter on a daily basis. They realize that the problem can be solved or overcome by a simple solution, which others do not see. They view the problem as an opportunity and try to create a business out of it. Entrepreneurs use all of their existing resources, knowledge, capacities and close network. Sometimes this results in the capability to start the venture, but often, entrepreneurs are not able to sustain it due to the lack of necessary finances (Gompers, 1994; Muñoz-Bullon, Sanchez-Bueno, & Vos-Saz, 2015; Packalen,
2007).
In order to sustain the business, entrepreneurs must obtain the resources elsewhere. They pursue investors and financing in order to grow. This poses a threat, because investors might have a different view on the business, want to go in a different direction or they are just to opportunistic. This can lead to an entrepreneur who is not willing to fully disclose his problem, opportunity and solution. Entrepreneurs are scared to give out the complete view and information they possess, because of the fact that investors can take on the venture by themselves (Shane & Venkataraman, 2000). This is where the problem of information asymmetry between the start-up and the investor starts. Information asymmetry is a situation in which one party has more information than the other. This is often the case with investors and entrepreneurs, especially during an initial public offering (IPO), where every investor can invest in the company. The information asymmetry problem in this case consists of vital company information which is not known to the investors. This is caused by the fact that prior to the IPO companies do not have to publish their information and performance, since they are private companies. IPO investors therefore do not exactly know how a company performed and future performance will therefore be hard to predict, this is especially
6 the case for recently founded companies. Investors could therefore be less interested in the companies, due to the lack of information. This could potentially drive down the return of the stock. A higher degree of information asymmetry can therefore harm the return of the stock.
The theory on information asymmetry has extensively been researched in academic literature, as well as the performance of IPOs. However how information asymmetry influences the performance of IPOs has not been researched extensively. It has been proven that information asymmetry influences the pricing of an IPO, however up until now there has not been proof of influence on stock returns after an IPO. This is important to explore, because previous research found that IPOs underperform relative to different benchmarks (Berk & Peterle, 2015; Ritter, 1991). This underperformance might in fact be caused by information asymmetry, and this process is therefore relevant to explore.
Additionally, looking at the difference between venture capital backed start-ups and non-venture capital backed start-ups is also a point of interest. The reason behind this is that having a venture capitalist on board increases trust of IPO investors, due to the fact that venture capitalists only invest in the best start-ups. On top of that, venture capitalists bring experience into the company, therefore it is expected that a connection between information asymmetry and venture capitalists exists. This makes the difference between venture capital backed start-ups and non-venture capital backed start-ups interesting to investigate.
What this study aims to do is to see if IPOs performed differently than two benchmarks and to see if there exists difference in returns if a company was backed by venture capital or not. These findings will contribute in understanding the most important parts of this research, namely the influence of information asymmetry on the
7 stock return after a start-up went public, and the influence of venture capital investment on information asymmetry. This will be especially interesting for start-ups that decide to go public shortly after they were founded, the reasoning behind this is that relatively not much is known about these companies to investors. This is because of their short track record, this will increase information asymmetry and interest of this research. From this focus on start-ups came an interest in the high-technology market, this is because start-ups within this industry tend to grow extremely fast, which makes this industry especially interesting to research.
These interests resulted in a main research question; does information asymmetry affect stock returns after an IPO? This question specifically addresses the impact of information asymmetry on the performance of IPOs, which is a contribution to the scientific literature. To investigate whether venture capitalists decrease information asymmetry a sub-question is proposed; how do venture capitalists play a role in information asymmetry after an IPO? This sub-question addresses the influence of venture capitalists on information asymmetry. By combining these two questions the impact of information asymmetry on IPO performance and the influence of venture capitalists on information asymmetry can be measured, and the research gap can be narrowed.
In order to answer these questions, this research will contain quantitative analyses. The dataset in this analyses contains a sample of 246 IPOs after cleaning the data. For each IPO specific information was collected regarding the proxy variables. The proxy variables were grounded on existing theory. The information of the proxy variables was obtained using ThomsonOne. The other variables were obtained using different databases, namely; Datastream, ThomsonOne and Kenneth French’s database. After which multiple different types of analyses were performed, such as an event study
8 and different multiple regression analyses.
The obtained results showed that the stock returns of companies one year after their IPO were significantly underperforming the benchmarks. However, the results indicated that this was due to the venture capital backed sub-sample, which performed worse than the non-venture capital backed sub-sample, which was the opposite result of what was expected. No clear results on the influence of information asymmetry on IPO performance was found doing this research. The results showed that venture capitalists do decrease information asymmetry. However, this does not show a clear meaning since information asymmetry does not appear to play a significant role in stock returns after an IPO.
This paper continues in the following way. The next section contains a review of existing literature, where the entrepreneur-investor relationship is extensively described. Thereafter, insights in IPO pricing and performance will be mentioned. After which, the theory on information asymmetry, information asymmetry between the entrepreneur-investor and information asymmetry during IPOs is described. Subsequently, the research questions and several hypotheses will be presented as a guideline for the rest of the study. The second section ends with a conceptual model, which illustrates the proposed hypotheses. The third section will consist of a discussion of the used measures, the sample selection criteria, an elaboration on data collection and summary statistics, as well as the model specifications. In the fourth section the results will be given. The fifth section contains a discussion of the obtained results, answers to the research questions, furthermore implications of the research and suggestions for future research are given in this section. While the last section summarizes the complete research and in addition contributions of the research will be given.
9
2. Literature review
This literature review contains different sub-sections, in which an aspect of prior literature is described. The aspects and sub-sections together form the base of this research and each section thus contributes to the overall picture of this study. The first section will describe the relationship that a start-up builds with an investor, in particular with a venture capitalist. The second section elaborates on the setting up, the pricing and the performance of an IPO. While the third section quickly recaps information asymmetry. The fourth section elaborates on the information asymmetry between the venture capitalist and the start-up. The fifth section discusses information asymmetry during an IPO. The last three sections elaborate on the research questions, the hypotheses and the conceptual model.
2.1 The investor-start-up relationship
This section describes the relation between the start-up and the investor. After viewing a problem as an opportunity and creating the solution an entrepreneur might need an investor, because he or she often lacks the proper resources to sustain the business (Gompers, 1994; Muñoz-Bullon et al., 2015; Packalen, 2007). To find an investor, let alone the right investor might be one of the most difficult tasks an entrepreneur faces. The reason for this is that different investors have different views and insights, and some might be more suited for him than others. Investors often give advice on business decisions, but they might give wrong strategic advice, and this can lead to decline in growth (Steier & Greenwood, 1995). In addition, an investor might be able to let a founding team member go, this might harm the relationship between the investor and the entrepreneur. This can also have an effect on the trust between the entrepreneur
10 and the investor (Rosenstein, Bruno, Bygrave, & Taylor, 1993). This might lead to business decisions, which are not profitable, and this can also decrease information supply to the investor. Getting mixed up with a powerful investor thus has its benefits but it also comes with a certain amount of danger. At the end of the day if the entrepreneur wants to grow his business then he should obtain the necessary resources to do so, and thus building a relationship with an investor, might be necessary, this is also viewed in literature as a necessary evil (Villanueva, Van de Ven, & Sapienza, 2012). The relationship they develop exists as long as the investor is invested in the business and his resources are still being used.
On the other side of the coin there is the investor that looks at this problem with a different perspective. The investor, which most likely will be a venture capitalist, a business angel or corporate investors, view this as an opportunity for both the investor as well as the start-up (Denis, 2004; Maula, Autio, & Murray, 2005). They will try to build a relation that goes further than just financing. Investors, especially investor firms, often add additional resources that can consist of recruiting new talented managers, new strategies and have a network that consist of multiple resources, which were unobtainable before (Ragozzino & Blevins, 2016). A venture capital firm has a network of professional and industry contacts, this will contribute to building the first contacts (De Clercq, Fried, Lehtonen, & Sapienza, 2006). The most used resources of a venture capitalist are; being a board to the founding team, obtaining more alternative sources of equity and debt financing, monitoring financial and operational performance, and interfacing with the investor group (Macmillan, Kulow, & Khoylian, 1989). In addition, they also have an excellent business sense, and they can help with structuring the next financing rounds, this can be both debt or equity financing, which can for example be an IPO (De Clercq et al., 2006). Venture capital investment thus brings additional
11 experience in the venture. Using these resources indicates a form of trust towards the future investors, and this shows signals of low information asymmetry between the venture capitalist and the entrepreneur. Indirect benefits from investor involvement are that it sends a vital message to the world, displaying the quality of the start-up (Brav & Gompers, 1997). In this research the focus will lie on venture capitalist firms as investors, in general they possess the most and best resources for entrepreneurs to use. In addition, venture capital firms only invest in around one percent of the actual new venture proposals they receive (Megginson & Weiss, 1991). They also build a reputation in which they signal to invest in a reliable and profitable new firm. Next to these advantages, a venture capital firm can also be of moral support, in which they act as a trustee (De Clercq et al., 2006). This clearly displays the quality of the start-up, which venture capital firms invest in. Venture capital firms are thus often seen as a certification of a good start-up.
A venture capitalist will thus only invest in high potential start-ups and in return they will receive equity in the start-up. The downside of any investor for the entrepreneur, and especially a venture capitalist, is that they have a clear exit strategy. They usually exit between four to seven years (Black & Gilson, 1998). The reason for this is that the venture capitalist wants return on their investment. This return is generated by investing in and growing the business, and then selling their stake for more money. There are usually two exit strategies for a venture capitalist, either going public by doing an IPO, or selling to another firm, which is called an acquisition. Unlikely types of an exit are management buyout, buyback, write-off or a secondary sale (Black & Gilson, 1998; Cumming & Johan, 2008; Guo, Lou, & Pérez-Castrillo, 2015). If the entrepreneur wants to stay in control of the firm, he should either buy out the venture capitalist or bring the company public. With an IPO the company stays independent, and
12 ownership of the company becomes differentiated, since every person can buy a share. This also means that the entrepreneur stays in his managerial role. While with a merger or acquisition the company changes owners. This can end up in a conflict between the entrepreneur and the existing venture capitalist (Black & Gilson, 1998). Thus often the best outcome for an entrepreneur is an IPO. An IPO also increases their liquidity; improve managerial skills because of the establishment of new teams and a scale up of their offering (Ragozzino & Blevins, 2016). An IPO is also the most favorable exit strategy for the venture capitalist, because this will leave him with the highest return on investment (Gompers, 1995). On top of that, the venture capitalist can choose between multiple return on investment strategies with an IPO. He can decide if he wants to sell all his equity, partially or no equity at all, in which he will end up being a majority shareholder. This strategy can also generate high returns, because share prices might increase the years following the IPO and this would mean that the value of his equity stake increases. This is often a strategy that a venture capitalist makes use of. In addition, this implies that they are not cashing out, which goes to show that the firms’ insiders stay within the business and this sends a positive message to other potential investors (Gompers & Lerner, 2001).
If an entrepreneur dreams of going public, then he should get mixed up with a venture capitalist. The reason for this is that a new venture is 42% more likely to go public with the support of a venture capitalist (Ragozzino & Blevins, 2016). As well as the fact that new ventures that went public did so when the firm was significantly younger and had a greater book value of assets when supported by a venture capitalist, compared to the non-venture capital backed new ventures that went public (Megginson & Weiss, 1991).
13 corporation when investing in or acquiring the venture. Thus information asymmetry will be higher with an IPO (Cumming & Johan, 2008). This is the main reason this research focuses on information asymmetry during an IPO. On top of that, the certifying role for a venture capitalist will be greater due to the dispersed ownership after an IPO. Thus IPOs will be most interesting to investigate. The next section will elaborate on the physics of an IPO.
2.2 IPO pricing and performance
This section will start with a short introduction of how an IPO takes place. The initial step for a company is to select an underwriter or multiple underwriters to help with the IPO (Katti & Phani, 2016). ‘An underwriter is a company or other entity that administers the public issuance and distribution of securities from a corporation or other issuing body’, this is often a bank or financial institution (Investopedia, 2017b). The underwriter starts by doing research, the underwriter does macroeconomic research and industry specific research. After this step, the actual underwriting takes place, in which the draft prospectus will be filed with the authority. The third step consists of simple marketing, where the underwriter creates awareness among investors about the IPO. The next part is the price determination, in which the investors send out bids to the underwriter. These bids are taken into consideration, but eventually the underwriter and the company will decide what the final offer price will be. In the next phase the shares will be allocated to the different investor types. The final phase is the actual listing; this usually takes place a week after the allocation phase is completed (Katti & Phani, 2016). IPO performance has been extensively researched in historical literature, and different outcomes have been measured. A famous study from 1991 showed that IPOs were underperforming the market. The three-year return after the closing price of
14 the first trading day was almost 30% lower than the returns of an average of 1526 companies, with an average return of 34% compared to 61% by the 1526 companies (Ritter, 1991). Compared to several indexes IPOs underperformed as well (Ritter, 1991). Another study showed that the fact that it was an IPO was not the reason for underperformance, but the type of firm played a more important role (Brav & Gompers, 1997). The next section contains general aspects regarding information asymmetry.
2.3 Information asymmetry
The origin of information asymmetry lies in agency theory. It is the problem where the principal cannot accurately observe what the agent is doing. The principal does not possess the same information the agent has (Eisenhardt, 1989). This can also be seen in the entrepreneur – investor relationship, where the agent is the entrepreneur and the investor is the principal. The next two sections contain aspects of information asymmetry that are vital to this research.
2.4 Information asymmetry between the start-up and the investor
This section elaborates on the information asymmetry between the start-up and the initial investor. As was mentioned in the introduction, some people recognize different opportunities because of their different views, and others cannot see those opportunities yet (Shane, 2000). If everyone had the same beliefs and information then opportunities would not exist. If this would be the case then the opportunity would immediately be removed because of competitors (Schumpeter, 1934). So, in the case of an investor, if he had the same information as the entrepreneur, then he would simply adjust the price of his resources to an extent that entrepreneurial profits would not exist anymore. In this way the investor can capture the entrepreneurial profits himself
15 (Shane & Venkataraman, 2000). Information asymmetry plays a big role at the start of the relationship for the investor. The investor does not know the entrepreneurs intentions before he makes the investment and uncertainty is thus high at this point. The investor will always be cautious during the relationship, but uncertainty will go down during the relationship (Amit, Glosten, & Muller, 1990; Chan, Siegel, & Thakor, 1990). Nearing the end of the relationship uncertainty switches towards the entrepreneur. Entrepreneurs do not know the exit strategy of the venture capitalist and this might end up in an unfortunate way for the entrepreneur (Black & Gilson, 1998).
The growth potential of venture capital involvement is something that is tempting for entrepreneurs, but the fact that venture capitalists have an exit strategy for every venture raises concerns for an entrepreneur. The relationship that a start-up builds with his investor is therefore interesting to explore. The venture capitalists wants profit in return for their initial investment and this has an impact on the relationship. Information asymmetry plays a big part in this relationship during the whole investment period. Due to information asymmetry the start-up will have an ever-changing relationship with the investor, especially nearing the end of the relationship where the entrepreneur faces the exit of the venture capitalist, by doing an IPO for example. This will be further discussed in the next section.
2.5 Information asymmetry during an IPO
Information asymmetry during an IPO can be observed at two different points. The first one is between the issuer or start-up and underwriter, and the second one is between the underwriter and potential investors (Katti & Phani, 2016). The underwriter can have an incentive to set a lower offer price, therefore attracting more investors, and this lowers the amount of marketing that needs to be done by the underwriter (Katti &
16 Phani, 2016; Rock, 1986). This paper does not try to see if information asymmetry has an effect on IPO pricing, therefore while conducting this research assumed was that the issuer and underwriter have the same role. The main purpose of this research is to see whether information asymmetry has an influence on IPO performance. Prior research has found that it has an influence on IPO pricing, but the effect on performance is still questionable (Katti & Phani, 2016; Rock, 1986). The next section contains the research questions.
2.6 Research questions
The presented literature review resulted in the following main question; does information asymmetry affect stock returns after an IPO? A sub-question is raised to see if a venture capitalist has influence on information asymmetry; How do venture capitalists play a role in information asymmetry after an IPO? The next section contains the hypotheses needed to answer the research questions.
2.7 Hypotheses
This research focuses on the returns of IPOs with regard to venture capital involvement and information asymmetry. These two aspects can only be measured and explained if the returns are calculated and compared. Therefore the first step in this research is to investigate how IPOs are performing, this can be done by comparing them to the return of the market. This is done by using two indexes or benchmarks. Section 2.1 described that venture capitalists only invest in high quality start-ups. While section 2.2 described that IPOs underperformed compared to the benchmarks, except for IPOs with venture capitalist backing, which do not underperform. These findings were backed up previous
17 literature (Berk & Peterle, 2015; Brav & Gompers, 1997; Olsson & Östman, 2015; Ritter, 1991). Taking previous research into account the following hypotheses were developed.
Hypothesis 1a: IPOs underperform compared to the benchmarks.
Hypothesis 1b: Venture capital backed IPOs have similar performance compared to the
benchmarks.
Hypothesis 1c: Non-venture capital backed IPOs underperform compared to the
benchmarks.
The above hypotheses compare the IPO returns with the benchmarks, however to compare the two sub-samples with each other another hypothesis is needed.
It has already been proven that start-ups are more likely to go public when they are backed up by venture capital (Ragozzino & Blevins, 2016). In addition, research has shown that stock performance increases when a venture capitalist has invested in the company (Brav & Gompers, 1997). This brings us to the next hypothesis.
Hypothesis 2: The start-up will have higher stock returns after going public if a venture
capitalist is involved
Up until now, the stated hypotheses will give insight into how IPOs perform relative to the benchmarks and how a venture capitalist influences stock returns. This results in a better understanding of the performance of IPOs, and hereby a more in-depth answer to the research questions can be developed. However to answer the two research questions additional hypotheses are required.
The relation with an investor consists of several points where information asymmetry plays a big part. The first part of information asymmetry is before the actual
18 investment is made. It is the part where entrepreneurs are looking for potential investors. Information asymmetry makes it difficult to acquire the needed resources. The reason for this is that the entrepreneur will not give out all the information because this will make his idea pursuable for others. The result of this is that investors do not have the same information as the entrepreneurs at the time of the initial investment (Shane & Cable, 2002). Therefore, investors might view the opportunity differently because of the lack of information, and this could result in a lower valuation of the opportunity, as was also explained in section 2.4. This problem can also occur during an IPO, potential new investors do not know how the company performed in the past and therefore it is hard to predict future performance. This can decrease interest of investors, lowering the return of the stock. In addition, it has been found that information asymmetry creates underpricing, which is a negative effect (Katti & Phani, 2016; Rock, 1986). The unavailability of knowledge about past performance and the general negative effect of information asymmetry might harm the return of an IPO. To investigate this effect, the following hypothesis is developed.
Hypothesis 3: Information asymmetry has a negative effect on stock return in the first
year after an IPO
Section 2.1 described that researchers found that a venture capitalist act as a quality certification of the start-up. That they can be seen as a quality guarantee of the start-up, otherwise known as certifier or certification. By representing a new venture, the venture capitalist guarantees the quality of a start-up, this lowers the information asymmetry in a way, which would not be done without the involvement of the venture capitalist (Megginson & Weiss, 1991). Section 2.5 described that in the case of an IPO information asymmetry leads to underpricing, and thus having a venture capitalist on
19 board, as a certification of the start-up will lower the costs of information asymmetry and decreases underpricing, thus lowers the cost of going public. On top of that, venture capitalists will also attract long-term investors and this increases return and share price (Lerner, 1994; Megginson & Weiss, 1991). This results in the following hypothesis.
Hypothesis 4: A venture capitalist decreases the negative effects of information
asymmetry.
The results of the first two hypothesis will show how the IPOs performed, this will help in understanding how information asymmetry can influence the IPOs. Combining the first hypotheses with the last will give the ability to answer the research questions accordingly.
The next section contains the conceptual model.
2.8 Conceptual model
The former hypotheses lead to a conceptual model with different aspects. The conceptual model illustrates the relations between the theories in this research, figure 1 displays this conceptual model. Most relations are assigned with a sign. The arrows indicate that an effect on a concept is expected while the lines indicate that the two concepts will be compared. The total sample of IPOs and the non-venture capital backed sub-sample are expected to underperform compared to the benchmarks, therefore they are both assigned with a minus. The venture capital backed sub-sample is expected to have similar performance, therefore no sign has been assigned. Venture capital backed IPOs are expected to outperform the non-venture capital backed IPOs, therefore a plus has been assigned. While information asymmetry is expected to have a negative influence on the stock returns after an IPO, therefore in the model an arrow with a
20 minus can be seen. Moreover, the venture capitalist is expected to decrease the negative effects of information asymmetry during an IPO, therefore another arrow with a minus can be observed.
Figure 1. Conceptual model of the hypotheses
21
3. Methodology
The methodology starts with a sub-section where the measures will be discussed, after which the sample and data collection will be highlighted. The third sub-section will consist of the summary statistics, while the fourth sub-section elaborates on the empirical models used in this research.
3.1 Measures
To be able to answer the different hypotheses several measures must be highlighted. The two measures that will be used to answer hypothesis 1 are the cumulative average benchmark-adjusted returns (CAR) and the wealth relatives (WR). The first measure shows how the IPOs perform compared to the benchmarks, while the second measure shows how much an investor will earn by holding a portfolio of stocks for a certain amount of time compared to holding the benchmarks for that same period (Ritter, 1991). The two benchmarks used are the S&P500 and the Nasdaq.
The second hypothesis can be answered by comparing the yearly returns of venture capital backed IPOs and the non-venture capital backed IPOs. This had to be done in order to see if venture capitalists have a positive influence on stock performance. In addition, the yearly returns will also be the dependent variable for hypothesis 3.
The former two measures relate to the performance of IPOs with and without venture capitalist backing, and thus to the first two hypotheses. If the test results of the first two hypotheses show that IPOs underperform, than information asymmetry might be the reason for this. Therefore, hypothesis 3 requires measures for information asymmetry. In this research three measures are used that act as a proxy for information asymmetry. The first variable that is added to act as a proxy for information asymmetry
22 is initial offer size. The size of the initial offer is based on the amount of shares offered times the initial offer price, this is the share price that was set by the underwriter and company before the IPO. There are several reasons for initial offer size as a proxy variable. Firstly, there is more information available to investors about IPOs that are large in offer size (Katti & Phani, 2016). Thus there exists less information asymmetry between the potential investors and the company. In addition, small individual investors are more likely to invest in small companies, since institutional shareholders often do not take positions in small companies, because they can only invest with a relatively low amount of money since there are trade restrictions of large shareholders (Brav & Gompers, 1997). While small individual investors do not have the resources to fully analyze an issuing company, this could also lead to information asymmetry. Therefore size will be negatively related to information asymmetry, the bigger the initial offer the less information asymmetry.
The second information asymmetry measure is underwriter reputation. Underwriters often acquire a reputation; this reputation attracts long-term investors (Katti & Phani, 2016). Underwriters with a high reputation signal better pricing of the offer and supply better information to investors, thus there will be less information asymmetry (Carter & Manaster, 1990). A variable will be added, that will consist of the reputation of the lead underwriter. The reputation will be a number that is based on the rank that was given to them in prior research (Loughran & Ritter, 2004). This ranking system starts with a 1 as the lowest possible score and a 9.1 as the highest possible score. If the lead underwriter does not have a rank, he will be given a score of 1. A higher rank thus implies less information asymmetry and higher expected returns.
The last measure that acts as a proxy of information asymmetry is the number of underwriters for the IPO. Underwriters are a signal to potential investors; together with
23 the company they determine the price and do marketing for the IPO. More underwriters mean more marketing, greater investor reach and more fair price determination (Carter & Manaster, 1990; Katti & Phani, 2016). Therefore the amount of underwriters will negatively affect information asymmetry, thus more underwriters are expected to increase stock returns.
The last two measures will be two control variables that will control for different market movements. They are the Fama-French factors market return minus risk-free rate (RMRF) and small minus big (SMB). The RMRF is the return of the global market minus the risk-free rate, better known as the market premium. While the SMB calculates the spread in return between large and small stocks, based on market capitalization, it is also known as the size premium or small firm effect, because small firms tend to outperform large firms.
The dependent variable will be the return of the IPOs, two different regressions will be performed, based on yearly returns and daily returns. More information about the complete model can be found in section 3.4.
In order to answer hypothesis 4 another variable had to be created, this was a dummy variable for venture capitalist investment. The dependent variables for this model are the information asymmetry proxy variables, while the venture capitalist variable is a dummy variable that takes on the value one if a venture capitalist has invested in the company and zero otherwise. This measure shows if and by how much a venture capitalist influences the effects of information asymmetry. Section 3.4 elaborates on the methods. Table 1 shows a small recap of all the measures and their abbreviations.
24 Table 1. Summary of the measures
Measure Summary Abbreviation in the model
Cumulative average benchmark-adjusted return
Shows the daily performance of IPOs compared to the benchmarks for event day 1 to 252.
CAR
Wealth relatives Shows how much an investor will earn by holding onto the IPO stocks over a certain period of time compared to the benchmarks.
WR
Yearly returns and daily returns
The yearly and daily returns will be the dependent variable for hypothesis 3, in addition, the yearly returns of the two sub-samples will also be compared for hypothesis 2.
Ri
Initial offer size Proxy variable for information asymmetry, which measures the initial IPO offer size. Will be measured using the natural logarithm of the offer size.
LnSize
Underwriter reputation
Proxy variable for information asymmetry, which measures the underwriter reputation by using the rank of the lead underwriter assigned by prior research. Will be measured using the natural logarithm of rank.
LnRitterRank
Number of underwriters
Proxy variable for information asymmetry, which measures the number of underwriters. Will be measured using the natural logarithm of the number of underwriters
LnNumuw
Market return minus risk-free rate or market premium
Controls for the movement of the market. RMRF
Small Minus Big or size premium
Controls for the spread in market capitalization. SMB Venture capitalist This will be a dummy variable set to one if a venture
capitalist has invested in the company, to measure the effect on the information asymmetry proxies.
VC
The next section contains information about the data.
3.2 Sample and data collection
This section elaborates on the sample and on how the data was handled. This research contains a sample of all high-technology firms that went public between January 1st, 2011 and January 1st, 2015, four full years. The ThomsonOne database had a function that selected all the IPOs within this industry. In order to focus this research on IPOs; follow-ons, spinoffs, and divestitures were excluded. This study focuses on information
25 asymmetry between entrepreneurs and their investors, therefore only start-ups were included. This research calls companies’ a start-up if a company went public within ten years of their founding date. Therefore only companies that were founded after January 1st, 2003 were included. This means that on average a company cannot be older than ten years, and thus is considered a start-up. ThomsonOne offered a function in which could be seen whether a venture capitalist had invested or not. After finding the companies using ThomsonOne further information on the companies had to be found. The total number of underwriters, the names of the lead underwriters and the size of the offer could be found using ThomsonOne. The ISIN codes and Sedol codes were also found, they were necessary to find data on stock prices
After importing the data to Excel, the data had to be cleaned, which resulted in a sample of 252 IPOs. In which 127 were backed by venture capitalists, and 125 were not. Using the ISIN and Sedol codes the stock prices could be found. This research used Datastream to do so, in particular, the Excel add-on. Using the daily stock prices the yearly and daily returns were calculated, this was done using equation 1. The first year return was calculated with a share price from the second day of trading, this is because on the first day of trading the market and investor show their response on the IPO and new stock. Therefore the initial offer price is often one that does not reflect the actual share value (Miller & Reilly, 1987).
𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 =𝑃𝑃1−𝑃𝑃0
𝑃𝑃0 ∗ 100% (1)
The returns for the IPOs and the two indexes could be calculated using equation 1, because the daily prices were known. For the two Fama-French factors this was not possible. The factors are returns already and can be found on a daily, monthly or yearly basis. The data on the two variables can be found on Kenneth French’s website, the daily
26 global factors were taken (French, 2017).
In order to exactly match yearly returns of the two factors to the yearly returns of the IPOs, the daily returns of the factors had to be transformed to yearly returns. This was done using Excel and Stata, first in Excel the daily returns were exactly matched to each other for each observation, so that the dates of the IPO exactly match the dates of the two factors. In Stata, the yearly returns for these factors were computed using the daily returns. The actual formula for doing this is given in equation 2.
𝑅𝑅𝑅𝑅𝑅𝑅𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦 = (∏ (𝑅𝑅𝑦𝑦𝑡𝑡𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
100 + 1) 𝑡𝑡=252
𝑑𝑑𝑦𝑦𝑦𝑦 − 1) ∗ 100% (2)
Which would eventually end up being:
𝑅𝑅𝑅𝑅𝑅𝑅𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦 = ��( 𝑅𝑅𝑦𝑦𝑡𝑡100𝑑𝑑𝑑𝑑𝑑𝑑1+ 1� ∗ �𝑅𝑅𝑦𝑦𝑡𝑡100𝑑𝑑𝑑𝑑𝑑𝑑2+ 1� ∗ … ∗ �𝑅𝑅𝑦𝑦𝑡𝑡100𝑑𝑑𝑑𝑑𝑑𝑑252+ 1)� − 1� ∗ 100% (3)
The daily return had to be divided by 100, because the returns were already given in percentages.
After finding the stock data for the two samples four outliers were found, two in each sub-sample. In the sub-sample with non-venture capital backed IPOs these were PhotoAmigo and Millenium with yearly returns of 3900% and 952.63% respectively. In the sub-sample of venture capital backed IPOs these were FFRi and ColoPL, with 749.26% and 712.76% respectively. To answer the hypotheses the sub-samples had to be of the same size, therefore two more observations of the venture capitalist backed sample had to be deleted. In order to keep the sample from human ‘cherry-picking’ two observations were randomly deleted. This was done in Stata, using the function; ‘Sample, count *number of observations’; this number was 123. Because this is the amount of IPOs in each sub-sample.
27
3.3 Summary statistics
This section shows the summary statistics. On average there are 252 trading days each year; this follows from the following calculation:
𝑇𝑇𝑅𝑅𝑇𝑇𝑇𝑇𝑇𝑇𝑅𝑅𝑇𝑇 𝑇𝑇𝑇𝑇𝑑𝑑𝑑𝑑 = 365.25 ∗ �57� − 9 = 252 (4)
Where trading days is calculated by the average amount of days within a year times the proportion of workdays every week minus holiday days; in which the stock market is closed. This is the average amount of trading days that is used by researchers for the Nasdaq and most American stocks; this research used this approach for every stock.
The trading days of different indexes were matched to each IPO, this had to be done in order to calculate the CAR.
To give an overview on the values that were found for most variables the summary statistics for the whole sample are given in table 2. The average first-year returns for the stocks are given in the first row, the second row is the market premium and the third row is the size premium. The last three rows are statistics on the yearly returns of the S&P500, the Nasdaq composite index and the statistics on the offer size; the offer size is in million dollars.
Table 2. Summary statistics total sample Variable Mean Standard deviation Median Yearly returns 0.16888 64.43579 -13.0929 Market premium 8.662602 10.16012 6.508185 Size premium -2.3178 3.584183 -3.03918 S&P500 return 11.20198 7.513652 11.79552 Nasdaq return 14.56293 8.172306 15.2107 Offer size 182.2911 1050.545 57.659
28 In table 3 and table 4 the statistics on the sub-samples can be found. The statistics for the indexes and Fama-French factors also change, because they are linked to start and end dates of IPOs.
Table 3. Summary statistics non-venture capital backed sample Variable Mean Standard deviation Median Yearly returns 7.481294 67.07907 -6.17284 Market premium 8.562943 10.96517 6.46436 Size premium -1.79459 3.645848 -2.49251 S&P500 return 10.95411 8.373901 11.75867 Nasdaq return 14.88737 9.267144 15.37148 Offer size 73.18125 167.7781 18.63
Table 4. Summary statistics venture capital backed sample Variable Mean Standard deviation Median Yearly returns -7.14353 61.07826 -22.4442 Market premium 8.76226 9.329901 6.55201 Size premium -2.84101 3.457432 -3.62933 S&P500 return 11.44985 6.567082 11.83236 Nasdaq return 14.23849 6.930251 15.0407 Offer size 291.4009 1471.116 88
From the statistics can be drawn that the returns of non-venture capital backed IPOs is higher than those of venture capital backed IPOs. While the size of the offer is on average higher for venture capital backed IPOs. In the results section more can be found on these observations.
29 Table 5. Correlation between variables
Yearly returns Market premium Size premium S&P500 return Nasdaq return Offer size Number of under- writers Under writer reputatio n Venture capitalist Yearly returns 1.0000 Market premium 0.2133 1.0000 Size premium 0.2085 0.2164 1.0000 S&P500 return 0.1949 0.8739 -0.0703 1.0000 Nasdaq return 0.1685 0.7019 0.1749 0.8513 1.0000 Offer size -0.0390 0.1128 -0.0525 0.1102 0.0603 1.0000 Number of underwriters -0.0506 0.0815 0.0096 0.1035 0.1317 0.5905 1.0000 Underwriter reputation -0.0968 0.0233 0.0231 0.0096 0.0833 0.1425 0.5379 1.0000 Venture capitalist -0.1137 0.0098 -0.1463 0.0331 -0.0398 0.1041 0.2427 0.5351 1.0000
These correlations are based on the actual values, while for some variables the natural logarithms will be taken in the model, this can also change correlations.
3.4 Method and model specification
This section elaborates on the different methods that are used to answer the hypotheses.
In order to answer hypothesis 1a, 1b and 1c the CAR and the WR had to be calculated. This type of research is called an event study, this can be done by using so-called ‘event periods’, where in this research event days are used, previous research followed the same method, however, by using event months. The CAR is calculated by following the next steps.
30 First, the benchmark-adjusted return (ar) needs to be calculated, this can be done using equation 5.
𝑇𝑇𝑅𝑅𝑖𝑖𝑡𝑡 = 𝑅𝑅𝑖𝑖𝑡𝑡− 𝑅𝑅𝑚𝑚𝑡𝑡 (5)
Where 𝑇𝑇𝑅𝑅𝑖𝑖𝑡𝑡 is the benchmark-adjusted return for stock i on event day t, 𝑅𝑅𝑖𝑖𝑡𝑡 is the return of stock i on time t and 𝑅𝑅𝑚𝑚𝑡𝑡 is the return of benchmark on time t. The average benchmark-adjusted return (AR) is calculated using equation 6:
𝐴𝐴𝑅𝑅𝑡𝑡= 1𝑛𝑛∑𝑛𝑛𝑖𝑖=1𝑇𝑇𝑅𝑅𝑖𝑖𝑡𝑡 (6)
Where n is the amount of IPOs on time t. The calculation of 𝐴𝐴𝑅𝑅𝑡𝑡 leads to the calculation of CAR. Where CAR is simply the summation of 𝐴𝐴𝑅𝑅𝑡𝑡 and this equation looks like:
𝐶𝐶𝐴𝐴𝑅𝑅𝑞𝑞,𝑠𝑠= ∑𝑠𝑠𝑡𝑡=𝑞𝑞𝑇𝑇𝑅𝑅𝑖𝑖𝑡𝑡 (7)
Where q and s are the start and end event days, in this research q and s are 1 and 252 respectively, and thus looks like:
𝐶𝐶𝐴𝐴𝑅𝑅1,252= ∑1𝑡𝑡=252𝑇𝑇𝑅𝑅𝑖𝑖𝑡𝑡 (8)
The second measure is the WR, WR shows how much an investor will earn by holding the IPOs compared to the benchmark for a certain period, this is done from event day 1 up until event day 252. For the calculation of WR the holding period return must be calculated, this can be done using the following equation:
𝑅𝑅𝑇𝑇 = ∏252𝑡𝑡=1 𝑅𝑅𝑖𝑖𝑡𝑡 (9)
Where 𝑅𝑅𝑖𝑖𝑡𝑡 is the return of stock i on event day t. Where 𝑅𝑅𝑇𝑇 is simply the product of the returns, this is done for the stock and the indexes. This formula basically calculates
31 yearly returns for each stock. This formula is similar to equation 2, but different from equation 1, because this formula allows calculating WR for each event day, not just after a complete year, so that the WR can be given for each event day after the IPO. Equation 9 is used as input for the calculation of WR, however, the formula for WR is given in equation 10.
𝑊𝑊𝑅𝑅 = 1+𝑦𝑦𝐴𝐴𝑦𝑦𝑦𝑦𝑦𝑦𝐴𝐴𝑦𝑦 𝑦𝑦𝑦𝑦𝑡𝑡𝑟𝑟𝑦𝑦𝑛𝑛 𝑜𝑜𝑛𝑛 𝑖𝑖𝑛𝑛𝑑𝑑𝑦𝑦𝑖𝑖1+𝐴𝐴𝐴𝐴𝑦𝑦𝑦𝑦𝑦𝑦𝐴𝐴𝑦𝑦 𝑦𝑦𝑦𝑦𝑡𝑡𝑟𝑟𝑦𝑦𝑛𝑛 𝑜𝑜𝑛𝑛 𝐼𝐼𝑃𝑃𝐼𝐼𝑠𝑠 (10)
A WR value of more than one would mean that holding IPOs would result in a higher return than holding on to the indexes. These calculations are made for the sub-samples and the total sample.
According to Brav & Gompers (1997) a venture capitalist will have a positive impact on stock returns. Therefore a one-sided paired t-test will be conducted. This will be done by using equation 1 to calculate the average yearly returns of both sub-samples and then test to see if a significant difference exists. The result of this will give an answer to hypothesis 2.
To be able to answer hypothesis 3 a regression analysis must be performed. The dependent variable is the yearly return on the IPOs, the variables that were added to the regression were the market and size premium, in the regression they are respectively 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 and 𝑆𝑆𝑆𝑆𝑆𝑆. The three proxy variables for information asymmetry were offer size, underwriter reputation and number of underwriters, which are respectively 𝐿𝐿𝑅𝑅𝑆𝑆𝑇𝑇𝐿𝐿𝑅𝑅, 𝐿𝐿𝑅𝑅𝑅𝑅𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑅𝑅𝐿𝐿 and 𝐿𝐿𝑅𝑅𝐿𝐿𝑅𝑅𝑅𝑅𝑅𝑅𝐿𝐿, where the Ln stands for natural logarithm. A summary of the meaning of the measures can be found in table 1. The first multiple regression is given in equation 11;
32 𝑅𝑅𝑖𝑖 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽1∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 + 𝛽𝛽2∗ 𝑆𝑆𝑆𝑆𝑆𝑆 + 𝛽𝛽3∗ 𝐿𝐿𝑅𝑅𝑆𝑆𝑇𝑇𝐿𝐿𝑅𝑅 + 𝛽𝛽4∗ 𝐿𝐿𝑅𝑅𝑅𝑅𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑅𝑅𝐿𝐿 +
𝛽𝛽5∗ 𝐿𝐿𝑅𝑅𝐿𝐿𝑅𝑅𝑅𝑅𝑅𝑅𝐿𝐿 + 𝜇𝜇𝑖𝑖 (11)
This is a multiple regression equation based on yearly returns. The yearly returns are calculated using equation 1 for 𝑅𝑅𝑖𝑖 and equation 2 for 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 and 𝑆𝑆𝑆𝑆𝑆𝑆. To normalize the data all the proxy variables are in natural logarithms, especially for the size coefficient this is important. The ability of the analysis might be harmed if this was not done; this is because of the great differences in offer sizes (Bali, Engle & Murray, 2016). This regression will be performed with heteroscedastic standard errors, because the Cook-Weisberg test for heteroscedasticity showed a p-value of 0.000 and thus heteroscedastic standard errors were required. This can easily be done in Stata with the option robust.
Equation 12 shows the next regression that will be performed, which is a panel regression based on daily returns.
𝑅𝑅𝑖𝑖𝑡𝑡 = 𝛼𝛼𝑖𝑖𝑡𝑡 + 𝛽𝛽1∗ 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 + 𝛽𝛽2∗ 𝑆𝑆𝑆𝑆𝑆𝑆 + 𝛽𝛽3∗ 𝐿𝐿𝑅𝑅𝑆𝑆𝑇𝑇𝐿𝐿𝑅𝑅 + 𝛽𝛽4∗ 𝐿𝐿𝑅𝑅𝑅𝑅𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑅𝑅𝐿𝐿 + 𝛽𝛽5∗
𝐿𝐿𝑅𝑅𝐿𝐿𝑅𝑅𝑅𝑅𝑅𝑅𝐿𝐿 + 𝜇𝜇𝑖𝑖𝑡𝑡 (12)
This will be a panel regression based on daily returns. The return of the company is given by 𝑅𝑅𝑖𝑖𝑡𝑡, while 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 and 𝑆𝑆𝑆𝑆𝑆𝑆 are the daily numbers of the two Fama-French factors. The three proxy variables will be dummies set to 0 except on the first return. On the first day of each IPO the natural logarithm of the offer size will be added. The natural logarithm of the Ritterrank and the natural logarithm of the number of IPOs are respectively 𝐿𝐿𝑅𝑅𝑅𝑅𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑅𝑅𝐿𝐿 and 𝐿𝐿𝑅𝑅𝐿𝐿𝑅𝑅𝑅𝑅𝑅𝑅𝐿𝐿. They are also dummy variables taking on the natural logarithm of their values on the first day and zero otherwise. In this research a random effects panel regression will be conducted. One reason for this is that the Hausman test results showed that the random effects estimator was the appropriate
33 estimator in this case. In general the random effects estimator is also more efficient than the fixed effects estimator. Heteroscedastic standard errors were used since the test result pointed out that the variance of the error term does not equal zero, and thus heteroscedastic standard errors were necessary.
To answer hypothesis 4 a one-way MANOVA and multivariate regression will be performed in Stata. These methods show if and how a venture capitalist influences the proxy variables of information asymmetry. A simplified version of this multivariate regression can be seen in equation 13.
𝐿𝐿𝑅𝑅𝑆𝑆𝑇𝑇𝐿𝐿𝑅𝑅; 𝐿𝐿𝑅𝑅𝑅𝑅𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑇𝑇𝑅𝑅𝐿𝐿; 𝐿𝐿𝑅𝑅𝐿𝐿𝑅𝑅𝑅𝑅𝑅𝑅𝐿𝐿 = 𝛼𝛼𝑖𝑖+ 𝛽𝛽1∗ 𝑉𝑉𝐶𝐶 + 𝑅𝑅𝑖𝑖 (13)
The MANOVA detects if differences exist between the average values of LnSize, LnRitterRank and LnNumuw when a venture capitalist has invested or not. While the multivariate regression will give three coefficients for the venture capitalist variable, one for each dependent variable. Since the venture capitalist dummy can only take on the values zero and one the coefficient will indicate if the dependent variables will be higher or lower when a venture capitalist is involved.
34
4. Results
In this section the results will be discussed extensively, the results of the CAR and WR equations will be discussed first, so that hypothesis 1 can be answered.
In order to answer hypothesis 1a, a look at the complete sample was needed, the CAR was calculated first after which the WR was calculated. Table 6 shows results of the two equations for the full sample. In Appendix 1A the complete set of results can be found. Table 6 shows the result of event day one, five and ten, but also results after one month, three months, six months, nine months and a full year, which are respectively event day 21, 63, 126, 189 and 252.
Table 6. Several observations within the CAR and WR calculations for the full sample, * indicates significance at the 10% level, ** indicates significance at the 5% level (two-sided) Event day CARsp500 t-test CARnasdaq t-test WRsp500 WRNasdaq
1 0.292 0.030 0.338 0.035 0.852 1.057 5 -0.826 -0.185 -0.754 -0.168 0.851 0.866 10 -2.290 -0.520 -2.250 -0.511 0.765 0.768 21 -0.918 -0.208 -1.014 -0.231 0.976 0.970 63 0.708 0.224 -0.232 -0.073 1.030 1.010 126 -2.408 -0.737 -3.693 -1.145 0.983 0.970 189 -4.451 -1.257 -6.697* -1.883 0.983 0.968 252 -8.228** -2.075 -11.465*** -2.899 0.971 0.955
The table shows interesting results, the results of the CAR calculation of the S&P500 show negative significant results after a whole year of trading. This means that the sample of IPOs underperforms relative to the S&P500. The results are significant with a one-sided alpha of 2.5%. Compared to the Nasdaq the IPOs underperformed with a significance of 1%, this means that hypothesis 1A can be accepted and the full sample of IPOs underperform the indexes. The t-statistic is found by calculating whether CAR is statistically different from zero. This is done by calculating the standard deviations of
35 the abnormal returns per benchmark for each event day and then using the standard deviations for the standard t-test calculation. Table 6 shows results of WR which are below one, this indicates that holding onto the indexes for a year would result in more money than investing in the IPOs. This is in line with the expectations (Ritter, 1991). In order to illustrate how the IPOs performed compared to the benchmark, graph 1 is created.
Graph 1. IPO performance relative to the benchmark of the complete sample
Graph 1 illustrates that IPOs almost perform similar compared to the benchmarks in the first 100 days after the IPO. After which the IPOs performance relative to the benchmarks decreases and underperforms to around ten percent after a full year. Hypothesis 1a can thus be accepted.
In order to answer hypothesis 1b, additional CAR and WR calculations were
-1 5 -1 0 -5 0 C AR 0 50 100 150 200 250
Days after IPO
carsp500 carnasdaq
36 needed. The complete results of the sub-sample are given in Appendix 1B and some results are given table 7.
Table 7. CAR and WR results of the venture capitalist sub-sample, * indicates significance at the 10% level, ** indicates significance at the 5% level and *** indicates significance at the 1% level (two-sided).
Event day CARsp500 t-test CARnasdaq t-test WRsp500 WRNasdaq 1 -0.491 -0.053 -0.447 -0.048 0.686 0.704 5 -0.740 -0.148 -0.573 -0.115 1.118 1.143 10 -1.836 -0.356 -1.675 -0.326 0.977 0.987 21 -0.284 -0.057 -0.260 -0.053 1.028 1.029 63 -0.236 -0.072 -0.988 -0.301 1.008 1.000 126 -7.697** -2.120 -8.848** -2.475 0.967 0.961 189 -12.725*** -3.208 -14.654*** -3.712 0.964 0.958 252 -16.302*** -5.015 -19.046*** -5.918 0.970 0.963
The results of the venture capitalist sub-sample show negative CAR values, this means that IPOs that were backed by venture capitalists perform worse than the benchmarks. This was not in line with the expectations (Brav & Gompers, 1997).
The WR also shows that investing and holding on to the sub-sample of IPOs would result in less value than holding onto the benchmarks. Although the WR is just below one this still indicates underperformance relative to the benchmarks. This was not in line with the expectations, and hypothesis 1b can be rejected.
To illustrate the performance of the venture capital backed IPOs a graph was created. Graph 2 shows exactly how much the IPOs underperform relative to the benchmarks. Almost immediately after going public the IPOs start underperforming compared to the benchmarks, this continues during the whole period of the study.
37 Graph 2. IPO performance relative to the benchmark of the complete sample
From these results, a conclusion can be drawn in which venture capital backed IPOs perform worse than the indexes and thus hypothesis 1b can be rejected.
In order to draw a conclusion on hypothesis 1c, the last CAR and WR calculations were conducted. The sub-sample with non-venture capital backed IPOs was investigated. Appendix 1C shows the complete results of this sub-sample, while table 8 shows a small part of the results.
-2 0 -1 5 -1 0 -5 0 C AR 0 50 100 150 200 250
Days after IPO
carsp500 carnasdaq
38 Table 8. CAR and WR results of the non-venture capitalist sub-sample, * indicates significance at the 10% level, ** indicates significance at the 5% level and *** indicates significance at the 1% level (two-sided).
Event day CARsp500 t-test CARnasdaq t-test WRsp500 WRNasdaq 1 1.075 0.107 1.123 0.112 1.591 1.641 5 -0.912 -0.237 -0.935 -0.240 0.919 0.916 10 -2.743 -0.785 -2.826 -0.800 0.859 0.855 21 -1.552 -0.409 -1.768 -0.467 0.962 0.955 63 1.652 0.540 0.523 0.169 1.012 1.000 126 2.881 1.002 1.462 0.513 1.018 1.011 189 3.823 1.247 1.259 0.403 1.020 1.011 252 -0.154 -0.034 -3.884 -0.847 1.000 0.990
The calculations ended up showing different results than what was expected. According to Ritter (1991) IPOs underperform indexes, but this result does not suggest that. In fact, there is not a single result that is significant this shows that IPOs, which were not backed up by venture capital do not underperform the benchmarks. Brav & Gompers (1997) specifically tested if non-venture capital backed IPOs underperformed the benchmarks, which they did, but this result shows that they have similar performance. Alongside this finding, the WR shows almost exactly the same performance as the indexes during the whole period. This confirms the CAR results and indicates that IPOs that are not backed up by venture capital firms, perform similarly to the benchmarks. To illustrate this similar performance graph 3 was created.
39 Graph 3. Performance of non-venture capital backed IPOs compared to the indexes
The above findings are illustrated by the graph and show that the non-venture capital backed sub-sample has similar performance as the two benchmarks. Thus hypothesis 1c can be rejected. Companies, which are not supported by venture capital, are not underperforming compared to the benchmarks.
The results given above state how well the IPOs performed relative to benchmarks, how well the two sub-samples performed compared to each other were not given, although the difference between the former outcomes somewhat suggests an outcome that was not expected. Another test to proof different outcomes was needed. Table 9 shows statistics on the two sub-samples.
Table 9. Returns and standard deviations of the two sub-samples Sub-sample N Mean return St. Dev return Mean
offer size
St. Dev Offer size VC-backed 123 -7.143534 61.07826 291.4009 1471.116 Non-backed 123 7.481294 67.07906 73.18125 167.7781 -4 -2 0 2 4 C AR 0 50 100 150 200 250
Days after IPO
carsp500 carnasdaq CAR Non-VC backed sub-sample
40 At first, the average yearly return for each IPO had to be calculated this was done using equation 1. After which a one-sided test was done to see if the venture capital backed IPOs had a higher return, this was tested by doing a paired t-test in Stata. The test result showed that the venture capital backed mean is lower in 94.95% of the cases. This suggests that venture capital backed IPOs perform worse than their non-venture capital backed counterpart, this is not in line with existing theory and an interesting finding (Brav & Gompers, 1997). With an alpha of 10% hypothesis 2 can be rejected. Before testing hypothesis 3 another test was conducted to measure difference in offer size between the two sub-samples. Another paired t-test was done, this time a two-sided test, the test result showed that the offer size is not equal to each other in 89.41% percent of the cases, in other words, the p-value was 0.1059. Although the result is not significant in a two-tailed test, economically it does suggest that the offer size of the venture capital backed sub-sample is on average higher.
In order to test hypothesis 3, multiple regression analysis had to be performed. The first results will be the results of equation 10, the yearly multiple regression. The result of this regression for the complete sample can be found in table 10.
Table 10. Yearly regression for the complete sample. * indicates significance at the 10% level, ** indicates significance at the 5% level and *** indicates significance at the 1% level (two-sided).
Variable Coefficient St.Error T-value P-value Market premium 1.127** 0.452 2.49 0.013 Size premium 3.276*** 1.079 3.04 0.003 Natural logarithm initial offer size 3.867* 2.282 1.69 0.091 Natural logarithm of Ritter rank -10.430 6.352 -1.64 0.102 Natural logarithm number of underwriters 3.684 7.774 0.47 0.636
Intercept -9.041 9.344 -0.97 0.334
This regression resulted in three coefficients were significance was found; these were the market premium, size premium and the natural logarithm of the initial offer size. If
41 the market return increases by 1 percent the return on IPOs will increase with 1.127 percent, if the spread in the size premium increases by 1 percent the returns of the sample of IPOs will increase by 3.867 percent. In fact, a high value of the SMB coefficient means that this sample most likely contains relatively ‘small market capitalization’ stocks. The proxy variable initial offer size is positive and significant at the 10% level; this indicates that a higher size of the initial offer increases one-year returns. The influence however, is low, considering that it is a natural logarithm. According to basic statistical theory this result indicates that if the initial size of the offer increases by 1 percent, the first year return is supposed to go up by 0.03867 percent. The other variables are insignificant, which indicate that they are not significantly different from zero and thus do not influence the returns. The natural logarithm of the Ritter rank shows a coefficient value of -10.430 with a p-value of 0.102; although the value is not significant it does indicate that it influences the return of the stock. A positive value was expected for the natural logarithm of the Ritter rank but the value turned out to be negative, this was not in line with the expectations. Since a higher underwriter reputation was expected to have a positive return on stocks. The number of underwriters appears to have no influence on the return of the stock, considering that the p-value is 0.636. In order to see if coefficients were different between the sub-samples two additional regressions were performed. These results can be found in table 11 and 12.
42 Table 11. Yearly regression for the non-venture capital backed sub-sample. * indicates significance at the
10% level, ** indicates significance at the 5% level and *** indicates significance at the 1% level (two-sided).
Variable Coefficient St.Error T-value P-value
Market premium 1.141* 0.664 1.72 0.088
Size premium 2.964* 1.533 1.93 0.056
Natural logarithm initial offer size 7.038** 3.032 2.32 0.022 Natural logarithm of Ritter rank -17.839* 9.418 -1.89 0.061 Natural logarithm number of underwriters 14.879 12.744 1.17 0.245
Intercept -21.549 15.616 -1.38 0.170
This result shows that the returns of non-venture capital backed IPOs are positively influenced by the market premium, the size premium and the natural logarithm of the offer size, and negatively by the rank of the underwriter. The offer size is significant at the 5% level, while the other three variables are all significant at the 10%. A negative coefficient for the rank of the underwriter was not expected. Table 12 shows the results of the yearly regression results for the other sub-sample.
Table 12. Yearly regression for the venture capital backed sub-sample. * indicates significance at the 10% level, ** indicates significance at the 5% level and *** indicates significance at the 1% level (two-sided).
Variable Coefficient St.Error T-value P-value Market premium 1.172** 0.589 1.99 0.049
Size premium 1.951 1.625 1.20 0.232
Natural logarithm initial offer size -1.537 5.136 -0.33 0.765 Natural logarithm of Ritter rank 2.629 7.016 0.37 0.709 Natural logarithm number of underwriters -6.856 8.030 -0.85 0.395
Intercept -0.941 18.895 -0.05 0.960
The table shows that for this sub-sample only the market premium coefficient is significant. This indicates that venture capital backed IPOs are not influenced by the size premium. All the other variables are also not significant and thus equal to zero. Their p-values are relatively high, which indicates that they are not close to being different from zero.
43 hypothesis 3, additional panel regressions on daily returns were performed. The results of the panel regression for the complete sample can be found in table 13.
Table 13. Panel regression for the complete sample. * indicates significance at the 10% level, ** indicates significance at the 5% level and *** indicates significance at the 1% level (two-sided).
Variable Coefficient St.Error Z-value P-value Market premium 0.804*** 0.051 15.72 0.000 Size premium 1.106*** 0.090 12.24 0.000 Natural logarithm initial offer size 0.299 0.265 1.13 0.259 Natural logarithm of Ritter rank -0.385 0.725 -0.53 0.595 Natural logarithm number of underwriters -0.640 0.792 -0.81 0.419
Intercept -0.005 0.015 -0.33 0.744
The results show that none of the dummy variables are statistically significant from zero. This means that they have no influence on the daily stock returns of the IPOs. Both Fama-French factors are highly significant, they play an important role in the direction of the stock. This is in line with expectations, because market movements are an important factor in stock returns. The result is different from the result obtained from the yearly regression, where one information asymmetry proxy was significant. The same regressions were done for the sub-samples, table 14 and 15 show results from the non-venture capital backed sample and venture capital backed sample respectively.
Table 14. Panel regression for the non-venture capital backed sub-sample. * indicates significance at the 10% level, ** indicates significance at the 5% level and *** indicates significance at the 1% level
(two-sided).
Variable Coefficient St.Error Z-value P-value Market premium 0.522*** 0.051 10.21 0.000
Size premium 0.745*** 0.115 6.46 0.000
Natural logarithm initial offer size 0.790** 0.359 2.20 0.028 Natural logarithm of Ritter rank -1.566* 0.951 -1.65 0.100 Natural logarithm number of underwriters -1.217 0.986 -1.23 0.217
