1 Influence of patents on IPO under-pricing
In this research, I examine the influence of the number of patents on the probability of IPO under-pricing in the United States of America from 2002 to 2008. I look at the number of granted patents a company owns before the IPO date, as well as the size of the company, the age of the company and the offer price-range. Consequently, I test the research based on the research question: do patents have an influence on the probability of IPO under-pricing in the United States of America from 2002 to 2008? By using the statistical program, STATA, I observe and conclude that the number of patents granted before the IPO date and the size of the company do not have influence on the IPO under-pricing. The age of the company and the offer price- range do have an influence on the IPO under-pricing.
Name: Daan Cuperus
ECTS: 138
Study direction: Finance and Organization University of Amsterdam
2 Statement of Originality
This document is written by Student Daan Cuperus who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
I. Introduction
An initial public offering (IPO) is the process of a company selling its shares to the public on a securities exchange. There are two reasons for a firm to perform an IPO, the first reason is to obtain new funds, the second reason is to refinance the firm (Rock, 1986). Under-pricing occurs when the offer price is lower than the price of the first trade. The reason for under-pricing is signalling expected prospects of growth to the investors (Adams, Thornton, & Hall, 2008).
In this study, I examine the effect of the number of granted patents before the IPO date on the probability of IPO under-pricing in the high-technology sector in the United States from 2002 to 2008. I focus on the number of patents because patents provide underlying information about the owner of the patents and its inventions (Lanjouw, Pakes, & Putnam, 1998). In this way patents reduce information asymmetry between the owner of the patents and the investors (Lanjouw, Pakes, & Putnam, 1998). Combining the theory of (Lanjouw, Pakes, & Putnam, 1998) with the theory of (Adams, Thornton, & Hall, 2008) namely that the reason for under-pricing is signalling expected prospects of growth to the investors. I test whether a company with patents before the IPO date, thus providing more underlying information than a company without patents before the IPO date, have a lower probability to perform an IPO under-pricing.
As I look at the influence of the number of granted patents on IPO under-pricing, I use the high-technology sector for it is the sector where the focus is the highest on research and development, therefore a higher focus on acquiring patents (Lippoldt & Stryszowski, 2009). The second reason to select the high-technology sector is that there is a correlation found between firm success in the technology sector and the availability of patents in the high-technology sector (Merges, 2006). And the third reason, high-high-technology-related patents have been found to be useful for new entrants to the high-technology sector. Such firms may use patents to signal their expertise to third parties, negotiate cross-licensing arrangements, increase
4
their value to potential buyers and convert tacit knowledge into a verifiable and transferable form (Mann, 2004).
I specifically orient myself on the United States of America, because the United States has the highest focus on research and development compared to any other country in the world (Lippoldt & Stryszowski, 2009). Also, when I look at the number of patents granted before the IPO date, I focus on the country with the highest high-technology patents growth rate, the United States has more high-technology patents issued than the combined number of the rest of the world (Lippoldt & Stryszowski, 2009). The patenting in the United States grows with 19% compared to the 16.1% in Japan and 18% in the other G7 countries (Lippoldt & Stryszowski, 2009).
I decide to use the time period of 2002-2008, because I exclude the Dot-com bubble (1992-2002). The Dot-com bubble was a time of great returns in the high-technology sector, thus companies rushed their IPO (Goodnight Thomas & Green, 2010). The companies wanted to enjoy the great returns and so the IPO’s were not performed rationally (Goodnight Thomas
& Green, 2010). I examine if patents have an influence on IPO under-pricing, therefore I do not want the Dot-Com bubble to affect my results, thus to create an omitted variable bias. For the same reason my time period ends at 2008, the financial crisis might affect my results as well by creating an omitted variable bias and preventing from forming a clear conclusion.
In this research, I conclude that the number of granted patents before the IPO date have no influence on the probability of IPO under-pricing in the high-technology sector in the United States from 2002 to 2008. So if a company has patents before the IPO date or not, this does not influence the IPO under-pricing. The size of the company also has no influence on the IPO pricing, this means that if a company is big or small, it does not influence the IPO under-pricing. However, the age of the company does have an influence on the IPO under-pricing, I found that older companies are less under-priced than young companies. Also the position the
5
company takes in the offer price-range has an influence on the IPO under-pricing, I see that companies who take in a high position in the offer price-range, are more under-priced than companies who do not take the high position in the offer price-range.
The main difference of my research in comparison to prior literature is the main independent variable. I use the main independent variable ‘patents’. Many prior literature use R&D expenditures. Patents are innovation output where R&D expenditures are innovation input (Lanjouw, Pakes, & Putnam, 1998). I look at the innovation output in this research as innovation output are successful innovations, where innovation input does not have to be successful, giving no assurance it provides underlying information about the firm (Lanjouw, Pakes, & Putnam, 1998).
The second reason why I look at innovation output is that R&D expenditures (innovation input) have risk of not being recorded consistently across companies, providing a bias estimation of the results (Lanjouw, Pakes, & Putnam, 1998). As patents provide underlying information about the owner of the patent, in this case the high-technology companies and it withholds the competition from duplicating the same invention (Lanjouw, Pakes, & Putnam, 1998). Therefore, a company with patents gives investors more underlying information than a company without patents. If investors have more information about the company, the information asymmetry is less than when a company does not have any patents providing no option for investors to receive underlying information. A high information asymmetry (no patents) between a company and its investors, gives the company an incentive to perform an IPO under-pricing (Adams, Thornton, & Hall, 2008). So, the contribution of my research to prior literature is looking at innovation output and the effect it has on IPO under-pricing.
The thesis consists of 6 sections. In section 1, I motivate why I use the number of granted patents before the IPO date as main focus, what sample selection I use for my research and contribution to prior literature. In section 2, I provide a literature background on an IPO, IPO
6
under-pricing and patents. In section 3, I explain the method I use for testing my research, I define the variables I use and provide information about the sample I use for the research. In section 4, first I show the correlation coefficients between the variables and second, I provide the results of my regression. In section 5, I perform a sensitivity analysis based on prior literature. Finally, in section 6, I give the conclusion of my research.
7
II. Literature review Initial public offering
An initial public offering (IPO) is the first time a firm enters the public capital market (Chemmanur, He, & Nandy, 2010). This is when a company offers shares on a security exchange for the first time and therefore accepts public ownership. The firm in the United States of America must register most public offerings of securities (including the initial public offering) with the SEC and receive SEC approval before selling the securities to the public (Cheng-Few & Alice C, 2006). There are two main reasons why a firm goes public. The first reason is to obtain new funds, by offering the shares of the company on a security exchange for a certain price, the company receives funds by those shares being bought by the public (Rock, 1986). The second reason for an IPO is to refinance the firm. Refinancing the firm through an IPO is the process by which a firm sells equity and therefore receives funds to, for example pay off debt (Rock, 1986). Refinancing the firm through an IPO gives the firm funds to pay the debt holders, so refinancing the firm through an IPO is beneficial for debt holders of the firm as they receive the owed money by the firm. Refinancing the firm through an IPO is also beneficial for shareholders. As the firm receives more funds, providing opportunities to pay off debt or expand the firm through increase in funds, the default risk of the company decreases by paying off the debt or shareholders value increases by the expansion of the firm by the funds (Rock, 1986).
The process of bringing an IPO requires a mechanism that would determine how many shares the IPO will offer, what price the investors would pay and who the initial investors will be. There are three ways to bring an IPO on the market. These three IPO mechanism are fixed price public offers, book building and auctions. In fixed price public offers, the price and allocation rules are set before information on demand is received, and shares are allocated according to the rules announced earlier (Jagannathan, Jirnyi, & Sherman Guenther, 2015). The advantage of a fixed price public offer is that the method is flexible of allocating the shares
8
among groups. The disadvantage of a fixed price public offer is that both the underwriter and issuer have no saying in shares allocation (Jagannathan, Jirnyi, & Sherman Guenther, 2015).
The second IPO mechanism is the book building. Book building, is the underwriter telling the issuer to attend a road show. Then collect interest from potential investors. The offer price is set only after the order book is full, giving the underwriter some idea of demand, and the issue size may also be adjusted based on demand (Jagannathan, Jirnyi, & Sherman Guenther, 2015). The advantage of book building is that the underwriter does have a say for the issuer about allocations or pricing.
The third IPO mechanism is the auction. Auctions as a IPO mechanism have many forms. The first auction is the uniform price auction where all winning bidders pay the same price. The price paid may be the market-clearing price (the highest price that allows all shares to be sold), or it may be below the clearing price, leading to increased rationing (Jagannathan, Jirnyi, & Sherman Guenther, 2015). The second auction is the pay for what you bid auction where the price is based on the investors bid. In this type of auction, the auction’s allocation is determined by a set of rules before the auction. So, leaving the underwriter out of the allocation process (Jagannathan, Jirnyi, & Sherman Guenther, 2015).
9
IPO under-pricing
IPO under-pricing is the phenomenon of an IPO, offering the company’s shares below market value and closing their first day with a share price above or at market value (Adams, Thornton, & Hall, 2008). The valuation of an IPO under-pricing can be done, first by calculating the ‘money left on the table. The issuing company (or major stockholders) will lose money if an
IPO is significantly under-priced. The loss (money left on the table) is determined as the number of shares in the offering multiplied by the difference between the first day opening price and the offer price (Adams, Thornton, & Hall, 2011). The other way of valuating an IPO under-pricing is simply calculating the initial return, this is the percentage return on the first day trading relative to the offer price (Adams, Thornton, & Hall, 2011).
IPO under-pricing has several benefits. The first benefit is that an IPO under-pricing can make the issue of the equity easier by setting a low offer price (Adams, Thornton, & Hall, 2008). The low offer price makes it less risky for the underwriter to have to pay for unsold shares. The second benefit of an IPO under-pricing is that it generates excess demand. The excess demand makes sure that the shares get widely dispersed across investors, making it impossible for one person to own the majority of the shares and therefore the majority of the company (Adams, Thornton, & Hall, 2008). The third benefit of an IPO under-pricing is to reduce information asymmetry between the company and its (potential) investors (Adams, Thornton, & Hall, 2011). An IPO under-pricing creates the costly situation of leaving money on the table for a firm. When a firm performs an IPO under-pricing, regardless of the money left on the table, it signals to the investors that it can bear the costs of the money left on the table thus expecting prospects of growth in the future (Adams, Thornton, & Hall, 2008).
There are two types of costs when performing an IPO under-pricing. The first cost is ‘leaving money on the table’, when shareholders prior to the IPO set the share value below
10
market value it creates the situation of leaving money on the table, missing out on extra funds by not setting the share value at market value (Adams, Thornton, & Hall, 2011). The second cost of an IPO under-pricing is the dilution of ownership. The shareholders prior to the IPO sell their shares on the security exchange, therefore selling the rights the shareholders prior to the IPO had over the company to the potential investors of the company (Adams, Thornton, & Hall, 2008).
Patents
The concept of patents has been there for already a long time. It is mostly used in the law by indicating that a product can be used only by the inventor who filed for the patent (Heshmat, 2014). The term ‘patent’ stands for a privilege granted by the Federal Government to an individual (Hamilton & Till, 1948). So, a patent is for an invention, as the invention must be separated from all that has been found out already. The definition of the invention has to be non-existent yet. Every country has its own patent-rules (Heshmat, 2014). For some countries only a definition that is non-existent is enough but for other countries a whole trial is needed to grant a patent (Heshmat, 2014).
By being able to patent your inventions, patents have several advantages. The first advantage of a patent is the incentive for research and development (R&D). Being able to file for patents, inventors will do more research to make an invention their own (Heshmat, 2014). The second advantage is that patents eliminate the idea of free-riding on each other ideas (Heshmat, 2014). If a company has high R&D expenditures, it can therefore have a lot of inventions or innovations which are patented and so belong to that company. There is no possibility for a company with low R&D expenditures to copy these ideas, inventions or innovations from companies with high R&D expenditures and so as well use the innovations for their production (Heshmat, 2014). The third advantage for patents is that when an innovation
11
gets patented, this patent will hold for, on average 20 years. After this patent-period, the innovation will be for the public and can be used by, for example starting companies to make improvements in the innovation. If there were no patents at all, this idea would be lost forever or would be kept secret by the inventors (Heshmat, 2014).
Hypothesis
Once again, my research question is: Do patents have an influence on the probability of IPO under-pricing in the high-technology sector in the United States from 2002-2008? Looking at the reasons for an IPO under-pricing and the literature review on ‘patents’ I can formulate a clear hypothesis for my research question.
My hypothesis is the following: patents do have an influence on the probability of IPO under-pricing in the high-technology sector in the United States of America from 2002-2008. As patents provide underlying information about the owner of the patents, the firm (Lanjouw, Pakes, & Putnam, 1998). A firm with many patents has more information available to (potential) investors than a firm with no patents. Therefore, there is less underlying information available to the (potential) investors. From this I formulate the hypothesis that firms with many patents, therefore providing more underlying information to (potential) investors, have less reason to reduce information asymmetry. As information asymmetry gets reduced by performing an IPO under-pricing (Adams, Thornton, & Hall, 2008). I expect that the number of granted patents before the IPO date does have an influence on the probability of under-pricing.
H0: β1=0 Patents do not have an influence on the probability of IPO- under-pricing
12
III. Method
To test my hypothesis, I run the following model:
Under-pricing = β0 + β1 * Patents + β2*LnMC + β3 * Age + d4 * Ceiling
In this model:
Under-pricing is the dependent variable and it is defined as ((First day closing price – Offer price)/Offer price) (Elston Ann & Yang Jimmy, 2010).
The number of granted patents before the IPO date is defined as the main independent variable ‘patents’. This variable is to see if it influences the probability of IPO under-pricing.
The size of the company is defined as the control variable ‘ Ln market capitalization at offer date’, this control variable is the natural logarithm of the offer price * shares offered at the IPO
date (Elston Ann & Yang Jimmy, 2010). This control variable is a natural logarithm to control for the skew in the data and to interpret the result better by comparing the results with the dependent variable in percentages than in absolute numbers.
The age of the company is defined as the control variable ‘age of the company’, it is the
difference between the year the company was founded and the year it is today ( 2017) (Jiang, Cai X, Keasey, Wright, & Zhang, 2014).
The position in the offer price-range is defined as the control variable ‘ceiling’. This variable is a dummy variable therefore can only accept the value 0 or 1. If the dummy variable is equal to 1, the offer price is set at the maximum position of the offer price-range. If the dummy variable equals 0, the position in the offer price-range is not at its maximum (Elston Ann & Yang Jimmy, 2010).
I expect a negative coefficient on the main independent variable ‘patents’ because firms with patents, provide more underlying information to its (potential) investors. As IPO under-pricing
13
is a way to reduce information asymmetry, I expect that when the number of granted patents before the IPO date increases, the probability of IPO under-pricing decreases.
I expect a negative coefficient on the control variable ‘market capitalization at the offer date’ because if the size of the firm is big, this means that the firm owns more assets, revenues and higher revenues than a relatively smaller company (Cheng-Few & Alice C, 2006). Companies with a bigger size have therefore more market power, making the bigger companies more familiar for investors in that specific market (Cheng-Few & Alice C, 2006). I expect that when the size of the company increases, the probability of IPO under-pricing decreases.
I expect a negative coefficient on the control variable ‘age of the company’, as older companies are more likely to be more experienced in the market, investors will therefore choose less risk instead of the upcoming, uncertain younger companies (Mcmillan Steven & Thomas, 2005). There is more information asymmetry between younger companies and investors than older companies and investors. I expect that when the age of the company increases, the probability of IPO under-pricing decreases.
I expect a negative coefficient on the control dummy variable ‘ceiling’ as when firms set their
offer price at the maximum position of the offer price-range, the chance for under-pricing is less than when a company (Elston Ann & Yang Jimmy, 2010). I expect that when the dummy variable ‘ceiling’ increases, the probability for IPO under-pricing decreases.
Sample
In my research I examine the influence of the number of granted patents on the probability of IPO under-pricing in the United States from 2002 to 2008. Therefore, my sample has to match the criteria I examine in my research. First of all, I use the high-technology sector because I
14
look at the number of patents, I look at the innovation output of a sector. I chose the high-technology sector as it has the highest focus on research and development, therefore the highest growth in acquiring patents (Lippoldt & Stryszowski, 2009). Another reason for the use of the high-technology sector is that small, new high-technology firms can use patents to signal to a third party about their expertise and knowledge and in this way increase their value (Mann, 2004), in this way provide underlying information about the company, this is an assumption in my research. I use the United States of America since the highest focus on research and development in the high-technology sector relatively to the other sectors in the country, is highest in the United States of America. The second reason for focussing on the United States is that this country has the highest patent-growth in the world (Lippoldt & Stryszowski, 2009). I limit the research period from 2002 to 2008, in order to exclude the Dot.com bubble since the IPO’s performed during the Dot.com bubble were not done rationally but based on the great
returns there could be profited from during the Dot.com bubble (Goodnight Thomas & Green, 2010). As I exclude the macro-economic shocks of the Dot.com bubble and the Financial Crisis, I get results based on the factors I am testing. There will not be an omitted variable bias.
My sample, which includes all IPO’s performed by high-technology companies in the United States from 2002 to 2008, is derived from the statistical database, Thomson One. By filtering the IPO’s based on sector, country and time period I obtain a clear list of companies. From this
database I observe my control variables, ‘market capitalization at offer date and my dummy variable ‘Ceiling’. The control variable ‘Age of the company’ comes from checking the annual
reports of all the companies for the year the company was founded. The main independent variable, ‘Patents’, comes from the law database Justitia. In this database I am able to find out
15
my final sample containing companies in the high-technology sector in the United States from 2002 to 2008, performing an IPO under-pricing.
Variable Obs Mean Std.Dev Min Max
Patents
LnMarketCapitalization Age of the company Dceiling 116 116 116 116 5.37069 5.891184 23.64655 0.5948276 11.68837 0.7927965 13.1891 0.4930552 0 3.12676 12 0 72 7.922769 130 1
Table 1: Descriptive statistics final sample
In table 1, I present the descriptive statistics of the final sample of my research. This table shows the number of observations, the mean, the standard deviation and the minimum and maximum of the variables I test in my research. In table 1, I observe that the minimum amount of granted patents before the IPO date a company owns is 0. I explain this number by the fact that as the high-technology sector in the United States has the highest growth rate compared to the world statistics, this gives an incentive for new high-technology companies to arise (Lippoldt & Stryszowski, 2009). And as a patent application takes a substantial amount of time for review and approval by the authorities (Lippoldt & Stryszowski, 2009), this leads to the fact that young start-up high technology companies have 0 granted patents before the IPO date as shown in the table 1. In table 1, the minimum and the maximum of the variable’ age of the company’ are relatively far apart. This difference can be explained by the fact that not all high-technology companies came from the growth period in the past decade, there were also companies founded before the growth period (Lippoldt & Stryszowski, 2009) explaining the big difference between the minimum and the maximum.
16
IV. Results
In table 2, I present the correlation table between the variables used in my research. These variables include the number of granted patents before the IPO date (Patents) , the size of the firm ( LnMarketcapitalization), the age of the company ( Age of the company) and the position of the offer price in the offer price-range (Dceiling).
Patents LnMarketcapitalization Age of the company Dceiling Patents
LnMarketcapitalization Age of the company Dceiling
1.000
0.0916 1.000 0.0018 -0.0905 1.000
0.0685 0.3148 -0.1372 1.000 Table 2: Correlation table of the variables
As you can see in table 2, the highest correlation is between the variable ‘LnMarketcapitalization’ and the dummy variable ‘Dceiling’. The correlation between the two
variables is 0.3148. This correlation can be explained by the fact that both the ‘LnMarketcapitalization’ and the ‘Dceiling’ are influenced by the offer price of the IPO. The ‘LnMarketcapitalization’ is defined as the natural logarithm of the offer price times the shares offered at the IPO and the dummy variable ‘Dceiling’ is defined as the position the company
sets their offer price in the offer price-range. Both variables are influenced by the offer price of the IPO, this explains the relatively high correlation.
The second highest correlation of my regression is between the variables ‘Age of the company’ and the dummy variable ‘Dceiling’. This correlation is a negative correlation of -0.1372. This
second highest correlation can be explained by the fact that investors prefer older companies above younger companies due to the risk of start-up failure (Mcmillan Steven & Thomas, 2005). In this way the younger companies have to set the offer price at the maximum of the offer price-range to avoid the risk of the underwriter having to buy the unsold shares (Mcmillan Steven & Thomas, 2005). The older companies who have more trust from the investors, run less risk of
17
unsold shares, therefore the older companies can set the offer price different from the offer price-range.
The third highest correlation of my regression is between the variables ‘lnMarketcapitalization’ and ‘Patents’. The correlation between the size of the firm and the number of granted patents
before the IPO date is 0.0916. This third highest correlation can be explained by the fact that patents can help small firms to signal their expertise to third parties (Lippoldt & Stryszowski, 2009). In this way, the number of granted patents is correlated with the size of the firm.
When looking at table 2, I examine the table for multicollinearity. Multicollinearity occurs when two or more variables in the same regression are linearly dependent or nearly dependent (Bahovec, 2014). If two variables are perfect positive linearly dependent, this means that the correlation between the two variables is 1. If two variables are near linearly positive dependent, the correlation between the two variables is close to 1. Also, a correlation can be negative, if a correlation is -1, this means perfect negative linearly dependent or if the correlation is near -1, this means near perfect negative linearly dependent (Bahovec, 2014). The consequence for multicollinearity is that the estimations in the regression become biased and inconsistent (Stock. H & Watson, 2015). The highest correlation coefficient in my regression is 0.3148, since 0.3148 is not 1 or not near 1, this means that there is no multicollinearity present in my regression.
18
In the second part of the results, I will show the results of my regression. Under-pricing Coefficient P> ǀ t ǀ Patents
LnMarketcapitalization
Age of the company
Dceiling Constant R2 Prob > F N -0.0007239 0.488 (0.0010396) 0.0366546 0.265 (0.0327379) -0.0031683 0.006 (0.0011295) 0.1154072 0.001 (0.0342189) 0.504187 0.805 (0.0342189) 0.1656 0.0001 116 Table 3: Robust regression output of my research
- Standard errors of the coefficients are given within the parentheses
In table 3, I present the outcome of the regression of my research. My regression consists of 116 observations. The R-squared of the regression is 0.1656, this means that 16.56 percent of the variance of the dependent variable is predictable from the independent variables (Stock. H & Watson, 2015).
The coefficient on the main independent variable ‘Patents’ is -0.0007 with a P-value of 0.488,
this variable is statistically insignificant at the significance level of 0.05, this means that I do not reject H0. This outcome suggests that the number of granted patents before the IPO date does not have an influence on the probability of IPO under-pricing in the high-technology sector in the United States from 2002 to 2008. The outcome of the regression for this variable does not support my hypothesis. The coefficient is negative, this is in line with my prediction.
The coefficient on the first control variable ‘Ln Marketcapitalization’ is 0.3665 with a P-value of 0.265, this variable is statistically insignificant at the significance level of 0.05, this means
19
that I do not reject H0. This outcome suggests that the size of the firm does not have an influence on the probability of IPO under-pricing in the high-technology sector in the United States from 2002 to 2008. The outcome of the regression for this variable does not support my predictions. The coefficient is not negative therefore also not in line with my predictions for the coefficient. In prior literature (Elston Ann & Yang Jimmy, 2010), the control variable for size of the firm is significant with significance level of 0.05, therefore the outcome of my regression is not in line with that of prior literature.
The coefficient on the second control variable ‘age of the company’ is -0.0032 with a P-value
of 0.006, this variable is statistically significant at the significance level of 0.05. This means that I do reject H0, this outcome suggests that the age of the company does have an influence on the probability of IPO under-pricing in the high-technology sector in the United States from 2002 to 2008. The outcome of the regression for this variable does support my predictions. The coefficient is negative, therefore also in line with my predictions for the coefficient. In prior literature (Jiang, Cai X, Keasey, Wright, & Zhang, 2014), the variable age of the company is not significant, therefore the regression is not in line with that of prior literature.
The coefficient on the third control dummy variable ‘Dceiling’ is 0.1154 with a P-value of
0.001, this variable is statistically significant at the significance level of 0.05. This means that I do reject H0, this outcome suggest that the position the company sets in the offer price-range does have an influence on the probability of IPO under-pricing in the high-technology sector in the United States from 2002 to 2008. The outcome of the regression for this variable does support my predictions. The coefficient is positive therefore not in line with my predictions for the coefficient. In prior literature (Elston Ann & Yang Jimmy, 2010), the variable is significant therefore my regression outcome is in line with that of prior literature.
20
V. Sensitivity Analysis
In section 5 of the thesis, I present the sensitivity analysis regarding my research. I use a sensitivity analysis as it measures the sensitivity of my estimations when a change of parameters occurs, this shows if the regression estimations are consistent (Caswell & Sanchez Gassen, 2015).
In this sensitivity analysis I change the parameters by adding a new control variable and removing observations. This new control variable is the P/E ratio, the Price / Earnings ratio of the companies in my sample. This control variable has been used in the prior literature (Huayang & Yan, 2011) to examine the IPO under-pricing influences in China. In the database Thomson One I find the data for the P/E ratios for the sample used in this research. Due to lack of information on the P/E ratio for some companies, the amount of observations has decreased from 116 to 89.
Under-pricing Coefficient P> ǀ t ǀ Patents
LnMarketcapitalization
Age of the company
Dceiling P/E ratio Constant R2 Prob > F N 0.0001122 0.938 (0.0014417) 0.0186361 0.639 (0.0395678) -0.0059803 0.005 (0.0020851) 0.1232785 0.005 (0.0428162) 0.0000116 0.006 (0.00000411) 0.2163437 0.405 (0.2585316) 0.1687 0.0000 89 Table 4: Sensitivity analysis
21
In table 3, the outcome of the regression of my research the variable ‘Patents’ was not
statistically significant with a significance level of 0.05. After performing the sensitivity analysis, I examine if the outcome has changed regarding the influence on IPO under-pricing. The variable ‘Patents’ still does not have an influence on the probability of IPO under-pricing
in the high technology sector in the United States from 2002 to 2008 since the insignificance has not changed.
In table 3, the outcome of the regression of my research the variable ‘LnMarketcapitalization’
was not statistically significant with a significance level of 0.05. After performing the sensitivity analysis, I examine if the outcome has changed regarding the influence on IPO under-pricing. The variable ‘LnMarketcapitalization’ still does not have an influence on IPO under-pricing in the high-technology sector in the United States from 2002 to 2008 since the insignificance has not changed. I still do not reject H0.
In table 3, the outcome of the regression of my research was that the variable ‘Age of the company’ was statistically significant with a significance level of 0.05. After performing the
sensitivity analysis, I examine if the outcome has changed regarding the influence on IPO under-pricing. The variable ‘Age of the company’ still does have an influence on the probability of IPO under-pricing in the high-technology sector in the United States from 2002 to 2008 since the significance has not changed. I still reject H0.
In table 3, the outcome of the regression of my research was that the variable ‘Dceiling’ was statistically significant with a significance level of 0.05. After performing the sensitivity analysis, I examine if the outcome has changed regarding the influence on IPO under-pricing. The dummy variable ‘Dceiling’ still does have an influence on the probability of IPO
under-22
pricing in the high-technology sector in the United States from 2002 to 2008 since the significance has not changed. I still reject H0.
I conclude that after adding the control variable P/E ratio and decreasing the amount of observations, the outcome of my regression does not change. Therefore, the estimations that I present in the regression in table 3 are consistent when adding a new control variable and decreasing the amount of observations.
23
VI. Conclusion
In this research I examined and analysed the influence of the number of patents on the probability of IPO under-pricing in the high-technology sector in the United States from 2002 to 2008. I tested this research using a sample containing high-technology companies in the United States performing an IPO from 2002 to 2008. The main independent variable is the number of patents granted before the IPO date, the control variables in this research were used in prior literature about IPO under-pricing. The control variables are the size of the firm, the age of the company and the offer price-range. To test this influence of the variables on the IPO under-pricing, I used the statistical program STATA performing a regression giving the significances of the variables on IPO under-pricing.
First of all, I compare the outcome for my main independent variable with the hypothesis of my thesis. The hypothesis in my thesis is as following: the number of granted patents before the IPO date does have an influence on the probability of IPO under-pricing in the high-technology sector in the United States of America from 2002 to 2008. For my main independent variable ‘patents’, the variable is not significant therefore we do not reject H0 and so the number
of patents granted before the IPO date has no influence on the IPO under-pricing. So the outcome of the regression does not support the hypothesis of my thesis.
Secondly, I conclude the main findings for the control variables. The control variable ‘market capitalization at offer date’ is not significant, we do not reject H0 and the variable has
no influence on the IPO under-pricing. The control variable ‘age of the company’ is significant, therefore we reject H0 and thus the age of the company is of influence on the IPO under-pricing. For the dummy variable ‘Ceiling’ the variable is significant, so the setting of the offer price at
the maximum of the offer price-range is of influence on IPO under-pricing.
My research focussed on the high-technology sector in the United States from 2002 to 2008. There would have been an interesting alternative option: a focus on the upcoming and
24
even faster growing high-technology sector in a country like India (Zorpette, 1994), many IPO’s in India could be more interesting to look at. As India takes over the United States as biggest high-technology country in the world, the IPO’s in India could be more lasting and of higher value (Zorpette, 1994). Future research should focus on the upcoming high-technology countries instead of the already established countries as they stand still and do not grow as much as they did before. Another limitation of my research is that I exclude events as the Dot.com bubble or the financial crisis. For high-technology companies the Dot.com bubble was a time of rapid growth and a recession after the bubble popped (Goodnight Thomas & Green, 2010). During the financial crisis, there was a time of uncertainty which decreased the amount of IPO’s. In future research, researchers should include either a crisis variable or a dummy variable
for the Dot.com bubble or the Financial Crisis to see if the variables get influenced by the crisis and therefore if the crisis affects the amount of IPO under-pricing.
25
VII. Appendix
Patents The term ‘patent’ stands for a privilege
granted by the Federal Government to an individual (Hamilton & Till, 1948)
Initial public offering (IPO) This is the first time a firm enters the public capital market (Chemmanur, He, & Nandy, 2010). It is when a company offers shares on a security exchange for the first time and therefore accepts public ownership.
Initial public offering under-pricing IPO under-pricing is the phenomenon of an IPO offering their shares below market value and closing their first day with a share price above or at market value (Adams, Thornton, & Hall, 2008).
The Dot.com bubble The Dot.com bubble was a bubble because of
the (unexpected) rapid growth of the high-technology sector, the share value of tech companies rose enormously and created a regression when the bubble popped. The Dot.com bubble lasted from 1997 to 2002 (Goodnight Thomas & Green, 2010).
Control variable A control variable is a variable used in prior
literature. Mostly held constant to help in giving the effect of the one variable on the other
26
Dummy variable A dummy variable is an interaction variable
that can get the values one or zero to see if the presence or absence of the variable influences the dependent variable (Stock. H & Watson, 2015).
R&D expenditures An expense related to the research and development of a company (Heshmat, 2014). Fixed price public offers The price and allocation rules are set before
information on demand is received, and shares are allocated according to the rules announced earlier (Jagannathan, Jirnyi, & Sherman Guenther, 2015).
Book building Book building is the underwriter telling the
issuer to attend a road show. Then collect interest from potential investors. The offer price is set only after the order book is full, giving the underwriter some idea of demand, and the issue size may also be adjusted based on demand (Jagannathan, Jirnyi, & Sherman Guenther, 2015).
Auctions The two auctions that can be used as an IPO
mechanism are uniform price auction and the pay-for-what-you-bid auction (Jagannathan, Jirnyi, & Sherman Guenther, 2015).
27 Uniform price auction Uniform price auctions are when all winning
bidders pay the same price. The price paid may be the market-clearing price (the highest price that allows all shares to be sold), or it may be below the clearing price, leading to increased rationing (Jagannathan, Jirnyi, & Sherman Guenther, 2015).
Money on the table Wealth losses by selling the shares too low in comparison to the actual market value of the company, the pre-IPO shareholders set a share value not reflecting the company’s
value (Adams, Thornton, & Hall, 2008). Information asymmetry If one side of the trade knows the actual
quality/size/ costs of the products and the other side does not know the quality/ size/ costs. Therefore, a situation where one side knows more than the other side, creates an information asymmetry (Lanjouw, Pakes, & Putnam, 1998).
Omitted variable bias An omitted variable bias occurs when a model created incorrectly leaves out one or more important factors. The "bias" is created when the model compensates for the missing factor by over- or underestimating the effect
28
of one of the other factors (Stock. H & Watson, 2015)
Multicollinearity Multicollinearity arises when two or more
variable of an equation are linearly dependent. If the correlation is 1, this means perfect multicollinearity. If the correlation coefficient is close to 1, this indicates near linearly dependency (Stock. H & Watson, 2015).
29
Bibliography
Adams, M., Thornton, B., & Hall, G. (2008). IPO Pricing Phenomena: Empirical Evidence Of Behavioral Biases. Journal of Business & Economics Research, 8-12.
Bahovec, V. (2014). Multicollinearity. International Encyclopedia of Statistical Science, 17-18.
Caswell, H., & Sanchez Gassen, N. (2015). The sensitivity analysis of population projections.
Demographic Research, 801-840.
Chemmanur, T., He, S., & Nandy, D. (2008). The Going Public Decision and the Product Market. Review
of Financial Studies, 28-34.
Cheng-Few, L., & Alice C, L. (2006). Encyclopedia of Finance. Encyclopedia of Finance, 257-258. Coackley, J., Hadass, L., & Wood, A. (2009). UK IPO underpricing and venture capitalists. The European
Journal of Finance, 421-435.
Ekkayokkaya, M., & Pengniti, T. (2012). Governance reform and IPO underpricing. Journal of Corporate
Finance, 238-253.
Elston Ann, J., & Yang Jimmy, J. (2010). Venture capital, ownership structure, accounting standards and IPO underpricing: Evidence from Germany. Journal of Economics and Business, 517-536. Elston Ann, J., & Yang Jimmy, J. (2010). Venture capital, ownership structure, accounting standards and
IPO underpricing: Evidence from Germany. Journal of Economics and Business, 517-536. Goodnight Thomas, G., & Green, S. (2010). Rhetoric, Risk, and Markets: The Dot-Com Bubble. Quarterly
Journal of Speech, 115-140.
Hamilton, W., & Till, I. (1948). What Is a Patent? Law and Contemporary Problems, 245-259. Heshmat, A. (2014). Patents. Journal of Elastomers & Plastics, 579-593.
30 Huayang, Y., & Yan, L. (2011). Empirical Study on IPO Underpricing of GEM Listed Companies in China.
International Conference on Management and Service Science, 1-4.
Jagannathan, R., Jirnyi, A., & Sherman Guenther, A. (2015). Share auctions of initial public offerings: Global evidence. Journal of Financial Intermediation, 283-311.
Jiang, P., Cai X, C., Keasey, K., Wright, M., & Zhang, Q. (2014). The role of venture capitalists in. Journal
of International Small Business, 619-643.
Koop, G., & Li, K. (2001). The valuation of IPO and SEO firms. Journal of Empirical Finance, 375-401. Lanjouw, J., Pakes, A., & Putnam, J. (1998). How to Count Patents and Value Intellectual Property: The
Uses of Patent Renewal and Application Data. Journal of Industrial Economics, 405-432. Lippoldt, D., & Stryszowski, P. (2009). Innovation in the Software Sector. OECD.
Mcmillan Steven, G., & Thomas, P. (2005). Financial success in biotechnology: company age versus company science. Technovation, 463-468.
Nielsson, U., & Wojcik, D. (2016). Proximity and IPO underpricing. Journal of Corporate Finance, 92-105.
Rushe, D. (2012, December 6). Technology sector found to be growing faster than rest of US economy.
The Guardian, pp. 17-21.
S, C. B., Jade S, D., & Ajay K, K. (2015). Marketing an IPO issuer in early stages of the IPO process. Journal
of the Academy of Marketing Science, 14-31.
Stock. H, J., & Watson, M. (2015). Introduction to econometrics. Harlow: Pearson.
Wang, X., & Wan. P, W. (2013). Explaining the Variance in Underpricing among Venture Capital-Backed IPOs: A Comparison Between Private and Corporate VC Firms. Strategic Entrepeneurship
31 Zorpette, G. (1994). Technology in India. IEEE Spectrum, 24-32.
