Bachelor Thesis 2017-2018 period 1
By Bas Koster (11026928)
Under-pricing of internet companies at IPO during the dot-com bubble
Supervisor: Andrej Woerner
Economics and Business
Statement of Originality
This document is written by Bas Koster 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.
Abstract
During the dot-com bubble in 1997-2000, many internet companies had their Initial Public Offering (IPO) with high percentage gains on the first trading day. These high percentage gains indicate that the companies were under-priced at their IPO. Speculations about the potential of the internet led to a real hype around internet companies. To see if internet companies which had their IPO during the dot-com bubble were really exceptionally under-priced, this research will dot-compare these internet companies with internet companies which had their IPO in a later period (2002-2016). A regression including control variables like asset size and sector was executed. The results show that internet companies which had their IPO during the dot-com bubble had on average 67% higher first day gains, compared to internet companies which had their IPO on a later date. Secondly, the percentage gain of internet companies over the period 1997-2016 is 33% higher, compared to non-internet companies. Finally, the total asset size of a company during the IPO year has no effect on the under-pricing of that company. These results can be used to predict IPO under-pricing of companies which operate in an economy with a similar hype going on, like the cryptocurrency market.
I. Introduction
During the years 1997 – 2000, the internet itself and internet companies were starting to rise to the massive internet network we know nowadays. This period of time is known as the dot-com bubble. Some of the biggest companies today are internet companies (like Facebook and Amazon), which shows how big internet companies have become in a relatively short period of time. In the early days of the internet, there was already a lot of speculation about the high potential of the internet and internet companies. The hype around internet companies during the dot-com bubble could have had an effect on the valuation of internet companies (and their stocks). This would be in contrast to the
efficient market hypothesis, according to which every form of information is already reflected in the price of a stock. Fama (1991) stated that market efficiency is hard to measure, but there is some direct evidence on efficiency based on mainly event studies. This market efficiency implicates that there would not be severe under-pricing at IPOs. The aim of this paper is to investigate the relationship between the hype around internet companies and the under-pricing of these companies at their IPO.
A comparison between internet companies and non-internet companies is made to investigate the effect of the hype on the under-pricing of internet companies. The findings on this effect of the hype on under-pricing can then be used for IPO under-pricing predictions of other markets with a similar hype going on (like the cryptocurrency market for example). A research which is quite comparable to this research has been done by Ljunqvist and Wilhelm (2003). They investigated the under-pricing at the IPO of companies during the dot-com bubble. However, due to the date of the research (2003), there did not exist much internet companies which had their IPO after the dot-com bubble. Therefore, they were not able to compare internet companies which had their IPO during the dot-com bubble with internet companies which had their IPO in a later period. This paper on the other hand will compare the dot-com internet companies to the internet companies which had their IPO in the period after the dot-com bubble (ranging from 2002-2016). This will make sure it is observable whether the under-pricing of internet companies in the dot-com bubble was really exceptional high. The central question is: ‘Did the hype about internet companies during the dot-com bubble lead to higher under-pricing of internet companies at their IPOs?’.
In this research, under-pricing at the IPO is defined as the percentage gain on the first day of trading. The data exists of a total of 105 internet and non-internet companies. The IPOs from two periods of time are investigated: IPOs during the dot-com bubble (1997-2000) and IPOs in the period after the dot-com bubble (2001-2016). During the dot-com bubble, there was a hype about, while there was no hype in the period after the dot-com bubble. Ofek and Richard (2001) came up with a theory that explains the beginning and end of the dot-com hype. They stated that the dot-com hype is caused by a large group of optimistic investors who were willing to pay a high price for the dot-com shares, this caused the share prices to rise exceptionally. At the end of 2000 did the lock-up periods of many of the dot-com companies expire. The group of pessimistic investors were able to sell their shares from that moment onwards. This mass selling brought the prices back down and the hype to decrease, eventually to the point when no hype existed at all. Non-internet companies on the other hand were not hyped in both periods, so these companies will be useful to examine the effect of a hype on IPO under-pricing. The non-internet companies will control for the possibility that under-pricing was in general higher during the dot-com bubble, compared to the period after the dot-com bubble.
The results show that internet companies which had their IPO during the dot-com bubble had an on average 67% higher first day gain, compared to internet companies which had their IPO in the
period after the dot-com bubble (controlling for asset size and sector). No significant difference in under-pricing is found between non-internet companies which had their IPO during the dot-com bubble and non-internet companies which had their IPO in the period after the dot-com bubble. This implies that there was no real difference in IPO under-pricing in general in the two periods of time. Comparing both types of companies (IPOs in the period 1997-2016) shows that internet companies have a 33% higher first day gain, compared to non-internet companies. No relationship between total asset value (during the IPO year of that company) and IPO under-pricing is found. All these findings are in accordance with the three hypothesis, which can be found in the third section. The following section reviews existing literature about under-pricing at IPOs. In the third section the three hypotheses are stated, which provides a guideline for the empirical approach in this paper. The fourth section consists of the regression results and the interpretation of these results. The last section concludes and discusses limitations of this paper.
II. Literature review
The under-pricing of a company at their IPO is researched a lot. A review of the existing literature about IPO under-pricing in general and of internet companies during the dot-com bubble is made. The hypothesis that are stated in the next section, are mainly based on the findings of these papers. Information asymmetry is an important condition for deliberate IPO under-pricing, according to Rock (1986) and Hoque (2014) who both discussed this relationship in their papers. Rock (1986) divided investors in two groups: one group of investors which have superior information and one group of investors that does not have this information available. He states that when the stock is priced at the expected value this superior group crowds out the other group. When the stock is priced below the expected value this group will withdraw from the offer, which will serve as a signalling device. The company can prevent this from happening by under-pricing the stock, in which both type of investors will be attracted. Hoque (2014) defined three groups of companies as companies with high information asymmetries: companies with small IPOs, IPOs underwritten by non-prestigious underwriters and IPOs which join the AIM (alternative investment market). His results show that there was more under-pricing at IPOs of companies that fits in at least one of the three groups named above, compared to other companies. The results of Rock (1986) and Hoque (2014) can be used to explain a possible higher IPO under-pricing of internet companies during the dot-com bubble. In some papers, it is stated that the IPO under-pricing can be seen as some kind of compensation for the investors, like the papers of Loughran & Ritter (2000) and Benveniste & Spindt (1989). The paper of Loughran and Ritter (2000) provides several reasons on why issuers are not upset
when they leave ‘money on the table’ by under-pricing their stock at the IPO. Their results show that the companies which miss out on the most money at their IPO typically already have a higher IPO price than the price that was first expected. Because of this higher price, these issuers already have more money raised through their IPO than they expected beforehand. This ‘extra gain’ makes up for the money which was left on the table, so the issuers get a winners feeling and therefore they will not get upset about the higher potential price they could have obtained. Benveniste and Spindt (1989) also stated that under-pricing arises as a compensation for the investors. These investors are compensated for the existence of information asymmetry: the investors knew obviously not everything about the company at the time of the IPO (through which their investment could be risky). Chambers & Dimson (2009) and Lowry & Schwert (2002) investigated IPOs over a longer period of time. Chambers and Dimson (2009) showed that under-pricing is a phenomenon which already occurred in 1917. The average under-pricing in that period was only 3.8%, which is really low compared to the levels of under-pricing of the internet companies included in the dataset for this research. The median of under-pricing in the period 1917-1945 was 0.42%, while the median in the period 1945 - 1986 was 8.79%. The under-pricing continued to increase in the period thereafter, with an equally weighted mean of 18.08% in the period 1990-1999 and 19.86% in the period 2000-2007. This implies that a possible difference which may be found in this research could be allocated to a general trend in time. A control group will be used in this paper to control for this possible general trend in time. Whereas Lowry and Schwert (2002) observed the pattern in which companies choose to go public and saw that companies of the same type typically go public in the same period of time. This can also be noticed in this research sample: companies which are active in the same sector often have their IPO in the same period as at least one of the other companies active in that sector.
In other papers, the relationship between IPO under-pricing and one other specific variable is investigated. Liu and Ritter (2011) analysed the relationship between IPO under-pricing and the coverage from an all-star analyst, which is defined as an influential analyst by the ‘Institutional Investor magazine’. In their research, IPOs from 1993 till 2008 were investigated and dummy variables for internet companies and companies which had their IPO during the dot-com bubble were included as control variables. They found that IPOs which are covered by an all-star analyst are 20% more under-priced. Above all, they noticed a positive relationship between first day returns and being an internet company or having an IPO during the dot-com bubble. This relationship will be further investigated in this paper. Furthermore, Boulton et al. (2011) investigated the effect of earnings quality (i.e. the quality of accounting information) in different countries on IPO under-pricing. They found that in countries with higher quality earnings information IPOs were less under-priced. In their research, Boulton et al. (2011) also used some control variables which will also be used in this research. For example, they looked at the offer size at the IPO and found no effect on the under-pricing at the IPO. In a later
research, Boulton, Smart and Zutter (2017) found out that there is a negative relationship between accounting conservatism and IPO under-pricing. This negative relationship was the strongest at IPOs of smaller companies, in which there is likely more information asymmetry. Aggarwal et al. (2002) looked at strategic under-pricing by managers. This type of under-pricing refers to the actions which managers take to let their stock have a very large first day return. Large first day returns attract more attention to the stock, which increases the demand of the stock. This increase in demand causes a higher stock price when the lock-up period expires. Therefore, will under-pricing at the IPO lead to higher stock prices on the medium term. Managers usually sell their owned stock after the lock-up period, instead of at the IPO. These managers thus benefit from a higher stock price after the lock-up period. Aggarwal et al. (2002) found a positive correlation between the amount of shares owned by the managers and the level of IPO under-pricing, which is consistent with the stated theory. From all the research that has been done on this topic, the research from Ljungqvist and Wilhelm (2003) is the most comparable to the research which will be executed in this paper. Ljungqvist and Wilhelm (2003) analysed the under-pricing during the dot-com bubble and saw that the median of the first day return was 40% in 1999 and 30 percent in 2000. They compared internet and technology companies to other companies which were not active in these two sectors. They found that internet and technology companies had an on average 5.6% and 14.5% higher first day return, compared to the other types of companies. The main drawback of this research is that the under-pricing of the internet companies with their IPO during the dot-com bubble is not compared to the under-pricing of internet companies which had their IPO in a later period. However, this was also not possible because of the date of their publication (2003). There was just not enough data for this comparison available. This is where the relevance of this paper is shown: internet companies which had their IPO during the dot-com bubble are much under-priced, but is this under-pricing also a lot dot-compared to the under-pricing of internet companies which had their IPO in the period after the dot-com bubble? Most of the internet giants of today had their IPO in the last ten years, for example Facebook in 2010, TripAdvisor in 2011 and Twitter in 2013. No research has been done which compares the internet companies with their IPO in the dot-com bubble to the internet companies which had their IPO in the last ten years. This comparison would show if the dot-com bubble from 1997 to 2000 was really a period of time with exceptional high gains of internet companies on their first trading day.
In summary, the existing literature shows some important findings which are useful for this research. For example, Liu and Ritter (2011) found a positive relationship between internet companies and IPO under-pricing. They saw the same positive relationship between companies which had their IPO during the dot-com bubble and IPO under-pricing. Ljungqvist and Wilhelm (2003) also showed the high first day gains of internet companies which had their IPO during the dot-com bubble. Based on these results, a higher first day gain for internet companies, compared to non-internet companies and
for com companies (IPO in 1997-2000), compared to companies which had their IPO after the dot-com bubble is expected. Above all, a difference in under-pricing per period of time was shown by Chambers and Dimson (2009). A control group existing of non-internet companies will be included to control for this difference in under-pricing per period of time.
III. Hypothesis
Three hypotheses can be generated, based on the existing literature about IPOs and internet companies in the dot-com bubble. The first hypothesis (H1) is about the internet companies which had their IPO in different periods: internet companies which had their IPO during the dot-com bubble have a higher percentage gain on the first trading day, compared to internet companies which had their IPO in the period after the dot-com bubble. The high first day gains of internet companies which had their IPO in the dot-com bubble was already shown by Ljunqvist and Wilhelm (2003). However, this paper dates back from 2003, when there were almost no internet companies with their IPO after the dot-com bubble which could serve as a control group. This limitation questions the strength of evidence of their research: the high first day gains can not be compared to the first day gains of internet companies which had their IPO in a later period. It is interesting to see how the under-pricing during the dot-com bubble holds, compared to the internet companies of the last ten years. The second hypothesis (H2) is about the differences between internet and non-internet companies: internet companies have a higher percentage gain on the first day, compared to non-internet companies. In the research of Liu and Ritter (2011) a dummy variable for non-internet companies was included as a control variable. The coefficient of this dummy variable was positive, therefore a higher first day gain of internet companies is also expected in this research paper. The third hypothesis (H3) is about the relationship between total asset value and percentage gain on the first day: the total asset value of a company has no effect on the percentage gain on the first day of that company. This third hypothesis is based on findings of Boulton et al. (2011), who showed that the size of an IPO does not matter for IPO under-pricing. The three hypotheses will be tested and the results can be found in the fifth section.
IV. Data and Model
The dataset which is used for this research consists of U.S. companies which had their IPO during or after the dot-com bubble. For each company, data was obtained on the following characteristics: type of company (internet or non-internet), sector of company (total of 8 different sectors), IPO date (dot-com or later), IPO price, closing price first trading day, percentage gain first day and the total asset value of the company in the year of the IPO. All these characteristics are needed for the tests of this research and will first consecutively be explained.
Two types of companies are defined in this dataset: internet companies and non-internet companies. A total of 105 U.S. companies are investigated in this research, from which 32 are noted on the NYSE and the other 73 are noted on the NASDAQ. This total number of companies can be split up in 64 internet companies and 41 non-internet companies. 37 Of the 64 internet companies are part of the Dow Jones Internet Composite Index (DJINET), which is an index consisting of 40 of the largest and most traded U.S. internet companies. For this index, an internet company is defined as a company which gets at least 50% of their revenue from the internet. These 37 companies from the DJINET are complemented by 27 other U.S. internet companies. The 41 non-internet companies are U.S. companies from various sectors, which are chosen based on their IPO year and asset size. This selection of companies is necessary to make sure a fair comparison between the two groups can be made. The sector in which a company is active is based on the classification by Yahoo finance. The 105 companies are divided in 8 different sectors (the number of companies is in parentheses): Applications software (24), Internet information providers (14), Internet software & services (9), Other internet (17), Biotechnology (18), Independent oil & gas (11), Scientific & technical instruments (6) and Other non-internet (6). Note that ‘other internet’ and ‘other non-internet’ are no real sectors, they contain companies which are active in a smaller sector or in sectors from which just a few companies are included in the dataset (not enough companies for a representative sector).
The IPO date of each company determines if the company had their IPO during the dot-com bubble. There is some discussion about the exact date in which the dot-com bubble occurred, but the majority (including the papers mentioned in the literature review) defines the period as roughly 1997 till 2000. Therefore, companies which had their IPO during these years are defined as a company which had their IPO during the dot-com bubble. However, there are also two companies included which had their IPO just before this period (one in 1995 and one in 1996). This has been done because some argue that the dot-com bubble already started in 1995 instead of 1997. The dataset which is used, has a total of 43 companies which had their IPO in this period, 26 of these companies are internet companies and 17 are non-internet companies. The other 62 companies had their IPO on a later date, ranging from 2002 to 2016. The exact composition of the dataset is also shown in table 1.
Table 1- Composition of Sample Group by IPO year/period
All companies Internet company Non-internet company Number of observations 105 64 41 IPO year/period 1995-1997 1998 1999 2000 2002-2004 2005-2007 2009-2012 2013-2014 2015-2016 Total # of firms % share 12 0.11 11 0.10 14 0.13 6 0.06 6 0.06 6 0.06 23 0.22 22 0.21 5 0.05 105 1 # of firms % share 7 0.11 7 0.11 9 0.14 3 0.05 4 0.06 3 0.05 15 0.23 13 0.20 3 0.05 64 1 # of firms % share 5 0.12 4 0.10 5 0.12 3 0.07 2 0.05 3 0.07 8 0.20 9 0.22 2 0.05 41 1
Note: No significant difference is found by comparing the two proportions.
The IPO price is the official price at which the stocks are offered to the public. The price range is determined by the underwriter, based on the company’s current financial situation and future cash flows. The final price of most IPOs lies in this range, but this is not always the case. This price was retrieved from the official NASDAQ site for almost every company in the sample. The IPO price of the remaining companies was obtained through DataStream or WRDS.
The closing price is the final price of the stock when the trading day is over. The closing price on the first trading day is used for this research. The closing prices in this dataset were gathered from the Wharton Research Data Services and were also checked by DataStream. This closing price minus the IPO price, divided by the IPO price (times 100%) is the percentage gain on the first trading day. This percentage gain is used as a measurement of the under-pricing of a company at their IPO. Figure 1 displays the mean percentage gain on the first trading day per year for each type of company. The mean percentage gain of internet companies is higher in every year (except for 2012), compared to the non-internet companies. This finding corresponds with the expectation that internet companies
on average have a higher first day gain, compared to non-internet companies (H2). The spike in the graph shows the extreme first day gains of internet companies during the dot-com bubble. The second spike around 2011 is most likely the consequence of a recovering economy after the financial crisis.
Figure 1 – Mean percentage gain first trading day per year for each type of company
Note: Figure 1 shows the mean percentage gain on the first trading day for each type of company. The years 1995, 1996, 2015 and 2016 are excluded from this figure due to the small number of companies in the dataset. N=98
The mean percentage gain per type of company and period are displayed in table 2. It is noteworthy that the mean of the percentage gain on the first day is much higher in both periods for internet companies, compared to non-internet companies. Something which is also worth mentioning is that companies which had their IPO during the dot-com bubble have a higher percentage gain, compared to companies which had their IPO after this period. These two differences are in line with the first two hypotheses, but these will first be tested on significance to really be able to say something about them. This is done in the following section.
Table 2 – Mean and Standard deviation of % Gain first day and Total asset value IPO during dot-com
bubble
IPO after dot-com bubble
All Companies Internet Non-internet Internet Non-internet % Gain first trading day Mean St. Dev. 45.03 (70.26) 95.88 (108.34) 35.04 (71.23) 31.95 (31.22) 17.74 (20.24) Total asset value Mean St. Dev. 1060.94 (2452.16) 585.41 (1632.66) 1361.67 (3903.52) 1077.17 (2502.59) 1337.37 (1792.90) N 105 26 17 48 24
Note: All gains are in percentages, so 45.03 indicates a first trading day gain of 45.03%. The total asset value is in million $.
The total asset value is the sum of the current and long-term assets, as noted on the balance sheet in the annual report of each company. Only the asset value of the IPO year is used in this research. This value is used because it is an objective measure, especially compared to the market capitalisation value which is partly based on the expectations of the market. The expectations of the market change a lot, which makes the market capitalisation also volatile. The total asset values are gathered from the WRDS-database and the mean and standard deviation are displayed in table 2. Remarkable is the large difference in total asset size of the internet companies from both periods of time. Internet companies which had their IPO after the dot-com bubble were on average almost twice as large as the internet companies which had their IPO during the dot-com bubble (table 2 row 2). This shows that the internet companies during the bubble were really new and therefore not as large in total asset size during their IPO year as the internet companies of the last ten years. The non-internet companies on the other hand were really similar in total asset size in both periods of time.
All the data is needed for the estimations in this research. The dependent variable is the percentage gain on the first trading day, which is the measurement for the under-pricing at the IPO. The independent variables are IPO in dot-com bubble (dummy), internet company (dummy), natural logarithm of total asset value and sector of company. The log of the total asset value is chosen, because no linear relationship between the asset value itself and the percentage gain on the first day is expected. This decision is in line with the research of Liu and Ritter (2011), who also used the logarithm of the size. A test in which each variable is used, looks like the equation below, in which 𝛽0 is the
constant, 𝐷𝑜𝑡𝑖 a dummy variable for IPO in the dot-com bubble (1 if true, 0 otherwise), 𝐼𝐶𝑖 a dummy
two dummy variables, 𝐿𝑜𝑔𝐴𝑉𝑖 the log of the total asset value, 𝑆𝑒𝑐𝑡𝑖 an indicator for the sector in which
the company is active (ranging from 1 – 8) and 𝜀 the error-term.
% 𝑔𝑎𝑖𝑛 𝑓𝑖𝑟𝑠𝑡 𝑡𝑟𝑎𝑑𝑖𝑛𝑔 𝑑𝑎𝑦 = 𝛽0+ 𝛽1∗ 𝐷𝑜𝑡𝑖+ 𝛽2∗ 𝐼𝐶𝑖+ 𝛽3∗ 𝐼𝐶𝑖∗ 𝐷𝑜𝑡𝑖+ 𝛽4∗ 𝐿𝑜𝑔𝐴𝑉𝑖+ 𝛽5∗ 𝑆𝑒𝑐𝑡𝑖+ 𝜀
V. Results
The results of the empirical testing of the three hypothesis are shown in this section. The first hypothesis stated that internet companies which had their IPO during the dot-com bubble have a higher percentage gain on the first trading day, compared to internet companies which had their IPO after the dot-com bubble. The regression in table 4 below checks this hypothesis. The dependent variable of this regression is the percentage gain on the first trading day and the independent variables are: IPO in dot-com bubble (1 if true, 0 otherwise), Internet company (1 if true, 0 otherwise), an interaction variable of internet company and IPO in dot-com bubble (both dummy variables), Log (total asset value) and sector (ranging from 1 - 8).
The main variable of interest here is the interaction variable between ‘internet company’ and ‘IPO in the dot-com bubble’. This variable is positive and significant in all three regressions in which it was included (two sided t-test with significance level 0.10 in 4&5, 0.05 in 6). This indicates that internet companies which had their IPO in the dot-com bubble had an on average 46% (or 67% column 6) higher first day gain, compared to the other companies (ceteris paribus). This difference is large and shows that internet companies during the dot-com bubble were more under-priced, compared to the other companies in the dataset. The variable ‘IPO in dotcom’ is not significant in the last three columns, which suggests that the non-internet companies are not more under-priced during the dot-com bubble. This can be explained by looking at the hype regarding both type of companies: internet companies were hyped during the dot-com bubble and not hyped in the period after the dot-com bubble. While no hype existed in both periods for the non-internet companies. The existence of a hype was expected to have a positive effect on the under-pricing of companies at their IPO. The results displayed in table 4 indicate that the hype has caused more under-pricing of internet companies during the dot-com bubble to occur. Our findings thus confirm the first hypothesis.
Table 4 – Regression output Regression (1) (2) (3) (4) (5) (6) Dot 0.45654*** 0.44638*** 0.39657*** 0.17306 0.16213 0.04081 (0.15082) (0.15865) (0.15042) (0.17580) (0.16227) (0.14581) 0.33392*** 0.33232*** 0.14215** 0.14031** (0.12061) (0.12370) (0.06555) (0.06655) 0.46620* 0.46658* 0.67168** (0.28040) (0.28074) (0.27843) -0.00906 0.01015 -0.00955 0.01548 (0.05274) (0.06339) (0.05144) (0.06116) 1) 0.27185 1) 0.09548 (0.18367) (0.14190) 2) 0.41875** 2) 0.13713 (0.20594) (0.17656) 3) 0.22666 3) 0.06666 (0.20923) (0.21391) 4) 0.33655 4) -0.04513 (0.27910) (0.26771) 6) -0.16703 6) -0.16890 (0.20464) (0.18705) 7) 0.34245 7) 0.51957 (0.39855) (0.39944) 8) -0.17490 8) -0.18915 (0.21473) (0.19803) IC IC*Dot AV Sect N 105 105 105 105 105 105 Constant 0.05983 0.11646 0.04630 0.17737 0.23715 0.13962 R2 0.1561 0.1565 0.1806 0.1817 0.1821 0.2218
Note: Robust standard errors are reported in parentheses under the coefficients. *** means that the coefficient is significant different from 0 using a two-sided t-test with significance level 0.01 (** with significance level 0.05 & * with significance level 0.1). Dependent variable: Percentage gain first day. N=105
The second hypothesis is about the differences between internet and non-internet companies: internet companies have a higher percentage gain on the first day, compared to non-internet companies. The main variable of interest for this hypothesis is the dummy variable for internet companies (IC). This dummy variable is 1 if the company is an internet company and 0 otherwise. The results in table 4 are significant and show the positive relationship between internet companies and IPO under-pricing. Internet companies in general have an on average 33% (column 2&3) higher first day gain, compared to non-internet companies. Internet companies which had their IPO in the dot-com bubble have an even higher first day percentage gain (column 4&5), dot-compared to non-internet
companies. In the third and sixth column is the sector in which each company is active included as a control variable. These sectors are compared to fifth sector (‘Biotechnology’), which is the reference group. This sector is chosen as the reference group because it has average first day gains (22.65%), compared to all the non-internet companies (24.91%) and enough companies (18) to give a good representation of the total sector. Only the ‘internet information provider’ sector (sector 2) is significantly different from the reference group. The companies active in this second sector have an on average 41.87% higher first day gain, compared to the companies which are active in the Biotechnology sector. Remarkable is that the other sectors (expect 6 & 8) all have a much higher percentage first day gain, compared to the biotechnology sector. However, these differences are not significant. These differences that are not significant can partly be explained through the relative small sample size. Note that the sector coefficients depend on the choice of the reference group, another sector as a reference group would lead to other results. Therefore, the effect of the sectors should mainly be used to get some insight about the different sectors (and to be included as control variables). The third hypothesis stated that the total asset value of a company has no effect on the percentage gain on the first day of that company. Column 2, 3, 5 and 6 all have the logarithm of the total asset value of a company as a variable included. In column 3 and 6 has this variable a positive coefficient, while in column 2 and 5 the coefficient of this variable is negative. The most important conclusion which can be drawn from the regression outputs is that all these coefficients are not significantly different from zero (two sided t-test with significance level 0.05). This implies that the total asset value of a company has no significant effect on the percentage gain on the first trading day of that company. Therefore, the third hypothesis does hold, based on the results of the regressions. This finding is in line with the research of Boulton et al. (2011), who also found that the size of the IPO has no effect on the under-pricing.
VI. Conclusion
In this paper are both internet and non-internet companies investigated, which had their IPO either during or after the dot-com bubble from 1997 to 2000. The results are in line with the three hypotheses stated in the beginning of this paper. Internet companies which had their IPO during the dot-com bubble had an on average 67 percent higher first day gain, compared to internet companies which had their IPO after the dot-com bubble (in line with H1). The second hypothesis that was tested, stated that internet companies have a higher percentage gain on the first trading day, compared to non-internet companies. The results show that non-internet companies have a 33 percentage higher first day gain, compared to non-internet companies. Possible explanations of this higher under-pricing are
uncertainty and information asymmetry. Hoque (2014) already showed that IPOs in which more information asymmetry was involved, have a higher percentage first day gain. Internet companies in general are more difficult to be valuated, compared to non-internet companies because they have more intangible and less tangible assets. This was especially the case during the dot-com bubble when internet companies were new and it was impossible to predict their potentials. This high uncertainty and information asymmetry led to exceptional first day gains of internet companies which had their IPO in the dot-com bubble. The total asset value of a company has no effect on the under-pricing, as predicted by the third hypothesis. In summary, this research shows clear results regarding IPO under-pricing of internet companies which had their IPO during and after the dot-com bubble.
However, some critical notes should be made regarding the research that has been done. First of all is the sample not extremely large with 105 U.S. companies. This number is large enough to give insight about the IPO under-pricing from internet companies in the U.S., but the validity of this research would be even better if the number of companies was increased. The dataset consists only of companies which are still active. The companies which are no longer active are not included in the dataset, which makes the dataset biased (survivorship bias). The observed effect is most likely lower than the true value (negative bias), because companies which had really exceptionally high gains were not able to live up to the high expectations in the period after the dot-com bubble. This would eventually lead to bankruptcy, through which these companies are not included in the dataset for this research. This bias holds for every type of company in the dataset, internet and non-internet companies but also dot-com and after dot-com bubble companies. The including of companies which are currently not active anymore would correct for this bias. Adjustments that would also improve the validity of this research are to include companies from different countries than the U.S. and more control variables like the type of underwriter at the IPO (prestigious or not). In the future, research on this topic could perhaps implement these adjustments to see if the findings are still the same. The research that was carried out could also be extended by looking at the long-term company performances of the under-priced companies. It would be interesting to see whether there exists a relationship between IPO under-pricing and company performances on the medium and long run. This would show whether the internet companies have justified their high first day gains in the long-run and if the dot-com bubble was really a bubble. Something which would also be interesting is to perform a similar research in another industry which has or had a certain hype around it. The cryptocurrency market could be suited for a similar research, but on the day of writing is it still way too early for this. Finally, the results mentioned in this paper can also be used as a warning for companies to leave ‘less money on the table’ at future IPOs.
Reference list
Aggarwal, R., Krigman, L., & Womack, K. (2002). Strategic IPO underpricing, information momentum, and lockup expiration selling. Journal of Financial Economics, 66(1), 105-137.
Benveniste, L., & Spindt, P. (1989). How investment bankers determine the offer price and allocation of new issues. Journal of Financial Economics, 24(2), 343-361.
Boulton, T., Smart, S., & Zutter, C. (2017). Conservatism and international IPO underpricing. Journal of
International Business Studies, 48(6), 763-785.
Boulton, T., Smart, S., & Zutter, C. (2011). Earnings Quality and International IPO Underpricing. The
Accounting Review, 86(2), 483-505.
Chambers, D., & Dimson, E. (2009). IPO Underpricing Over the Very Long Run. Journal of Finance, 64(3), 1407–1443.
Fama, E. (1991). Efficient Capital Markets: II. Journal of Finance, 46(5), 1575-1617.
Hoque, H. (2014). Role of asymmetric information and moral hazard on IPO underpricing and lockup. Journal of International Financial Markets, Institutions & Money, 30, 81-105.
Liu, X., & Ritter, J.R. (2011). Local underwriter oligopolies and IPO underpricing. Journal of Financial
Economics, 102(3), 579-601.
Ljungqvist, A., & Wilhelm, W. (2003). IPO Pricing in the Dot‐com Bubble. Journal of Finance, 58(2), 723-752.
Loughran, T., & Ritter, J. (2002). Why Dont Issuers Get Upset About Leaving Money on the Table in IPOs? Review of Financial Studies, 15(2), 413-444.
Lowry, M., & Schwert, G. (2002). IPO Market Cycles: Bubbles or Sequential Learning? Journal of
Finance, 57(3), 1171-1200.
Ofek, E., & Richardson, M. (2003). DotCom Mania: The Rise and Fall of Internet Stock Prices. Journal of
Finance, 58(3), 1113-1137.
