The IPO Underpricing and the Relationship between Foreign IPO. Underpricing and Economic Freedom

Transcription

1 The IPO Underpricing and the Relationship between Foreign IPO Underpricing and Economic Freedom by Xiaoxuan Zhou [254009] MSc Tilburg University A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Finance Tilburg School of Economics and Management Tilburg University Supervisor: Prof. dr. F.C.J.M. de Jong August 23, 2014

2 Acknowledge I gratefully and sincerely thank my supervisor Prof. dr. F.C.J.M. de Jong for his guidance, generous time and commitment. He helped me to think independently and improve research skills, and he also greatly assisted me with scientific writing. I also want to thank my family and friends, who supported me a lot and brought me much love during my studies. 1

3 Abstract I use the updated sample of IPOs in the US market to do empirical studies to investigate the factors that lead to IPO underpricing and find out that the ex-ante uncertainty risk of the firms: Firm Specific Risk, IPO Features and Industry Risk do have a strong relationship with IPO underpricing and the results show clearly that the IPOs carried out during internet bubble period (01/ /2000) experience more IPO underpricing than other periods. What s more, when it comes to foreign IPOs, the results show that the initial return (underpricing) is negatively related to the economic freedom, that is, foreign IPOs from countries with lower economic freedom are more underpriced than those from other countries. 2

4 I.Introduction and Literature Review Many researchers have documented an important anomaly that is the underpricing in the initial public offering (IPO) of common stocks. In early researches, such as Logue (1973) and Ibbotson (1975), document that the close price on the first day of trading increases substantially and this means that the price of stocks tends to be underpriced when companies go public. In the 1980s and in 1990s, a large theoretical literature tries to explain why IPOs are underpriced. To sum up, theories explaining IPO underpricing considering key IPOs transaction (the issuing firm, the underwriting bank, the deal characters and investors) can be divided into four groups: asymmetric information explanations, institutional explanations, ownership control explanations, and behavioral explanations (Ljungqvist, 2004). Asymmetric information can be used to explain the IPO underpricing and its assumption is that one party concerning IPOs holds more information than other parties. In the literature written by Baron (1982), it assumes that the bank has more information than the issuer about demand conditions of the market, so in order to solve agent problem the underpricing is generated. Rock (1986) argues that due to the existence of asymmetric information among investors, the relatively uninformed investors tend to receive a greater proportion of the overpriced issues on average. Institutional theories focus on the features of the market to explain the IPO underpricing. For example, Hensler (1995) argues that firms would like to underprice intentionally in order to avoid securities litigation, in which case firms use underpricing as insurance. Benveniste and Wilhelm (1996) point out that the underpricing of IPOs benefits mainly the institutional investors who participate in book building of the IPO. Ownership control theories argue that underpricing is used to reduce intervention by investors outside so as to help build the shareholders base when the firms go public. Brennan and Franks (1997) state that in order to avoid monitoring by a large outside shareholders, firms use underpricing to decide the control of managers and the costs of agents. What s more, Stoughton and Zechner s (1998) also gives suggestion that by encouraging monitoring, agency costs may be minimized by underpricing. Behavioral finance shows interest in the influences of irrational or sentiment investors on stock prices. And behavioral theories assume that the existence of sentiment investors and issuers suffered from behavioral biases force the underwriting banks to reduce underpricing (Ljungqvist, 2004). Ljungqvist et al., (2004) assume sentiment investors always hold optimistic attitude towards the IPO firms future because investors want to maximize the excess valuation (the difference between price and fundamental value) of their stocks. But the price reverts to fundamental value at last, Ritter (1991) and others find that IPO returns are negative in the long-run, meaning the stock value actually revert to its fundamental value at last. What s more, Loughran and Ritter think that IPO underpricing can be explained by behavioral biases among the parties who made decisions of the IPO firms. Empirical evidence proves this theory, for example, Ofek and Richardson (2003) show that during the internet bubble period, high initial returns burst and eventually reversal to sizeable level at the end of Ljungqvist and Wilhelm (2004) test whether the subsequent decisions made by CEOs of IPO firms are consistent with their behavioral prospects of the outcome of IPOs. 3

5 Generally speaking, asymmetric information approaches are the best established among these four theories. And the empirical evidence provided by researchers mostly confirms the view that IPO underpricing is related to information frictions. And nowadays literature concerning the empirical IPO is getting mature and most theories use rigorous empirical testing to study IPO underpricing. Ritter and Welch (2002) study the IPO underpricing concerning the reasons for going public, the pricing and allocation of shares of IPOs, and long-run performance of IPOs. Loughran et al. (2004) use empirical method to find that the average initial return doubles to almost 15% during , and jumps to 65% during (the internet bubble period) and then reverts to 12% during And they think that most of the abnormal higher underpricing during the internet bubble period is resulted from the behavioral bias of issuers, which is same as the results in Ofek and Richardson (2003). Nowadays more and more firms avoid stock exchanges in their own countries and seek public offerings on stock exchanges outside their own countries. Chemmanur and Fulghieri (2006) point out that the number of firms giving up domestic exchanges and practicing IPOs outside their own countries is increasing. And they think this is mainly due to the accelerating pace of international capital market integration. Although there is very little literature studying the foreign IPOs, there is still a growing research on the foreign IPOs. Foerster and Karolyi (1999) point out that there are an increasing number of foreign firms who choose to go public in the US market. And some investigations try to understand why foreign IPOs choose to list on exchanges outside instead of their own countries. Ding et al. (2010) examine the choices between foreign listing (Hong Kong) and home listing (Shenzhen) made by Chinese firms. Similarly, Hursti and Maula (2007) find out that the reasons that many European IPOs give up going public on their domestic exchanges and make offerings on foreign exchanges are mainly due to the international experience of the firms. In addition, some investigations extend to include capital market as an explanation to foreign IPO success. Moore et al. (2010) provide a model studying solely how institutional environments influence foreign IPOs success. And Bell et al. (2012) investigate how institutional environments affect the changing relationship between corporate governance and foreign IPOs success. In addition, some investigations study the underpricing of these foreign IPOs. Francis et al. (2001) compare the underpricing of domestic IPOs and foreign IPOs in the US market during the period And find out that foreign IPOs significantly experience more underpricing than domestic IPOs in this comparable sample. Information asymmetry can be used for explaining foreign IPO underpricing because foreign IPOs suffer more liability that is liability of foreignness than domestic IPOs (Moore et al., 2010). Thus, Petersen and Pedersen (2001) think that foreign IPOs experience considerable more difficulties in seeking offerings in foreign capital markets. Moore et al. (2012) think that the foreign IPOs underpricing performance has a relationship with their original countries institutional factors, including the ability to protect the rights of minority shareholder legally. And they also think that the legal environment plays an important role in limiting agency problems in market. All in all, they suggest that the underpricing of a new foreign IPO is influenced by institutional factors as well as the legal environment. What s more, Bell et al. (2008) extend their research to include macroeconomic of the country and societal issues that may lead to the foreign IPO underpricing in developed capital 4

6 markets such as US and set out to investigate emerging economies which is one of the research questions that researchers care about. Emerging economies are those develop rapidly in economic and often adopt a free-market system, and government policies in emerging economies are often appeared as economic liberalization (Hoskisson et al., 2000). Emerging economics are having more and more influence on the whole world economy nowadays. And researchers such as Bruton et al. (2002), Shenkar and Glinow (1994), Hoskisson et al. (2000) and Wright et al. (2005) have studied the behaviors of firms from emerging economies. In fact, country-specific macroeconomic and societal issues decide to which degree a country promotes competition in free market and they help build the country s economy. And they are also significant factors used to predict performance and survival conditions of firms in developed capital markets when they seek IPOs in foreign countries (Porter, 1990). Bell et al. (2008) find that when foreign firms go public in developed markets, economic freedom has significant impacts on the underpricing levels and is negative related with the foreign IPO underpricing. Firms from countries whose economic freedom level is similar to those of the US experience significantly less underpricing, because those IPOs have relatively high economic freedom level. In the comparative way, IPOs from countries who have lower economic freedom level experience more underpricing. This means that foreign IPO firms country of origin (COO) acts as signals in developed financial markets. Through judging the products or the brands from the countries, foreigners generate generalizations and perceptions of the countries and those are called countries of origin (COO) effects (Lampert and Jaffe, 1996; Roth and Romeo, 1992). I also use characterizes of country of origin (COO) effects such as economic factors and country-level institutional factors to examine the relationship between COO effects and underpricing. First part of my thesis, I will use empirical method to examine the IPO underpricing in the US market according to the newest time period ( ). And I also want to see whether the initial return (underpricing) bursts during internet bubble period (01/ /2000) as behavioral theories suggest (Ofek and Richardson, 2003; Loughran, Ritter, 2004). Then I follow the step of Bell et al. (2008) to examine the relationship between foreign IPO underpricing and economic freedom of country over the newest time period covering internet bubble period which is new in academic. And I will take more other factors that may influence IPO underpricing into consideration, both for the whole sample and foreign IPO sample. The remaining parts of my thesis are organized as follows. Part II presents the hypotheses and the development of methodology. The following Part III describes the data; the empirical results are presented in Part IV. The final Part V concludes the thesis. 5

7 II. Hypotheses and Methodology Development My thesis is going to study the factors that lead to IPO underpricing and the relationship between economic freedom and foreign IPO underpricing based on the updated data. When I say a foreign IPO in this thesis, it refers to an IPO that goes public in the US but whose domicile nation is other countries rather than US. In the similar way, a domestic IPO is an IPO that goes public in the US whose domicile nation is also US. I define underpricing or the initial return (IR i1 ) as same as most other researchers do, as the change from the initial offer price (P i0 ) to the close price on the first day trading (P i1 ) for firm i: IR i1 = (P i1 /P i0 ) 1 The greater the initial return is, the more underpricing the IPO experiences, and at the same time, the more money is left on the table by issuing firms and selling shareholders particularly. So we can use the amount of money left on the table (the number of shares sold at offering multiplied by the difference between the close price of first trading day and the offer price) as an alternative way to calculate underpricing. Ritter (1984) and Beatty and Ritter (1986) provide a key empirical implication: underpricing increases at the same time ex-ante uncertainty risk of the IPO firm s value increases. In order to study the relationship between ex-ante uncertainty risk and IPO underpricing, I use proxies in 3 aspects for ex-ante uncertainty risk of the firms: Firm Specific Risk, IPO Features and Industry Risk. As for firm specific risk, I use age, market capitalization, and firm valuation, EPS (earing per share) at offering, assets at offering and intangible assets at offering as proxies. Many researchers such as Ritter (1984), Megginson and Weiss (1991), Ljungqvist and Wilhelm (2003), and others all use age as the proxy variable and they all find that old firms experience less underpricing than young firms. In my thesis, I also add firm age at the time of IPO as an explainable variable and firm age is calculated from the difference between the year when the underlying firm founds and the year when the firm takes an IPO. Chambers and Dimson (2009) add market capitalization as a proxy of firm specific risk, which is measured as the voting stock market capitalization based on offer price of IPO, the bigger IPO market capitalization is, the less IPO underpricing the firm will experience. Firm valuation is measured by the ratio of offer price per share to book value where book value is calculated according to an audited year before IPO. The ratio indicates how much equity investors are paying in net assets for each dollar, so the higher the ratio is, the less underpricing the IPO will be (Chambers and Dimson, 2009). What s more, EPS before offering is the earing per shares at the offering calculated by an audited year before IPO. Lager number of EPS means that the firm has good profitability performance, thus leading to less underpricing when the firm goes public. If a firm has more total assets before offering, the firm is expected to provide more future economic benefits, in this way the firm will experience less IPO underpricing. If firms want to gain new scientific or technical advantage, they tend to carry out original and planned investigations; in this case, intangible assets increase. Thus more intangible assets may result in less IPO underpricing. Additional measures of IPO underpricing ex-ante uncertainty risks include IPO features. 6

8 Many literatures such as Booth and Smith (1986), Carter and Manaster (1990), Michaely and Shaw (1994) etc. all point out that hiring a prestigious underwriter is a good way to relief the IPO underpricing and all results are empirical significantly. But in my thesis I will study whether IPO underpricing is related to the number of book runners (when a company use book building to go public, it select one or more managers as the lead underwriters and in most cases lead underwriters are book runners.). As pointed in Loughran and Ritter (2002, 2004), book runners leave more money on the table than the amount adequately demanded for the issue because they expect to receive soft dollar commission in the aftermarket. But if there are multiple book runners, there will be less money left on the table. The reason is that, at the pricing meeting, the issuers of IPOs carried out by multiple book runners have more bargaining power. And this situation may lead to a higher offer price so the IPO is less underpriced in this case. This is same with Corwin and Schultz (2005) who point out that when multiple book runners are employed during the IPO, more price adjustment and less IPO underpricing occur. Considering these research, I assume IPO underpricing is negative related to the number of book runners. Yeoman (2001) shows that there existents a trade-off between IPO underpricing and gross spread. He argues that firms experience lower underpricing when they incur a higher gross spread during the IPOs. Because firms whose gross spreads are high in IPOs want to participate highly so as to reduce IPO underpricing when going public. Habib and Ljungqvist (2001) also indicate that the IPO underpricing is negatively related to the marketing costs of IPOs. Based on these prior researches, I expect that there is a negative relationship between IPO underpricing and the percentage of gross spread to principle amounts. In the paper done by Franzke (2004), Lee and Wahal (2004) and many other researchers, they all find that IPOs backed by venture capitals experience more underpricing than comparable those who are not backed by venture capitals. And I also hope to see this phenomenon in my thesis. As the newest behavioral explanation of IPO underpricing suggest: the existence of sentiment investors and issuers suffered from behavioral biases force the underwriting banks to reduce underpricing (Ljungqvist, 2004). Ofek and Richardson (2003) show that during the internet bubble period, high initial returns burst and eventually there are sizeable reversals at the end of Loughran et al. (2004) find similar result that IPOs carried out during internet bubble period (01/ /2000) are more underpriced. And in the research done by Loughran et al. (2004) and Chambers and Dimson (2009), they all find that IPOs which are carried out by companies who belong to technology industry experience more underpricing than others when taking initial public offerings. And all these findings help explain IPO underpricing through behavioral theories. I also hope to check these points in my thesis. Francis et al. (2001) find out that foreign IPOs tend to experience significantly more underpricing than domestic IPOs when they seek offerings in developed markets. And this is mainly due to the asymmetry information among investors in developed markets. Moore et al. (2012) think that the foreign IPOs underpricing performance has a relationship with their original countries institutional factors, including the ability to protect the rights of minority shareholder legally. And they also think that the legal environment plays an important role in limiting agency problems in market. What s more, compared to domestic firms, foreign firms have disadvantages when seek IPOs in developed markets and we generally refer these disadvantages as liability of foreignness (Hymer, 1960). Liability of foreignness are usually 7

9 generated because investors in the domestic market lack knowledge of the foreign firms and this causes costs such as regulatory costs, information costs etc. to the foreign firms when they seek initial offerings in the developed markets (Bell et al., 2010). What s more, this liability of foreignness may simply result from domestic investors home bias towards foreign IPOs, since domestic investors always prefer domestic IPOs better than foreign IPOs. Thus, I assume here that the IPOs tend to be underpriced more when it is carried by foreigners. Table 1 summarizes all the hypotheses related to underpricing developed. Table 1 Testable Hypotheses Regarding IPO Underpricing The hypotheses below are used in this thesis as explanations for IPO underpricing or the initial return, IR, which is defined as the percentage change from the initial offer price to the close price on the first day trading, Hypotheses Impact on Underpricing (a) Firm Specific Risk Age at the Time of IPO Negative Market Capitalization Negative Firm Valuation Negative EPS at Offering Negative Assets at Offering Negative Intangible Assets at Offering Negative (b) IPO Features Number of Book Runners Negative Percentage of Gross Spread of Principle Amounts Negative Venture backed Positive (c) Industry Risk Technology Company Positive (d) Internet Bubble Risk Period during 01/ /2000 Positive (e) Domicile Nation Foreign IPO Positive I employ the explanatory hypotheses listed in Table I in a linear model of initial returns or the underpricing of IPOs, IR, as follows: IR=b 1 * ln(age+1) +b 2 * lnmcap +b 3 * lnpbv +b 4 * lneps + b 5 * lnassets + b 6 * lnintangible Assets + b 7 * Nobookrunners + b 8 * PerGroSpread + b 9 * VBDummy + b 10 * TechDummy + b 11 * BubbleDummy + b 12 * ForeignDummy +ε The error term, ε, is assumed to be i.i.d. normal. And the definitions of other variables are shown in Table 3. Nowadays more and more foreign firms especially some from emerging economies increasingly hope to be listed in developed markets such as US. Are there any differences between foreign IPOs and domestic IPOs and what factors may affect foreign IPO underpricing? 8

10 Both individual investors and institutional investors make decisions about whether foreign firms go public outside their own countries in developed markets. What s more, when foreign firms undertake IPOs outside in developed markets, investors in those markets face more uncertainty risk such as the institutional environment and macroeconomic of the firms country of origin (COO). In order to determine a given country s institutional environment and macroeconomic, perceived liberty of a given country acts as a key element here and the perceived liberty of a given country is measured using a country s economic freedom (Doucouliagos, 2005). If a country s government poses more constraints on the production, distribution, or consumption of goods, economic freedom of that country tend to be in the lower level (Doucouliagos, 2005). Economic freedom measures the degree to which a country s government and institutions help protect investors to make their choices independently and exchange goods voluntarily. In addition, it s also an indication of the degree of freedom to compete in free markets. Economic freedom is an increasingly important element in the researches determining institutional environment and macroeconomic of a given country. Bell et al. (2008) study the sample of foreign IPOs listed on US stock exchanges and get the result that firms from countries whose governments and institutions doing their best to help build high economic freedom of people experience significantly less underpricing. In the opposite side, IPOs of firms from countries whose economic freedom is low experience significantly more underpricing. To sum up, the IPO underpricing decreases as a country s economic freedom level increases. At last I state the hypothesis formally as follows: foreign IPO underpricing is negatively related to the country s economic freedom level (Gwartney et al., 2005). Table 2 Testable Hypotheses Regarding Foreign IPO Underpricing The hypotheses below are in this paper as explanations for foreign IPO underpricing or the initial return, IR, which is defined as the percentage change from the initial offer price to the close price on the first day trading. Hypotheses Impact on Underpricing Economic Freedom Negative Emerging Market Positive I employ the explanatory hypotheses listed in Table 2 in a linear model of initial returns or underpricing of IPOs, IR, as follows: IR=a*lnEconomicFreedom+b* EmergingDummy+c* Controls+ε The error term, ε, is assumed to be i.i.d. normal. Control variables here are the factors that influence the initial return (underpricing) such as firm specific risks, IPO features, industry risk and internet bubble period etc. which I have discussed earlier. And the definitions of other variables are shown in Table 3. 9

11 Variable Ln(Age+1) LnMcap LnPVB LnEPS LnAssets LnIntangibleAssets Nobookrunners PerGroSpread VBDummy TechDummy BubbleDummy ForeignDummy LnEconomicFreedom EmergingDummy Table 3 Variable Definitions Definition The natural logarithm of the age plus 1, measuring age at the time of IPO The natural logarithm of market capitalization at the offer price of IPO The natural logarithm of the ratio of book value to the offer price per share The natural logarithm of earning per share before offering The natural logarithm of assets one year before offering The natural logarithm of intangible assets one year before offering Number of book runners Percentage of gross spread to principle amounts Dummy variable, equaling 1 if IPO is venture backed, and equaling 0 otherwise Dummy variable, equaling 1 if IPO is technology stocks, and equaling 0 otherwise Dummy variable, equaling 1 if IPO is carried out during 01/ /2000, and equaling 0 otherwise Dummy variable, equaling 1 if the IPO is a foreign IPO, and equaling 0 otherwise The natural logarithm of average of Heritage Foundation Index of Economic Freedom of the country Dummy variable, equaling 1 if a foreign IPO is from emerging economies, and equaling 0 otherwise 10

12 III. Data I get the date about IPO from Thomson One (T1). My sample comprises all firms that listed in the US stock exchanges between 1990 and 2013 for which I can obtain both Data from CRSP and Compusta database. Just as Ritter and Welch (2002) did, I exclude IPOs with an offer price below $5.00 per share, unit offers, REITs (real estate investment trusts), closed-end funds, banks and savings and loans (S&Ls), ADRs (American Depository Receipts, issued by foreign firms that list in at least one other market outside the US), and IPOs not listed on CRSP within six months of issuing. The date of the company s age is obtained from the Jay R. Ritter s website 1. After combining all these data I get 5872 IPOs meeting the standard. To identify companies whether they are in technology industry, I use SIC code to distinguish them. I do as Loughran et al. (2004) do, defined IPOs as Technology stocks if their SIC codes are 3571, 3572, 3575, 3577, 3578 (computer hardware), 3661, 3663, 3669 (communications equipment), 3671, 3672, 3674, 3675, 3677, 3678, 3679 (electronics), 3812 (navigation equipment), 3823, 3825, 3826, 3827, 3829 (measuring and controlling devices), 3841, 3845 (medical instruments), 4812, 4813 (telephone equipment), 4899 (communications services), and 7371, 7372, 7373, 7374, 7375, 7378, and 7379 (software). I use the same Index as Beach and Miles (2004) and Bell et al. (2008) to measure economic freedom, which is the Heritage Foundation Index of Economic Freedom. This is an annual ranking index to measure the degree of economic freedom in the world's countries and this created by the Heritage Foundation and The Wall Street Journal in The index use statistics from organizations like the World Bank, the International Monetary Fund and the Economist Intelligence Unit to score countries on 10 broad factors of economic freedom: Business Freedom, Trade Freedom, Monetary Freedom, Government Size/Spending, Property Rights, Financial Freedom and Labor Freedom. Each of the 10 factors is graded using a scale from 0 to 100, where 100 represent the maximum freedom and 0 represents the minimized freedom. These 10 factors are averaged equally into a total score. The higher the index is, the higher the economic freedom level the country has. 2 The Heritage Foundation Index of Economic Freedom to measure the degree of economic freedom can be obtained from the Heritage s official website since My foreign IPO sample comprises all the firms that go public in the US stock exchanges but whose domicile nation are not US in , which I call them foreign IPOs here. The definition of emerging markets is referred to the list created by the International Monetary Fund (IMF) who the labels the "emerging economies" as follow countries in of July 16, 2012: Argentina, Brazil, Bulgaria, Chile, China, Colombia, Estonia, Hungary, India, Indonesia, Latvia, Lithuania, Malaysia, Mexico, Pakistan, Peru, Philippines

13 IV. Empirical Results A.IPO Underpricing A.1 The Initial Return I plots the average initial return and the volume of the IPOs each year over periods in Figure 1. Table 4 shows the means of the initial returns (underpricing) by offering year and by sub-period. Throughout all the analysis, my sample is always split into four sub-periods: 01/ /1998, 01/ /2000 (the internet bubble period), and 01/ /2013. We can see from the figure 2 that, The average initial return is almost modest during 1990 and 1998, but then jumps to 69% during the internet bubble period (01/ /2000) which is same as I assume in Part II and this is due to the irrational behavior of investors during this period (Ofek and Richardson, 2003; Loughran et al., 2004). In the following 2 years of internet bubble, annual IPO volume drops to 70 issues with an average initial return of approximately 10%. Then the initial return lasts stable again for about 8 years, and drops dramatically to about 5% in 2008 with only about 26 IPOs largely due to the big US sub-prime crisis, correspondingly, the percentage of negative initial return climbs to the highest which is almost 54% in 2008 while the percentage of negative initial return is almost stable during other period. This means that during internet bubble period IPOs have more underpricing and have less IPO underpricing, since behavioral theories assume that the existence of sentiment investors and issuers suffered from behavioral biases force the underwriting banks to reduce underpricing (Ljungqvist, 2004). And the results are consistent with the results done by Ofek and Richardson (2003) and Loughran et al. (2004). 12

14 % 60.00% 40.00% 20.00% 0.00% No. of IPOs Average IR % of IPOs with Negative IR Figure 1: Number of IPOs, Average Initial Returns (Underpricing), and Percentage of IPOs with Negative Initial Return The average initial return (underpricing) is defined as the average of the initial return (the percentage change from the initial offer price to the close price on the first day trading) of IPOs of each year. Data are from Thomson One (T1) and other sources. IPOs with an offer price below $5.00 per share, unit offers, REITs (real estate investment trusts), closed-end funds, banks and savings and loans (S&Ls), ADRs (American Depository Receipts, issued by foreign firms that list in at least one other market outside the US), and IPOs not listed on CRSP within six months of issuing are excluded. The data plotted are reported in Table 4. Table 4 shows that the average initial return is 23.11% during the whole period. The average initial return in each year from 1998 to 2000 has higher value than in any other year. The IPO underpricing as well as the amount of money left on the table fly up high in the internet bubble period (01/ /2000). After about 2 years of the internet bubble period, the number of IPOs drops dramatically to around 70 with initial return at the normal level. In year 2008, IPO underpricing undergoes a big depression and the amount of money left on the table appears to be negative and the percentage of negative IPOs seems to have the highest value, meaning that all the IPOs carried out in 2008 experience less underpricing. In the following several 2 year the number of IPOs is also below the average due to the effects of economic depression of US in

15 Table 4 Number of IPOs, Average Initial Returns, Percentage of IPOs with Negative Initial Return, Standard Deviation of Average Initial Return, Amount of Money Left on the Table, by Cohort Year, 1990 to 2013 The average initial return (underpricing) is defined as the average of the initial return (the percentage change from the initial offer price to the close price on the first day trading) of IPOs of each year. Money left on the table is calculated as the number of shares sold at offering multiplied by the difference between the offer price and the close price of first trading day. Data are from Thomson One (T1) and other sources. IPOs with an offer price below $5.00 per share, unit offers, REITs (real estate investment trusts), closed-end funds, banks and savings and loans (S&Ls), ADRs (American Depository Receipts, issued by foreign firms that list in at least one other market outside the US), and IPOs not listed on CRSP within six months of issuing are excluded. Year Aggregate SD of No. of % of IPOs with Money Left on Proceeds, Average IR Average IPOs Negative IR the Table, $mil Initial Return $mil % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % Year Periods % % % % % % Total % %

16 One more step further, when I get tri-sectional quantiles of all the variables (sort the variables from small value to big value and divide them into three sections equally) that affect the initial return (underpricing) as I assume in Part II, I get the results as shown in Table A.1 (Appendix). In panel A, the average initial return (underpricing) decreases as the quantile of firm s age increases and the results are significant at 1% level, which indicating that there is a negative relationship between underpricing and age, same as the hypothesis in part II. What s more, initial returns (underpricing) are lower when the quantiles of earning per share at the time of offering, intangible assets, the number of book runners at the issuing, and the percentage of gross spread of principle amounts are higher, which are also significant at 1%. All these results are consistent as the hypotheses in Part II. As for market capitalization, the initial return (underpricing) in high quantile is higher than the initial return (underpricing) in low quantile, indicating a positive relationship between IPO underpricing and market capitalization, which is different as Chambers and Dimson (2009). In order to see where this difference comes from, I set the quantiles of market canalization into smaller section, let s say, 5. The result is quite confusing; the initial return in the middle quantile of market canalization seems to have the highest value and decreases as the quantile increases and decrease. But when it comes to firm valuation which is measured by the ratio of offer price per share to book value where book value is calculated according to an audited year before IPO, the result shows that the IPO tends to be underpriced more when one firm has too high or too low offer price per share to book value. The relationship between initial return (underpricing) and assets is reversal of that of firm valuation; the IPO tends to be unpriced less when the firm has too many or too few assets. In Panel B, I sort the initial returns by whether the IPOs are backed by venture capitals, whether the IPOs belong to technology industries, whether the IPOs are carried out in internet bubble periods ( 01/ /2000) and whether the IPOs are foreign IPOs (those that go public in the US but whose domicile nations are other countries rather than US). The IPOs backed by venture capitals have higher initial return than those who are not. So venture backed IPOs appear to be more underpriced than non-venture backed IPOs, consistent with former literatures. Technology IPOs and IPOs that are occurred in internet bubble are also more underpriced than those are not and these results are all significant at 1% level. Foreign IPOs are underpriced a little bit more than domestic IPOs and the result is significant at 5% level. All these results are consistent with the hypotheses. 15

17 A.2 Multivariate Results The descriptive statistics of variables for the sample of IPOs underpricing among are shown in Table 5. The table shows no very high correlation between variables, so there won t be any threat to the robustness of the results due to serious collinearity problem. Table 5 Descriptive Statistics and Correlation Matrix, 1990 to 2013 Statistic ln(1+age) lnmcap lnpbv lneps lnassets lnintangible Nobook assets runners PerGroSpread Panel A: Descriptive Statistics Mean Median Maximum Minimum Std. Dev Skewness Kurtosis No. Obs Panel B: Correlation Matrix ln(1+age) lnmcap lnpbv lneps lnassets lnintangible Nobook assets runners PerGroSpread ln(1+age) lnmcap lnpbv lneps lnassets lnintangibleassets Nobookrunners PerGroSpread Table 6 shows the OLS regressions of initial return (underpricing) and factors that affect initial return (underpricing) which is stated as the first model in Part II. In Panel A, from first to the eighth regressions, I regress on the variables independently and the results show that all variables are of the expected sign as assumed in Part II except the PBV and Percentage of gross spread to principle amounts. Especially for Percentage of gross spread to principle amounts, its relationship with IPO underpricing is always reversal with other researches through my whole thesis; there occurs a positive relationship between gross spread and IPO underpricing. The coefficients of Age of the company at offering, EPS before offering, Assets and Intangible Assets at issuing and Number of book runners are significant at highest level (1%). In addition, Percentage of gross spread to principle amounts is significant at 5% and Market capitalization is a little lower, significant at 10.3% level. In the ninth and tenth regressions, again the signs are almost as expected apart from Market capitalization, this time 16

18 the results show a positive relationship between market capitalization and IPO underpricing and are significant at 5% and 1% level, respectively. The relationship between age and underpricing the relationship between percentage of number of book runner and IPO underpricing are still negative, but neither of the results is significant. EPS before offering are significant at 5% and 10% level respectively, meaning that if firms have high EPS before IPOs, they tend to be less underpricing. Intangible Assets is significant at ninth regression but become insignificant when I remove the PBV and Assets variables. Percentage of gross spread to principle amounts shows a reversal relationship with IPO underpricing as expected in Part II, becoming significant after removing the PBV and Assets variables. Since there are so many values missing of PBV (only 2580 available), EPS (2061 available) and Intangible assets (1258 available), I regress without these three variables in the eleventh OLS regression. This time there are totally 4679 and the results are quite precise; all the coefficients are significant at 1%. The coefficients of Market capitalization and Percentage of gross spread to principle amounts are still positive, indicating strongly positive relationships with underpricing. And other variables are as expected. When it comes to dummy variables in Panel B: venture backed IPOs, technology IPOs, IPOs carried out during internet bubble period and foreign IPOs, all sign of these results are exactly the same as I assumed in Part II. In the first to the fifth regressions, I can see that the variables are almost significant apart from the Foreign Dummy. This means that IPOs backed by venture capitals, technology IPOs and IPOs carried out during internet bubble period (01/ /2000) tend to be underpriced more than other IPOs. Although the coefficient of Foreign Dummy is not significant, the sign still indicates that the foreign IPOs underprice more than domestic IPOs. In the first regression of Panel C, the results of age, EPS before offering, Assets, Intangible Assets at issuing, VBDummy and Bubbledummy are consistent as hypotheses in Part II, among which EPS before offering, Intangible Assets at issuing and Bubbledummy are statistically significant. In the opposite side, Market capitalization and ForeignDummy show statistically significant reversal relationships with IPO underpricing as expected in Part II. And other variables are all not significant. In the second regression, I remove the PBV and Assets variables based on Table A.1 to see whether this modifies the model. The results are almost same as that of first regression. The coefficients of Number of book runner and Technology Dummy become negative again as assumed and the coefficients of VBDummy and percentage of gross spread to principle amounts become significant at the first regression, meanwhile the coefficients of Intangible Assets again becomes insignificant as same in the tenth regression in Panel A. As there are so few values of PBV (only 2580 available), EPS (2061 available) and Intangible assets (1258 available), I regress without these three variables in the third OLS regression and the results are quite robust as first and second and highly significant. There are all 4679 IPOs in this regression, among which 238 are foreign IPOs and 4441 are domestic IPOs. All the sign of coefficients are expected expect Market capitalization and Percentage of gross spread to principle amounts are still positive, indicating strongly positive relationships with underpricing. And other variables are as assumed. What s more, the relationship between underpricing and foreign IPOs revers in this regression, significant at 5% level, indicates a negative relationship between them. When I regress on the foreign IPOs and domestic IPOs individually in the fourth and fifth regressions, I can see a little difference 17

19 between them. The coefficients of age is positive for foreign IPOs and negative for domestic IPOs, meaning that, for foreign IPOs, firms with old age tend to be underpriced more, but unfortunately the coefficient of age for the foreign IPOs is not significant. Another strange point I can get here is that for foreign IPOs which are backed by venture capitals are underpriced less while domestic IPOs which are backed by venture capitals are underpriced more which is consistent as expected, but again the coefficient of VBDummy for foreign IPOs is not significant. 18

20 Table 6 OLS Regressions of Initial Returns, 1990 to 2013 The dependent variable, IR, is the initial return or underpricing, measured as the percentage change from the initial offer price to the close price on the first day trading. Other variables are defined in Table 3. All regressions are described in the text. All regressions are OLS and standard errors are calculated using White s (1980) heteroskedasticity-consistent method. Panel A regresses on all variables excluding dummy variables. Panel B regresses on all the dummy variables. Panel C regresses on all variables and the fourth and fifth regressions of Panel C are the OLS regressions of foreign IPOs and domestic IPOs respectively. ***, **, and * indicate statistical significance of coefficients at the 1%, 5% and 10% levels respectively. And all the P-values are in parentheses. Dependent Variable ln(1+age) lnmcap lnpbv lneps lnassets lnintangibleassets Nobookrunners PerGroSpread Intercept *** 0.390*** (0.103) 0.285*** (0.002) (0.110 ) 0.193*** *** *** Panle A *** 0.296*** *** 0.303*** *** 0.275*** 0.032** (0.030) (0.965) (0.234) 0.077** (0.019) (0.728) ** ( 0.015) (0.168) * (0.056) (0.313 ) ( 0.274) (0.549) (0.232 ) 0.052*** ( 0.002) *** (0.005) (0.137) (0.412 ) 0.046*** (0.004) ** (0.020) N Adjusted-R *** 0.151*** *** *** 0.079*** *** 19

21 Dependent Variable VBDummy TechDummy Bubbledummy ForeignDummy Intercept *** 0.183*** 0.199*** 0.164*** 0.460*** 0.171*** (0.277) 0.228*** 0.035** (0.048) 0.127*** 0.418*** (0.383) 0.115*** N Adjusted-R Dependent Variable ln(1+age) lnmcap lnpbv lneps lnassets lnintangibleassets Nobookrunners PerGroSpread VBDummy TechDummy Bubbledummy ForeignDummy Intercept (0.134) 0.084** (0.031) (0.760) ** (0.032) (0.220) ** (0.032) (0.621) (0.144) (0.374) (0.716) 0.201** (0.025) ** (0.018) (0.189) Panel B Panel C (0.223) 0.050*** (0.006) ** (0.037) (0.307) (0.822) 0.049*** (0.006) 0.067** (0.012) (0.976) 0.164** (0.023) (0.425) ** (0.017) *** 0.097*** ** (0.016) *** (0.006) 0.085*** 0.053*** (0.002) 0.147*** 0.403*** ** (0.045) *** (0.659) 0.060** (0.037) (0.309) (0.433) 0.065** (0.011) (0.965) 0.126** (0.031) 0.557*** (0.006) _ ** (0.032) *** 0.101*** ** (0.020) *** (0.008) 0.087*** 0.057*** (0.001) 0.148*** 0.394*** _ *** N Adjusted-R

22 0 Initial Return.5 1 B.Foreign IPO Underpricing The hypothesis in Part II predicts that there is a negative relationship between economic freedom and foreign IPO underpricing. In the figure 2, I draw the scatter diagram and fitted line of initial return (underpricing) and economic freedom. The points with lower economic freedom and highest numbers are IPOs from China, whereas the points with highest economic freedom are IPOs from Hong Kong. The figure apparently shows that there is a negative relationship between initial return (underpricing) and economic freedom, indicating that foreign IPOs from countries with lower economic freedom level experience more underpricing than those from other countries when going public in developed markets. From Table 7, I can get that there are all 282 foreign IPOs from 41 different countries during And 27.64% of my sample of foreign IPOs is from countries belonging to emerging economies referring to International Monetary Fund (IMF). I can also see that all the average economic freedom index of the nations from emerging markets is lower than those who are not. Therefore, the emerging markets represent the majority of countries with low economic freedom level ln(economic Freedom) Initial Return Fitted values Figure 2: The Scatter Diagram and Fitted Line of Initial Return and Economic Freedom The Initial return or the underpricing is defined as the percentage change from the initial offer price to the close price on the first day trading. The economic freedom is the average of the Heritage Foundation Index of Economic Freedom Index from for each country. High initial return values are truncated. 21

23 Table 7 Descriptive Statistics of Economic Freedom, by Countries The economic freedom index is the Heritage Foundation Index of Economic Freedom which can be obtained from the Heritage s official website since And I take the average of the index from of each country. The definition of emerging markets is referred to the list created by the International Monetary Fund (IMF). Country Average Economic Freedom Index Emerging Economy Numbers Argentina Australia Bahamas Belgium Bermuda Brazil Canada China Cyprus France Germany Greece Hong Kong India Indonesia Ireland-Rep Israel Italy Kenya Luxembourg Mexico Netherlands New Zealand Panama Peru Philippines Russian Fed Singapore South Korea Spain Sweden Switzerland Taiwan United Kingdom Utd Arab Em Total 27.64%