Initial Public Offering Allocations

Home , Equity issuance, Initial public offering, Issued shares, Listing (finance), Public offering, Stock

by Sturla Lyngnes Fjesme

A dissertation submitted to BI Norwegian Business School for the degree of PhD

PhD specialization: Financial Economics

Series of Dissertations 9/2011

BI Norwegian Business School

Sturla Lyngnes Fjesme Initial Public Offering Allocations

Series of Dissertations 9/2011

ISBN: 978-82-8247-029-2 ISSN: 1502-2099

BI Norwegian Business School N-0442 Oslo Phone: +47 4641 0000 www.bi.no

Printing: Nordberg Trykk

The dissertation may be downloaded or ordered from our website www.bi.no/en/Research/Research-Publications/

Abstract Stock exchanges have rules on the minimum equity level and the minimum number of shareholders that are required to list publicly. Most private companies that want to list publicly must issue equity to be able to meet these minimum requirements. Most companies that list on the Oslo stock exchange (OSE) are restricted to selling shares in an IPO to a large group of dispersed investors or in a negotiated private placement to a small group of specialized investors. Initial equity offerings have high expected returns and this makes them very popular investments. Ritter (2003) and Jenkinson and Jones (2004) argue that there are three views on how shares are allocated in the IPO setting. First, is the academic view based on Benveniste and Spindt (1989). In this view investment banks allocate IPO shares to informed investors in return for true valuation and demand information. Informed investors are allocated shares because they help to price the issue. Second, is the pitchbook view where investment banks allocate shares to institutional investors that are likely to hold shares in the long run. It is argued, by investment banks, that buy-and-hold investors will create price stability that is good for the issuing companies. Finally, is the rent seeking view, or profit sharing view, where investment banks allocate shares to investors in return for kickbacks. There are four types of IPO rent seeking that have been investigated by U.S. regulators (the SEC and the NASD), see Liu and Ritter (2010). IPO allocations can be tied to future corporate business for the banks (IPO spinning), after-listing purchases of the IPO shares (IPO laddering) and stock-trading commissions. Investment banks and companies can also agree on high underpricing in return for after-listing company share coverage from a star analysts provided by the bank (analyst conflict of interest). Underpriced shares are then allocated to bank clients that generate high stock-trading commission for the investment bank. In the paper 'Laddering in Initial Public Offering Allocations' it is investigated if IPO allocations are tied to after-listing purchases of the IPO shares (IPO laddering). In the paper 'Using Stock-trading Commissions to Secure IPO Allocations' it is investigated if IPO allocations are tied to investor stock-trading commission. Private companies that want to list publicly can, as an alternative to the IPO allocation, issue shares in a negotiated private placement to a small group of specialized investors. Most theoretical papers on equity offerings, however, show that IPOs will almost always be preferred to the negotiated private placement by the seller, see Bulow and Klemperer (1996), Bulow and Klemperer (2009) and French and McCormick (1984). Why some companies use private placements has therefore been the focus of many empirical studies in finance, see Wruck (1989), Hertzel and Smith (1993), Barclay et al. (2007), Anshuman et al. (2010) and Cronqvist and Nilsson (2005). The research question addressed in the paper 'Initial Public Offering or Initial Private Placement?' is whether private placements are used, instead of IPOs, to transfer private benefits of control from sellers to buyers. A common contribution of all papers is that we introduce new and unique data on private company share ownership. This data allow us to investigate share allocations questions it has previously been difficult to investigate.

Acknowledgements I am deeply indebted to Professor Øyvind Norli, my supervisor, for all the continued support, guidance and encouragement throughout my time as a PhD student. I would also like to thank Professor Roni Michaely for help and guidance, and for making my stay at Cornell University such a great experience. I am very grateful to François Derrien and Øyvind Bøhren, who gave me many helpful and detailed suggestions on my pre-doctoral defense and who helped me with the job market process. I am grateful to Bruno Gerard for supervising my master degree thesis and for helping me with the job market process and my PhD thesis. I would also like to thank Karin Thorburn, Diane Denis, William Megginson, Paul Ehling, Christopher Vincent, David De Angelis, Alyssa Anderson, Maury Saslaff, Yelena Larkin, Gideon Saar, Jay Ritter, Dag Michalsen and Richard Priestley for support and for commenting on the thesis. I would like to thank my fellow PhD students, Limei Che, Christian Heyerdahl-Larsen, Morten Josefsen, Siv Staubo, Siri Valseth, Nam Huong Dau, Ignacio Garcia de Olalla Lopez, Junhua Zhong, and my friends, Per Helmer Thorkildsen, Henrik Hasner, Kjell Olav Dalen, Jan Kenneth Evanger, Dag Djurovic, Martin Jensen, Per-Eilert Vierli and Øystein Larsen, for support and many interesting economic discussions. Finally, I would like to thank my family, Sølvi Lyngnes, Torbjørn Fjesme, Arvid Lyngnes Fjesme, Sunniva Victoria Fjesme and Hanna Kristiansen, for all the help and support during my time as a PhD student.

Contents

1 Introduction 3 1.1 Laddering in Initial Public Oﬀering Allocations ...... 4 1.2 Using Stock-trading Commissions to Secure IPO Allocations ...... 4 1.3 Initial Public Oﬀering or Initial Private Placement? ...... 4

2 Laddering in Initial Public Oﬀering Allocations 7

2.1 Introduction...... 9 2.2 Related literature ...... 11 2.3 Predictions and testable implications ...... 12 2.3.1 The IPO laddering hypothesis ...... 13 2.3.2 Other testable implications of IPO laddering ...... 14 2.4 The listing process and the incentives to engage in IPO laddering . . . . . 15 2.4.1 Why investment banks use IPO laddering ...... 15 2.4.2 Why laddering investors agree to buy more shares ...... 16 2.4.3 Why IPO laddering is a problem ...... 16 2.5 Datadescription ...... 17 2.5.1 The IPO sample ...... 18 2.5.2 The remaining IPOs ...... 18 2.5.3 Aggregate laddering ...... 18 2.5.4 Variable explanations ...... 19 2.6 Empiricalresults ...... 21 2.6.1 Optimal holdings ...... 24 2.6.2 The eﬀect of IPO laddering ...... 24 2.6.3 Robustness and aggregate IPO laddering ...... 24 2.7 Conclusion...... 24

3 Using Stock-trading Commissions to Secure IPO Allocations 43

3.1 Introduction...... 45 3.2 Related literature ...... 46 3.3 Theoretical predictions and testable implications ...... 47 3.3.1 The rent seeking view of IPO allocations ...... 48 3.3.2 The pitchbook view of IPO allocations ...... 49 3.3.3 The academic view of IPO allocations ...... 50 3.4 Data...... 50 3.4.1 IPO allocations ...... 51 3.4.2 After-listing ownership ...... 51 3.4.3 Variable description ...... 52 3.5 Empiricalresults ...... 53 3.5.1 The rent seeking view of IPO allocations ...... 54 3.5.2 The pitchbook view of IPO allocations ...... 55 3.5.3 The academic view of IPO allocations ...... 55 3.5.4 Robustness ...... 56 3.6 Conclusion...... 56

1 4 Initial Public Offering or Initial Private Placement? 73 4.1 Introduction...... 74 4.2 Literaturereview ...... 75 4.3 Theroadtothelisting ...... 77 4.3.1 The formal listing process ...... 77 4.3.2 A public or a private offering? ...... 78 4.4 Theoretical predictions and testable implications ...... 79 4.4.1 The private benefits of control hypothesis ...... 80 4.4.2 Alternative explanations ...... 81 4.4.3 Other control measures ...... 82 4.4.4 Private benefits of control also after the listing ...... 83 4.5 Data and descriptive statistics ...... 83 4.5.1 Descriptive statistics ...... 84 4.5.2 Variable description ...... 84 4.6 EmpiricalResults...... 85 4.6.1 The private benefits of control hypothesis ...... 85 4.6.2 Alternative explanations ...... 86 4.6.3 Private benefits of control also after the listing ...... 86 4.7 Conclusion...... 87

5 Summary 100

2 1 Introduction

This dissertation consists of three papers; ’Laddering in Initial Public Offering Alloca- tions’, ’Using Stock-trading Commissions to Secure IPO Allocations’and ’Initial Public Offering or Initial Private Placement?’ The rest of this section is organized as follows. I first discuss the common feature of the papers, namely the allocations of Initial Public Offering (IPO) shares. I then briefly discuss the main results in each of the papers. Stock exchanges have rules on the minimum equity level and the minimum number of shareholders that are required to list publicly. Most private companies that want to list publicly must issue equity to be able to meet these minimum requirements. Most companies, that list on the Oslo stock exchange (OSE), are restricted to selling shares in an IPO to a large group of dispersed investors or in a negotiated private placement to a small group of specialized investors. Initial equity offerings have high expected returns and this makes them very popular investments. Ritter (2003) and Jenkinson and Jones (2004) argue that there are three views on how shares are allocated in the IPO setting. First, is the academic view based on Benveniste and Spindt (1989). In this view investment banks allocate IPO shares to informed investors in return for true valuation and demand information. Informed investors are allocated shares because they help to price the issue. Second, is the pitchbook view where investment banks allocate shares to institutional investors that are likely to hold shares in the long run. It is argued, by investment banks, that buy-and-hold investors will create price stability that is good for the issuing companies. Finally, is the rent seeking view, or profit sharing view, where investment banks allocate shares to investors in return for kickbacks. There are four types of IPO rent seeking that have been investigated by U.S. regulators (the SEC and the NASD), see Liu and Ritter (2010). IPO allocations can be tied to future corporate business for the banks (IPO spinning), after-listing purchases of the IPO shares (IPO laddering) and stock-trading commissions. Investment banks and companies can also agree on high underpricing in return for after-listing company share coverage from a star analysts provided by the bank (analyst conflict of interest). Underpriced shares are then allocated to bank clients that generate high stock-trading commission for the investment bank. In the paper ’Laddering in Initial Public Offering Allocations’it is investigated if IPO allocations are tied to after-listing purchases of the IPO shares (IPO laddering). In the paper ’Using Stock-trading Commissions to Secure IPO Allocations’it is investigated if IPO allocations are tied to investor stock-trading commission. Private companies can, as an alternative to the IPO, issue shares in a negotiated private placement to a small group of specialized investors. Most theoretical papers on equity offerings, however, show that IPOs will almost always be preferred to the negotiated private placement by the seller, see Bulow and Klemperer (1996), Bulow and Klemperer (2009) and French and McCormick (1984). Why some companies use private placements has therefore been the focus of many empirical studies in finance, see Wruck (1989), Hertzel and Smith (1993), Barclay et al. (2007), Anshuman et al. (2010) and Cronqvist and Nilsson (2005). The research question addressed in the paper ’Initial Public Offering or Initial Private Placement?’ is whether private placements are used, instead of IPOs, to transfer private benefits of control from sellers to buyers. A common contribution of all papers is that we introduce new and unique data on private company share ownership. This data allow us to investigate share allocations questions it has previously been diffi cult to investigate.

3 1.1 Laddering in Initial Public Offering Allocations IPO laddering is the process where share allocations are tied to the after-listing purchases of the company shares. IPO laddering has been known by regulators for a long time (the SEC sent out warnings to investment banks that laddering is illegal the first time in 1961), but there has been limited empirical research on IPO laddering. A potential reason for this is that it is very diffi cult to investigate laddering because investment banks rarely distribute information about allocation practices. In this paper we use unique data from the Oslo Stock Exchange (OSE) that allow us to observe the after-listing trading of investors that are allocated IPO shares. The data consists of 16,593 combinations of investor IPO allocations, stock-trading commission and after-listing trading on the OSE in the period from 1993 to 2007. This data allow us to investigate laddering at the investor level. The main contribution of this paper is that we show a strong and robust relationship between IPO allocations and the number of shares that are purchased after new listings at the investor level. This relationship is stronger for investors that sell all shares again right after the listing, in underpriced IPOs and in IPOs with a positive drift in the share price after the listing. These are the investors and the IPOs that the existing research identifies as the most likely laddering investors. These findings are consistent with the suspicion that IPO shares are allocated to investors that buy shares dictated by the investment bank after the listing (laddering). This finding extends to Hao (2007) and Griffi n et al. (2007).

1.2 Using Stock-trading Commissions to Secure IPO Alloca- tions Another concern for regulators is that IPO allocations are tied to excessively large stock- trading commissions and that such a practice is illegal kickbacks from investors to investment banks. Using the same data as in ’Laddering in Initial Public Offering Allocations’, we are able to link stock-trading commission and IPO allocation at the investor level. The main finding of the paper is a strong and robust positive relationship between the level of stock-trading commission generated by an investor prior to the IPO and the number of shares the same investor receives through the IPO allocation. This finding indicates that investors are able to buy IPO allocations by trading excessively to generate commission. The finding extends to Reuter (2006), Nimalendran, Ritter and Zhang (2006), Ritter (2003) and Jenkinson and Jones (2004) who all argue that investment banks are likely to allocate IPO shares in return for stock-trading commission.

1.3 Initial Public Oﬀering or Initial Private Placement? Companies can, as an alternative to the IPO, sell shares in a negotiated private placement. Most theoretical research on equity oﬀerings show that auctions, that are similar to IPOs, will in most cases be preferred by the seller of a company. In practice, however, there are many companies that use negotiated private placements to raise equity. Several studies have proposed explanations to this private placement choice. Some papers argue that private placements are used to attract certain investors, to keep management in control, to reduce undervaluation or to reduce problems associated with information asymmetry (Wruck, 1989; Hertzel and Smith, 1993; Barclay et al.,2007; Anshuman et al., 2010;

4 Cronqvist and Nilsson, 2005). Other papers suggests that private placements are used when buyers value private benefits of control over the stand alone cash flow value of the company (Zingales, 1994; Zingales, 1995; Zwiebel, 1995 and Damodaran, 2005). The main contribution of our paper is that we show a strong and robust relationship between private benefits of control, before the initial offering, and the use of private placements. This indicates that private placements are used to transfer private benefits of control from sellers to buyers. This finding supports Zingales (1995) in that private placements are used to transfer company control rights.

5 .

6 2 Laddering in Initial Public Oﬀering Allocations

Sturla Lyngnes Fjesme1 BI Norwegian Business School

JEL classiﬁcation: G3; G24 Keywords: IPO allocations; Laddering; Tie-in agreements; Rent seeking; Equity oﬀer- ings

1 I am very grateful to Øyvind Norli (supervisor), François Derrien, Roni Michaely, Øyvind Bøhren, Bruno Gerard, Karin Thorburn (discussant), Diane Denis (discussant), William Megginson (discussant), Paul Ehling, Christopher Vincent, David De Angelis, Alyssa Anderson, Maury Saslaﬀ, Yelena Larkin, Gideon Saar, and seminar participants at Cornell University, BI Norwegian Business School, the Nordic Finance Network (NFN) workshop in Lund 2010, the Financial Management Association (FMA) Doc- toral Student Consortium in Hamburg 2010, Stockholm University, the University of Gothenburg, the University of Warwick and the University of Melbourne for valuable suggestions. I thank the Oslo Stock Exchange VPS for providing the data, the Financial Supervisory Authority of Norway (Finanstilsynet) and the companies and investment banks that helped locate the listing prospectuses. Part of the article was written while I was a visiting PhD student at the S.C. Johnson Graduate School of Management at Cornell University. I also thank the American-Scandinavian-Association and the Norwegian Central Bank for ﬁnancial support. All errors are my own. Correspondence: BI Norwegian Business School, Nydalsveien 37, 0484 Oslo, Norway, Email address: [email protected], Telephone (USA): +1-607-793-6911, Telephone (Norway): +47-957-722-43.

7 Abstract

Tying Initial Public Offering (IPO) allocations of common stock to after- listing purchases in the IPO shares, a process referred to as IPO laddering, has resulted in large-scale investigations of the major investment banks by the SEC and the National Association of Securities Dealers (NASD). This process is claimed to drive after-listing share prices above their fundamental values, and is illegal under the laws against market manipulation and fraud. As a result, investment banks are reluctant to distribute information about their allocation practices, so investigating the alleged laddering and its implications has proven to be diffi cult. With a new and unique data set of 16,593 IPO allocations on the Oslo Stock Exchange (OSE), we confirm the SEC’ssuspicion that IPO allocations are dependent on after-listing trading. Allocations to after-listing purchasing investors has been combined with allocations to high stock-trading commissions generating investors that can take advantage of the IPO laddering, thereby allowing investment banks to recapture some of the money left on the table in IPOs. Allocated IPO investors buy more shares after new listings because they are rewarded for doing so with more IPO allocations.

8 2.1 Introduction On December 6, 2000 the Wall Street Journal (WSJ) reported that the SEC and the NASD were investigating some of the major investment banks for tying IPO allocations to after-listing purchases. An investment banker interviewed for the article admits that IPO allocations to investors with after-listing interest could occur, but explains that after-listing interest is a signal that the investor is of the buy-and-hold type. Since banks strive to allocate shares to buy-and-hold investors to create price stability, after-listing purchases are related to IPO allocations. An investor confirms that expressing an interest in after-listing purchases is one way of obtaining more IPO allocations. Three U.S. investment banks have been sued by the SEC over allegations of IPO laddering after the WSJ article, though all three later settled (without admitting guilt).2 The allegations made by the SEC are that the banks promised investors that they would receive an increased allocation in current hot IPOs if they bought additional shares after the listing of the same IPOs.3 The banks, allegedly asked IPO applicants if they would be interested in buying more shares after the listings and at what price and quantity. Since IPO laddering is illegal, there are no formal records of tying IPO allocations to after-listing trading, as agreements are likely to be made over the phone or in person rather than in a written agreement.4 It is, however, possible to see if there is a positive and consistent relationship between IPO allocations and after-listing trading by investors. Such a relationship would strongly indicate that IPO allocations are tied to after-listing buy trades, although this data is very hard to obtain in the U.S. (even for the SEC and NASD). Using data from the Oslo Stock Exchange (OSE), we are able to observe the after-listing trading of investors that were allocated shares in IPOs. The data consists of 16,593 IPO allocations with stock-trading commissions and after-listing trading on the OSE in the period from 1993 to 2007. Stock ownership by investor ID is observed for all companies throughout the listing process, and is used to calculate actual IPO allocations. It is, from this data that the relationship between IPO allocations, after-listing purchases, commissions and future IPO allocations is investigated. The main contribution of this paper is that we show a strong and robust relationship between the number of shares that are purchased after new listings and IPO allocations by laddering investors. This is consistent with the SEC’s suspicion that IPO shares are allocated to investors that buy shares dictated by the investment bank. We define laddering as allocated IPO investors that continue to buy shares right after the listing before they sell all shares within six months of the listing date. This sales requirement is included to remove rationed investors that buy shares to reach optimal holding levels after the listing. We also show that IPO laddering benefits both investors and investment banks and that the specified trading can not be attributed to other explanations such as share rationing. In the 50% IPOs with the highest laddering there is an average aggregate IPO allocation to laddering investors of 4%. On average these investors buy 6% more of the aggregate IPO shares after the listing, and then sell on average 10% of the aggregate

2 See the litigation releases made by the SEC at http://www.sec.gov/litigation/litreleases/lr18385.htm, http://www.sec.gov/litigation/litreleases/lr19050.htm, and http://www.sec.gov/litigation/litreleases/lr19051.htm. 3 There are many news articles and web pages that cover laddering and the laddering cases in the U.S. For excellent overviews please see Deneen and Hooghuis (2001), Aggarwal et al. (2006) and the IPO securities litigation websites at http://www.iposecuritieslitigation.com/, http://www.dandodiary.com/articles/ipo-laddering-cases/ and the articles by Susan Pulliam and Randall Smith, the journalists that ﬁrst published the laddering scandal in the Wall Street Journal series in 2000. http://www.pbs.org/wgbh/pages/frontline/shows/dotcon/interviews/pulliam-smith.html 4 In both Norway and the U.S. IPO laddering is illegal under the law against market manipulation.

9 IPO shares shortly after the listing. As a consequence of this, we are not able to reject that IPO allocations are tied to after-listing purchases of IPO shares. The SEC is investigating IPO laddering because laddering falsely increases the price and demand of specific shares (price manipulation). In addition to being abusive and discriminatory, IPO laddering is undesirable because it increases adverse selection problems (by deterring non-laddering investors from applying for IPO shares).5 Investment banks use IPO laddering because this practice will boost share prices after the listings. IPO shares that will go up in price for sure can also be allocated to bank clients that provide high levels of stock-trading commissions, thereby ensuring a future relationship between banks and investors that generate high levels of income for the banks. We show that investment banks and laddering inventors earn money on IPO laddering, while most companies with high levels of IPO laddering fall in price in the first six months after the listing (8 out of 11). IPOs generally have high first day returns (on average 8% in Norway in the sample period) and IPO shares are therefore very popular investments. Most IPOs are many times oversubscribed and few investors are allowed to buy IPO shares. Investment banks are reluctant to distribute information about their allocation practices, and the continued investigation by the SEC and the NASD on investment bank allocation practices has not made data collection any easier. Ritter (2003) and Jenkinson and Jones (2004) argue that there are three main views on how IPOs are allocated. First, the academic view based on Benveniste and Spindt (1989) is that investors obtain IPO allocations in return for revealing their true valuations of the IPO shares. These investors help to price the issue. Second, the pitchbook view argues that IPO shares are allocated to buy-and-hold investors, and long-term buy-and-hold investors will create price stability. Finally, the rent seeking view argues that IPOs are allocated in return for kickbacks. The types of rent seeking that have been under SEC investigation are to condition IPO allocations on generated stock-trading commissions, future corporate business (IPO spinning) or after-listing purchases of IPO shares (IPO laddering), see Liu and Ritter (2010). IPOs can also be intentionally underpriced in exchange for future analyst coverage (analyst conflict of interest). There are many articles that have studied both the academic and pitchbook view, but a lack of data has limited the number of articles which have studied the rent seeking view.6 Cliff and Denis (2004) show that IPO underpricing is related to after-listing analyst coverage, Liu and Ritter (2010) reveal that IPOs are allocated in return for IPO spinning and Fjesme, Michaely and Norli (2011) document that IPOs are allocated in return for stock-trading commissions. No empirical papers have been able to establish a relationship between IPO allocations and after-listing purchases of IPO shares (IPO laddering). Hao (2007) identifies the incentives to engage in IPO laddering and the implications of IPO laddering theoretically. Griffi n, Harris and Topaloglu (2007) show empirically that it is likely that IPO laddering is used by studying aggregate after-listing trading at the brokerage house level. Griffi n et al. (2007) find that after-listing buy trades primarily go through lead managers, whereas after-listing sell trades go through other managers in the weeks after new listings. This is consistent with IPO laddering because laddering investors will place their orders through the lead manager as evidence that the trades have been made. Previous research has not been able to study the relationship

5 Laddering is not new. The SEC sent out warnings that laddering was illegal in 1961, 1984 and 2000 (Griﬃ n et al., 2007). 6 See, amongst others, Jenkinson and Jones (2004), Ritter (2003) and Fjesme, Michaely and Norli (2011) for papers that summarizes studies on IPO allocations.

10 between IPO allocations and after-listing trading of the IPO shares at the investor level due to data limitations.7 The main research question addressed in this paper is whether investors are able to increase allocations in IPOs by committing to buy more shares after the listing of the same IPOs. We also investigate whether future IPO allocations are tied to after-listing purchases in past IPOs. The rest of the paper is organized as follows: Section 2.2 describes related literature. Section 2.3 describes predictions and testable implications. Section 2.4 describes the IPO process and the factors that create the incentives to engage in IPO laddering. Section 2.5 describes the data set. Section 2.6 describes the empirical results, and Section 2.7 concludes.

2.2 Related literature The two main theoretical papers that model IPO laddering are Hao (2007) and Aggarwal et al. (2006). Hao (2007) first show the factors that create the incentives to engage in IPO laddering. Then, the effects of IPO laddering on companies are identified. Hao (2007) argue that IPO laddering can benefit the underwriter from two sources. First, IPO laddering can boost the after-listing market price. This will reduce the underwriters expected cost of price support after the listing. From this it is expected that IPO laddering will be stronger when there is a positive drift in the after-listing share price. Second, IPO laddering can benefit the underwriter through rent seeking. If some allocated investors pay a part of their profit from IPO allocations back to the underwriter through stock-trading commission payments, then a part of the laddering generated profits will go back to the underwriter. Hao (2007) argue that when the underwriter share in on the profit from the underpricing, then laddering is stronger when the realized percentage underpricing is higher. Hao (2007) also show that expected underpricing increases IPO laddering. From this it is expected that laddering will be stronger when there is a positive underpricing. Hao (2007) also predicts that laddering is positively related to IPO allocations to high stock-trading commission generating investors. IPO laddering will inflate prices after the listing, so investment banks use laddering to make share prices go up after the listing (more than they otherwise would have). Shares that go up in price for sure are then be allocated to clients that generate high stock-trading commissions. Hao (2007) finally predicts that laddering will increase the IPO offer price, the first day closing price, the money left on the table and the long-run underperformance of the newly listed companies. Aggarwal et al. (2006) predict that IPO laddering increases underpricing, turnover and long-run underperformance of the newly listed companies. These are all effects of an increased demand of the IPO shares right after the listing that will fall in the long-run. There are three main empirical papers that provide indirect evidence of the existence of IPO laddering. Griffi n et al. (2007) look at investors who buy shares through lead and other underwriters in the three weeks after the listing of 1,294 Nasdaq IPOs in the period 1997 to 2002. As opposed to this study, they examine aggregate trading at the brokerage house level. They argue that the after-listing buy trades through the lead

7 Griffi n, Harris and Topaloglu (2007) find that it is very likely that investment banks tie IPO allocations to after-listing purchases. The major difference is that Griffi n et al. (2007) study the after-listing trading through co and lead managers at the brokerage house level, and we study actual IPO allocations and after-listing trading on the investor level. Griffi n et al. (2007) show that it is likely that laddering is being used by investigating through what manager after-listing buy orders are placed, and we show that after-listing buy orders are related to current and future IPO allocations by investors.

11 manager (main underwriter) in the weeks after the listings are likely to be part of laddering agreements, while buy trades through other managers (co-underwriters that help to spread the issue) in the same period are likely to not be part of the agreements. The paper ﬁnds that it is likely that IPO allocations are tied to after-listing purchases because there are unproportional high levels of buy trades through lead managers after new listings. Aggarwal et al. (2006) study IPOs that have been sued on laddering allegations to test the implications of laddering. The data includes 33 IPOs sued by the SEC, 140 class action law suits and 735 non-laddering IPOs on Nasdaq, NYSE and AMEX in the period 1998 to 2000. The main ﬁndings are that IPO laddering leads to underpricing and long- run underperformance. Ellis (2006) investigates the trading volume in IPO shares after the listing for 311 Nasdaq IPOs in the period 1996 to 1997. She shows that investor buy trades through the lead underwriter account for 22% of trading volume after IPOs, and this is consistent with laddering being used.8

2.3 Predictions and testable implications An IPO investor is rationed when the number of shares sought in the IPO is larger than the allocation. Rationing will lead to a smaller IPO allocation than the applied for shares for most investors. Rationed investors may buy more shares after the listing to get to the desired holding level. This has similar implications as IPO laddering. Griﬃ n et al. (2007) argue that investment banks may strategically allocate toe-holds to investors that the bank knows have higher optimal holding levels (share rationing)9. The bank does this in hopes that the investor will buy more shares after the listing to reach the optimal holding level. It is expected that most optimal holding investors will reach their determined holding level and then hold this in the longer run. Laddering investors, on the other hand, buy shares right after the listing to fulﬁll an obligation.10 Many laddering

8 There are also three other types of IPO rent seeking that have led to investigations and subsequent settlements with the SEC or the NASD (Liu and Ritter, 2010). IPO allocations can be dependent on future corporate business (IPO spinning), stock-trading commissions or companies can agree to underprice IPOs in exchange for after-listing company coverage from a star analyst provided by the investment bank (analyst conflict of interest). All of these allocation practices have been investigated in empirical papers. Liu and Ritter (2010) investigate 56 U.S. IPOs in the period 1996 to 2000 and show that IPO shares are allocated to corporate executives in return for future corporate business (IPO spinning). Cliff and Denis (2004) show that IPO underpricing is positively related to the after-listing coverage by the lead underwriter and an all star analyst (analyst conflict of interest). Nimalendran, Ritter and Zhang (2007), Reuter (2006) and Fjesme, Michaely and Norli (2011) show that IPO allocations are related to stock- trading commissions. 9 The SEC makes a big point about selling shares early in their cases. It is claimed that banks frequently allocated shares to investors that had no plans of holding the shares in the long run. Apparently, the banks asked the investors if they would agree to buy more shares after the listing and not if the investors were planning to hold the shares in the long run. If the reason for the after-listing purchases is to increase allocations, then this is laddering.

10 Allocated IPO investors that buy more shares after new listings can be explained by either IPO laddering or by IPO share rationing. Most of the IPO first day return takes place between the offer price and the first day opening price (not between the first day opening and the first day close). This means that any additional purchased shares have an expected return commensurate with risk and nothing more. It is therefore expected that investors that buy more shares after the listing do so because they want to hold more of the specific stock in their portfolio. If there is laddering, there should then be a stronger relation between after-listing purchases and allocations for short term investors. Short term investors are more likely to be laddering investors than long term investors.

12 investors will therefore sell their shares when the agreement is completed. The argument is not that laddering investors will always liquidate their holdings early. The argument is that investors that buy more shares because of optimal holding are more likely to hold their shares in the long-run. Some laddering investors are likely to hold their shares in the long-run as well, but some laddering investors will also liquidate their shares early because they have no interest in holding the shares. It is important to note that the intention of the after-listing buyer to buy-and-hold does not remove the possibility of IPO laddering (Griﬃ n et al., 2007).11 Optimal holding is also not a very good explanation for the observed after-listing buying in Norway. Investment banks rank investors on A, B and C lists before the IPO allocations.12 We do not know how investors are placed on the lists, but we believe that it is related to the investors’past trading characteristics. Investors on the A list are likely to be rationed less than investors on the B list, and investors on the B list are likely to be rationed less than investors on the C list. It is therefore expected that IPO applicants on the A list are awarded a big allocation and will buy few shares after the listing. Investors on the C list will be allocated few shares and will therefore buy many shares to reach their optimal holding level. This will create a negative correlation between the number of shares allocated and the number of shares purchased after the listing for these investors.

2.3.1 The IPO laddering hypothesis Hao (2007) argue that there are two reasons why underwriters use laddering. First, Hao (2007) argue that banks use laddering to boost prices after the listing. Boosted prices are good for investment banks because the expected price support cost is then reduced. IPOs with boosted after-listing prices will also be viewed as more successful. Second, Hao (2007) argue that when the underwriter share in on the profit from the underpricing, laddering is stronger when the realized percentage underpricing is higher. Hao (2007) also show that expected underpricing increases IPO laddering. (It is likely that the expected underpricing is highly related to the realized underpricing). If there is IPO laddering, it is expected that the relationship between allocations and after-listing purchases is stronger when the realized underpricing is higher. From Griffi n et al. (2007) and the first argument in Hao (2007) we expect that laddering is more likely when there is a positive drift in the share price after the listing (boosted price) and after-listing investors sell their shares soon after the listing date. This is formalized in H0.1. From Griffi n et al. (2007) and the second argument in Hao (2007) we expect that laddering is more likely when there is a positive underpricing and after-listing investors sell their shares soon after the listing. This is formalized in H0.2. If the relationship between IPO allocations and after-listing purchases is explained by share rationing, there is no reason why the relation should be stronger in IPOs where investors sell their shares soon after the listing, the price increase in the first week after the listing and the IPO is underpriced. This is formalized in HA.

11 Griffi n et al. (2007) test between IPO laddering and optimal holding by studying how the aggregate institutional holding percentage evolves from the listing date to the first quarter and the first year after the listing. They argue that laddering investors are mainly institutional, so the aggregate institutional holding percentage should go down in companies with IPO laddering - since laddering investors will reduce their holding percentage and optimal holding investors will not. In the Norwegian data we observe that the investors are allocated IPO shares buy more IPO shares after the listing and then sell shares soon after the listing. It is more likely that investors that follow this three stage IPO share investment process are laddering investors than optimal holding investors. 12 Information about allocation practices are obtained from meetings with former investment bankers in Norway.

13 H0 and HA are tested by regression equation (1).13 If the relationship between allocations and after-listing shares is signiﬁcantly stronger for (After-listing shares/shares issued)%i * D1 * D2 * D3 than for (After-listing shares/shares issued)%i, then we are not able to reject H0. This will, however, reject HA.

H0.1: The relationship between allocations and after-listing purchases is stronger when investors sell all shares within six months after the listing and the price after one week exceed the ﬁrst day closing price.

H0.2: The relationship between allocations and after-listing purchases is stronger when investors sell all shares within six months after the listing and the ﬁrst day closing price exceeds the oﬀer price. HA: The relationship between IPO allocations and after-listing purchases is the same for all investors.

(1) (Allocated shares/shares issued )%i= α+ β1(After-listing shares/shares issued)%i + β2(After-listing shares/shares issued)%i * D1 * D2 * D3 + β[Control variables] + i

2.3.2 Other testable implications of IPO laddering There are two other testable implications of IPO laddering besides that relation between after-listing purchases and IPO allocations. First, it can be tested if IPO laddering is beneficial for investors. In hot IPOs it is expected that investors that buy more shares after the listing will earn money because they are allocated an increased portion of hot shares. It is possible that the investors either lose or earn money on the additional shares purchased after the listing (this is uncertain and can go both ways according to an e-mail by Goldman Sachs referred to in the SEC release), but it is expected that buying more shares should be profitable overall. Money earned on the hot IPO allocations should outweigh any loss on the additional shares. This is tested by investigating if laddering investors earn money overall. In cold IPOs it is expected that the investors earn money on future IPO allocations. Although investors are not enthusiastic about cold IPOs it is expected that investors will follow through with the laddering to not be excluded from future IPOs, see Griffi n et al. (2007). This is tested by regressing past laddering on future IPO allocations. Second, it can be tested if IPO laddering is beneficial for investment banks. Investment banks tie allocations to after-listing purchases partly to earn money on stock-trading commissions. Laddering investors buy more shares after new listings so total IPOs with laddering increase more in price than IPOs with no laddering. Investment banks can then charge more stock-trading commissions for IPO allocations with laddering, see Hao (2007). If this is the case, there will be a relation between stock-trading commission generated before IPOs, by non-laddering allocated investors, and the aggregate after- listing purchases made by laddering investors. This is tested by regressing the aggregated

13 D1: A dummy that takes the value of one if the investor have sold all allocated and after-listing shares within six months of the listing date. D2: A dummy variable that takes the value of one if there is a positive drift in the share price in the week following the listing (from the ﬁrst day closing price to the ﬁrst week closing price). D3: A dummy variable that takes the value of one if the IPO have a positive underpricing.

14 IPO after-listing purchases made by the laddering investors on the average commission generated per share before the IPO (by the non-laddering allocated investors).

2.4 The listing process and the incentives to engage in IPO laddering The OSE requires that companies have suffi cient levels of equity to survive for 12 months without a positive cash flow after a listing. The OSE also requires that public companies must have a minimum number of owners before they can list (500 for the main list).14 This means that most companies need to issue equity before they are able to list publicly. Table 1 gives the annual distribution of IPOs on the OSE in the sample period. Most companies are assisted by an investment bank in their equity issuance and in the listing process. The investment bank makes a list with proposed IPO allocations that is given to the board of the issuing company for approval. Anecdotal evidence suggests that this list typically is approved without adjustments. Investment banks and investors have different reasons for why they participate in IPO laddering. Regulators investigate IPO laddering because it is manipulative.15

2.4.1 Why investment banks use IPO laddering IPO laddering can be advantageous for investment banks in both hot and cold IPOs. There are two main reasons why investment banks use IPO laddering in hot IPOs. Firstly, investment banks can earn money on combining allocations to investors that generate high stock-trading commission and to laddering investors. IPO laddering will boost prices after the listing. This will give the companies attention as more successful IPOs (Hao, 2007; Aggarwal et al., 2006; Griffi n et al., 2007). Secondly, IPO laddering will increase underpricing. The IPO allocations will then be valued higher by investors that are willing to pay stock-trading commissions to obtain allocations (Hao, 2007; Aggarwal et al., 2006). In related papers, Reuter (2006), Nimalendran et al. (2006) and Fjesme, Michaely and Norli (2011) show that stock-trading commissions are related to IPO allocations. Laddering can also be beneficial for investment banks in cold IPOs. IPO laddering will reduce the after-listing price uncertainty in cold IPOs. This is good for investment banks because IPOs that fall in price may cause reputation damage (and price support if used without over allotment options is potentially expensive) (Hao, 2007; Aggarwal et al., 2006; Griffi n et al., 2007). Investment banks use IPO laddering to earn more money on stock-trading commissions, to increase the likelihood of successful IPOs and to reduce the risk of after-listing price falls.16 The after-listing purchases will also increase direct commission from the extra trades. According to Griffi n et al. (2007), it is uncertain whether laddering is more beneficial for the investment banks in hot or cold IPOs.17

14 The information about the listing process is obtained from the seminar “The road to the listing” November 3, 2009 by Deloitte Public Accountants and the Oslo Stock Exchange and from meetings with former investment bankers in Norway. 15 Figure 1 describes the incentives to engage in IPO laddering for the different market participants. 16 It is probably more common that bidders will offer laddering than that banks require laddering. Investors will offer laddering if they believe that this will increase allocations and lead to future allocations. Hao (2007) argues that it does not matter for the effect of laddering if it is bidder or investment bank initiated. 17 Laddering in cold IPOs creates a relation between after-listing purchases and future allocations (not necessarily between allocations and after-listing purchases). Laddering in hot IPOs will create a relation between allocations and after-listing purchases in specific IPOs.

15 2.4.2 Why laddering investors agree to buy more shares Investors agree to buy more shares after cold IPOs to get future allocations in hot IPOs. Investors are not likely to be enthusiastic about laddering in cold IPOs, but investors who want continued access to future hot IPO allocations are likely to follow through with the agreements (Griffi n et al., 2007). Investors accept laddering in hot IPOs in order to get more allocations in the specific IPOs. Laddering investors may either earn or lose money on the extra shares purchased after the listing, but it is expected that the return of the hot allocated shares will outweigh any loss on the additional shares.18 Investment banks do not require laddering by all investors. Griffi n et al. (2007) argue that laddering is pre arranged buying support by large institutional clients. It is easier to control that the shares are purchased when there are only a few investors involved.

2.4.3 Why IPO laddering is a problem The reason why the SEC is investigating IPO laddering is because laddering violates both anti-price-manipulation and anti-fraud regulations. Laddering will falsely increase price and demand in specific shares. Investors who are not aware of the laddering will buy shares on false market demand information. Regulators (the SEC) try to ensure that the IPO allocation process is fair and open to all investors. Any abusive allocation practices will not be tolerated. Laddering is a problem because it is discriminatory against investors that are not willing to engage in price manipulation to receive IPO shares. In a fair IPO with high demand the offer price will increase and more money will go to the issuing company. In an IPO with laddering the price will go up after the listing and more money will go to the allocated investors. Other investors can also lose money on IPO laddering. The investors that are allocated less (or no) IPO shares because the laddering investors are allocated more hot shares are missing out on good investment opportunities. Non-allocated investors that buy shares after the listing lose money if the laddering investors sell their shares so that prices fall after the listing. IPO laddering will also increase adverse selection problems. When investors know that it is possible to buy allocations with after-listing trading, it is not likely that investors will participate in IPOs. Investors that do not provide any form of kickback will not want to participate in IPOs because they expect shares to be overpriced whenever they are offered allocations. The allegation made by the SEC is that investment banks have promised investors that they will get favorable IPO allocations if they buy additional shares after the listing of the same IPO.19 The banks have, allegedly, asked IPO share applicants if they are interested in buying more shares after the listings and at what prices and quantities. The banks have also allocated shares to investors with after-listing interest -investors the banks

18 See the litigation releases made by the SEC at http://www.sec.gov/litigation/litreleases/lr18385.htm, http://www.sec.gov/litigation/litreleases/lr19050.htm, and http://www.sec.gov/litigation/litreleases/lr19051.htm. 19 Three U.S. Investment banks that have been sued and later settled with the SEC on IPO laddering allegations. None of the banks have admitted to the laddering charges, but all banks have agreed to pay penalties of $40 million (Morgan Stanley), $40 million (Goldman Sachs) and $25 million (J.P. Morgan). The charge by the SEC is that the banks have violated Rule 101 of Regulation M under the Securities and Exchange Act of 1934. This rule is, among other things, in place to prohibit underwriters in a restricted period, prior to their completion of the distribution of the IPO shares, from bidding for or attempting to induce any person to bid for or purchase any offered security in the aftermarket. Regulation M is designed to prohibit activities that can artificially influence the market and the perceived demand of the IPO shares.

16 knew were likely to sell their shares soon after the listing (laddering investors). The banks have made follow up calls to investors that indicated after-listing interest to make sure the purchases are made. Arguably, the only reason investors have provided after-listing interest is because the investors understand that this will help them get favorable IPO allocations. Banks and investors have agreed that investors will buy after-listing shares proportional to the allocations they receive.20

2.5 Data description There are 403 new listings on the OSE in the period January 1993 to September 2007 (210 of the 403 companies listed through IPOs)21. New listings are identified from the annual statistics published by the OSE. Allocation dates are collected from the IPO listing prospectuses. One listing requirement on the OSE is that all shareholders must be registered in the Norwegian Central Depository (VPS) before the listing. The number of shares owned by each investor must be given to the VPS before any company can list publicly. This database is 100% accurate, as it is not possible to list otherwise. The VPS database includes month end ownership by all shareholders in all companies that are publicly listed or intend to list publicly. Some companies list in the VPS database years before the listing. Other companies list in the VPS as part of the listing process. See Figure 3 for a detailed description of the timeline in the listing process. IPO allocations are obtained from the VPS database by taking the difference in company ownership before and after IPO allocation dates. We only investigate IPO allocations to new shareholders. More allocations to existing shareholders, if any, are not included in the analysis. All companies list in the VPS, sell shares in the IPO and list on the OSE. There are three dates that are important in the listing process to determine IPO allocations,: -when companies list in the VPS ownership database, when companies distribute shares in the IPO and when companies list on the OSE. All three dates influence data on IPO allocations. Companies do this process in different orders, and this leads to different levels of the obtained IPO allocations. There are 15 savings banks (PCC list) out of the 210 IPOs on the OSE in the sample periods. In total, 14 and seven of these savings banks are in the 185 IPOs with allocation data and in 30 exact sample respectively. These banks are owned by the bank guarantee fund before they are publicly listed. All results remain unchanged if the banks are included in the analysis or not. These savings banks are removed in the main analysis because it can be argued that these banks are causing the results. The savings banks does not have previous owners before the listing, so it can be argued that different investors try to gain control over these companies after the listing. Investors who are not able to get control will eventually sell their shares. This will create similar findings, as the laddering hypothesis, for these savings banks.

20 In addition to these allegations, the NASD claims that J.P. Morgan tied cold IPO allocations to hot IPO allocations and that J.P. Morgan allocated hot IPO shares to investors in the return for accepting cold IPO allocations. This is also part of the J.P. Morgan settlement. Hao (2007) explains that IPO order books often have investors that are marked with the number of shares that will be purchased after the listing. 21 In total 14 savings banks listed on the PCC list of the OSE are removed from the analysis. Most of the PCC companies are listed by the Norwegian bank guarantee fund. When including the PCC companies the ﬁndings remain unchanged.

17 2.5.1 The IPO sample When the listing in the VPS database, the IPO allocation and the listing on the OSE are in separate calendar months, we are able to calculate exact IPO allocations (the ownership data is in monthly observations). Group one companies list in the VPS in good time before the IPO. These companies also list on the OSE in a separate calendar month from the IPO (for most companies, the IPO is in the calendar month right before the listing month). For group one companies the IPO allocations are completely accurate. There are 16,593 IPO allocations in group one companies (23 IPOs). After-listing purchases are the increase in the number of shares by the allocated investors from the IPO allocation to the end of the listing month (and to the end of the month after the listing).22

2.5.2 The remaining IPOs The data set also includes 158,789 IPO allocations in 148 IPOs that are used in robustness tests.23 The allocations in these IPOs include either some existing owners or some after- listing trading. Group two companies list in the VPS in good time before the IPO, but they list on the OSE in the same calendar month as the IPO allocation month. These companies have allocations that include the actual IPO allocations and some after-listing trading. These IPO allocations includes from one to 30 days of after-listing trading. The companies in group two are used to test the relationship between past and future after-listing IPO holdings.

2.5.3 Aggregate laddering There are 317 investors who sell all allocated and all after-listing shares within six months of the listing date in IPOs with a positive underpricing (in the 50% IPOs with the highest laddering). The aggregate allocations to these investors is 4% of the IPO shares. They buy in aggregate 6% of the IPO shares after the listing. Within six months they have sold all IPO shares (in aggregate 10% of the IPO shares). There are 174 investors who sell all allocated and all after-listing shares within six months of the listing date in IPOs that appreciate in price in the week after the listing (in the 50% IPOs with the highest laddering). The aggregate allocations to these investors is 5% of the IPO shares. They buy in aggregate 8% of the IPO shares after the listing. Within six months they have sold all IPO shares (in aggregate 13% of the IPO shares).

22 Shares sold over the counter (OTC trading) in the period between the allocation day and the end of the allocation month will not be detected in the data. Investors that buy shares in the OTC market between the allocation day and the end of the allocation month will be treated as allocated investors. OTC trading is, however, expected to be a very small issue. It is unlikely that many investors that have been allocated IPO shares will sell these shares in the weeks before the listing. The average number of days between payment date in the IPO (when shares are transferred) and the listing date is just below two weeks 23 The reason it is 148 IPOs and not 172 (195-23=172) is because in 15 IPOs it has not been possible to calculate IPO allocations from the ownership data. These companies are listed in the VPS in the same month as the listing month. These companies are therefore removed from the sample. In 6 IPOs it has not been possible to locate the pricing information. These IPOs are therefore not included in the analysis. There are three privatizations in the period that are removed.

18 2.5.4 Variable explanations IPO level characteristics are given in Table 2. Market value is the total market value (in USD) at the listing date of the IPO company. This is calculated as the number of outstanding shares times the first day closing price. Book/Market is the book to market ratio of the IPO company at the listing date. This is calculated as the book value of equity, after the IPO, divided by the market value. Offer price is the IPO offer price (in USD) reported in the listing prospectus or in the newspapers. VC dummy is a dummy variable that takes the value of one for companies with venture capital backing. High- tech dummy is a dummy variable that takes the value of one for IT -companies. The IPO company variables are used to control that the results are not driven by company specific characteristics. Market value and the book to market ratio are included in the regressions to make sure that company size is not driving the results. Offer price is included to make sure that it is not very high or low priced IPOs that drive the results. The VC dummy and the high-tech dummy are included to make sure that the results are not driven by venture capital backing or high technology companies. All regressions include IPO and year fixed effects. These are dummy variables that take the value of one for each of the companies and sample years. Investor characteristics, for the investors on the OSE in the period 1993 to 2007, are described in Table 3. (After-listing shares/shares issued) % is the additional shares purchased after the listing divided by the total number of shares issued in the IPO.24 The after-listing shares are calculated as the share increase from the IPO allocation to the end of the listing month for the 23 sample IPOs. (We also include the share increase to the end of the month after the listing because some companies list late in the month and IPO laddering may go on as long as three weeks after the listing, see Griffi n et al., 2007). For the remaining IPOs the share increase is measured from the end of the listing month to the end of the month after the listing. This is likely to underestimate the after-listing purchases in the IPOs used for robustness. D1 is a dummy variable that takes the value of one if the investor have sold all allocated and after-listing shares within six months of the listing date. D2 is a dummy variable that takes the value of one if there is a positive drift in the share price in the week following the listing (from the first day closing price to the first week closing price). D3 is a dummy variable that takes the value of one if the IPO have a positive underpricing. (Allocated shares/shares issued) % is allocated shares to each investor divided by the total number of shares issued in the IPO.25 This is the percentage allocation of shares given to each investor in each IPO. Previous laddering is the accumulated number of times an investor has laddered divided by the accumulated number of times the investor has participated in IPOs. This is a measure of how frequently

24 The number of shares sold in the IPO is the number of actual shares sold to new shareholders from the VPS database. In the listing prospectuses the number of shares sold is often listed as a range. E.g. in the Aqua Bio IPO the listing prospectus says that the number of shares sold will be between 1.2 million and 4 million shares. It is also uncertain if Over Allotment Options (OAO) is used or not. This may increase the number of shares sold from the listing prospectus to actual shares sold up to 20%. E.g. in the Nutri Pharma IPO the minimum number of shares sold is 10 million. The lead manager is given 2 million extra shares in an OAO. From the prospectus it is impossible to know the exact number of shares that will actually be sold. This number is observable in the VPS database. 25 (Allocated shares/shares issued) % is trimmed at 1% at the total 171 IPO level to remove the highest IPO allocations. These allocations are not likely to be made to investors based on trading characteristics. This is included to be consistent with Fjesme, Michaely and Norli (2011). This trimming has no inﬂuence on the ﬁndings in this article.

19 an investor engages in laddering, relative to its total participations in IPOs.26 Commission is the accumulated commission (in USD) generated by each investor in the two years before the IPO allocation dates.27 Commission is calculated as the monthly portfolio turnover times share prices and a fixed percentage commission rate (0.075%). The 0.075% commission rate is the average used by 15 Norwegian brokerage houses. Commission is calculated as buy generated commission only. Generated commission below the minimum rate is replaced by the fixed minimum fee for one transaction ($15). Portfolio value is the total investor portfolio value (in million USD) for each allocated investor at 31.12.xx in the year before the IPO allocation date. This is calculated as the shares held at 31.12.xx times the appropriate share prices. Financial institution dummy is a dummy variable that takes the value of one for investors that are either Norwegian or foreign financial institutions. Previous IPOs is the accumulated previous IPO participations by the investors divided by the accumulated number of IPOs in the sample.28 This is used to measure how many IPOs, out of all possible in the sample, each investor has participated in. Pre- vious buy-and-hold is the accumulated previous number of times the allocated investor has been a buy-and-hold investor divided by all previous IPO participations. This is the number of times, out of all previous IPO participations, the investor has held some of the IPO allocated shares for more than six months after the listing. Previous flipping is the accumulated number of times the investor has flipped previous IPOs divided by all previous IPO participations. Flipping is when all shares are sold within one month after a listing. This is the number of times, out of all previous IPO participations, the investor has held all IPO allocated shares for less than one month. The previous trading variables are used to control that the results are not driven by investor size, trading activity or holding periods. Other control variables includes the Percentage change in pricing range that is the change from the midpoint in the pricing range to the offer price in book-building IPOs. This variable measures price information collected in the book-building period, see Ljungqvist and Wilhelm (2002). Number of sentiment investors is the number of allocated retail investors that buy less than 1,000 shares in the IPO. We use this as our sentiment measure as we believe that small retail investors are more sentiment driven in their IPO applica- tions as they spend less time on fundamental analysis, see Kumar and Lee (2006). Average commission per share is calculated as the total commission generated by non-after-listing purchasing investors in the 24 month period before the IPO divided by the number of shares allocated in the IPO. This is the average dollar generated commission per share be-

26 An investor that has participated in one IPO and bought more shares after the listing and then sold shares will take the value of 1 (1/1). An investor that has participated in two IPOs and bought and sold more shares after the listing in one of these IPOs will take the value of 0.5 (1/2). 27 Commissions are generated from monthly data and not daily data. 28 Many IPOs are underwritten by more than one investment bank. If there is more than one investment bank involved in the IPO, the bank that appears on the top left of the front page of the listing prospectus is assumed to be the lead investment bank. Carter and Manaster (1990) use the investment bank that appears top left on the tombstone as the lead investment bank. In most IPOs there are also co-managers that help with spreading the shares. Co-managers will allocate shares to their own clients. Investment banks can be co-managers in many IPOs, and this creates the situation where investors can be allocated shares as a reward in an IPO by another lead bank. There are also some mergers between investment banks in the period and this will also create the situation where award shares can come from other lead banks. Because of this, we investigate past trading behavior in all past IPOs in relation to current IPO allocations. We also study IPOs by one single bank separately. When this is done, we only investigate past trading in the IPOs where the one bank has been the lead.

20 fore the allocation (by non-laddering investors). Combined commission % is calculated as the commission generated by all the allocated investors in the 24 month period before each IPO divided by the accumulated commission generated by all the allocated IPO investors in the 24 month period before all IPOs. Average commission and Combined commission are used to measure how important stock-trading commission is for allocations in each specific IPO. These variables measure if there is a relationship between commission generated before an IPO (by the allocated investors) and the aggregated after-listing purchases of the IPO shares. We do not know the exact oversubscription numbers in each IPO. Normally, oversubscription numbers are used to define if IPOs are hot (popular/oversubscribed) or cold (less popular/undersubscribed). We proxy for hot/cold by a dummy that takes the value of zero if there is negative first day return (cold) and one otherwise (hot). We expect that underpriced IPOs are hot and non-underpriced IPOs are cold.

2.6 Empirical results From Table 4 it can be seen that there is a positive relationship between IPO allocations and after-listing purchases (regression 1). This relationship is significantly stronger for investors who sell their shares soon after the listing (regression 2). The relationship is also significantly stronger for investors that sell all shares soon after the listing in IPOs with a positive drift in the share price in the week after the listing (regression 3). This is consistent with H0.1. The relationship is also significantly stronger when investors sell shares soon after the listing and the IPO have a positive realized underpricing (regression 4). This is consistent with H0.2. The relationship between allocations and after-listing purchases is also significantly stronger for investors that sell all shares soon after the listing, in IPOs with a positive underpricing, and in IPOs with a positive drift in the share price after the listing (regression 5). The point estimate for the allocation and after-listing purchase relationship is typically two to five times as large for the cases where H0 specify that the relationship should be stronger. The relationship is also economically significant. The coeffi cient between allocation and after-listing purchases is about 0.25. This means that for each 1% of the issues that is allocated these investors buy 4% more after the listing, controlling for all other variables. The average number of shares purchased after the listing is close to 7,000 shares for the 427 laddering investors. This indicates that the allocation rule is that investors who commit to buy 7,000 shares after the listing are allocated close to 2,000 more shares in the IPOs. The results are robust to how many shares and how early the shares must be sold for investors to be regarded as laddering investors. The results remain unchanged when investors who have sold 50% of their shares within three months of the listing date are regarded as laddering investors. The relationship between IPO allocations and after- listing purchases is significantly stronger for investors that sell 50% of total shares within three months after the listing, in IPOs with a positive underpricing, and in IPOs with a positive drift in the share price after the listing than for other investors (regression 6). The relationship is also significantly stronger for investors that sell 50% of total shares within six months after the listing, in IPOs with a positive underpricing, and in IPOs with a positive drift in the share price after the listing (regression 7). This is consistent with H0.29 Most of the control variables are unrelated to the level of allocations. Generated 29 Both allocated shares and after-listing shares are scaled by the number of shares issued in the IPOs.

21 stock-trading commission is positively related to allocations. This indicates that laddering investors are active investors. To make sure that the results are not driven by the other allocations views suggested by Ritter (2003) and Jenkinson and Jones (2004) we control for these views in all regressions. To control for the pricing information view (the academic view) we include a dummy variable that takes the value of one for all professional investors (ﬁnancial institution dummy). If there is allocation to buy-and-hold type investors, there will be a relation between holding periods and IPO allocations (buy-and-hold view). This is controlled for by including the past IPO holding period of the allocated investors in all regressions (past buy-and-hold and past ﬂipping). Neither of these variables are consistently related to allocations. It is also possible that allocations are made to commission generating investors only (rent seeking view). This view is controlled for, and ruled out by including the portfolio value and the generated commission before the IPOs by the allocated investors in the regressions.30 In Table 5 the relation between past IPO laddering and future ownership of IPO shares is investigated more closely. If there is IPO laddering, it is expected that investors may be rewarded with allocations in future IPOs as well. Testing the relation between past laddering and future allocations is hard in the 23 IPO sample because there may be some time between each observed IPO. This is therefore tested on the full sample where IPO allocations include after-listing trading. Here we test whether investors that buy more (and then sell) shares after the listing of IPOs also hold shares after the listing of future IPOs. In Table 5 all 171 IPOs (with 175,382 IPO allocations) are investigated. Most of these IPOs are of group two allocations. This means that the IPO allocations may be overestimated and the after-listing purchases may be underestimated in these IPOs.31 Therefore, we are not studying allocations. Rather, this table investigates whether past after-listing buying leads to future after-listing holding of IPO shares. In Table 5 we regress after-listing holdings of IPO shares on the number of times in the past (out of all IPO participations) allocated investors have bought (and then sold) more shares after IPOs. There is a strong relation between past IPO laddering and shares held after future IPOs. This indicates that banks tie IPO allocations together. This indicates that IPO shares are also rewards for past laddering in IPOs.32 There is a consistent

There are very different numbers of shares sold in each IPO. Capital raised depends on both the number of shares and on the offer price in the IPO. The numbers we are interested in are therefore allocated shares and after-listing shares in percent of issued shares. This tests the relationship regardless of the number of shares issued in the IPO. We also regress allocated shares on after-listing shares directly without adjusting for issued shares in all regressions. This does not alter the findings. There are some changes to significance levels and adjusted R —squares, but the results remain the same (not reported). 30 We are not able to control for IPO spinning. IPO spinning is when IPO shares are allocated to company executives for future corporate business. Spinning will not generate the same implications as IPO laddering, so we argue that this is not a problem. 31 These shares are still purchased by the investors. Aftermarket purchases for group two IPO allocations are calculated as the share increase from the end of the listing month to the end of the month after the listing. This means that all of these investors have an increase in the IPO shares in this period. All of these investors are buying shares after the listing of IPOs. These investors also hold significantly more IPO shares in subsequent IPOs. Table 5 shows that investors who hold shares after the listing of IPOs, before they buy more shares in the following month, also hold more shares of future IPOs. This is consistent with the laddering story. We cannot show that IPO allocated investors who buy more shares after a listing are allocated more hot IPO shares, but we show that investors who buy more (and then sell) shares after the listing of an IPO have more IPO shares in their future portfolios. 32 Past aftermarket buying is less statistically and economically related to IPO allocations in the 20 IPOs by the least active investment banks (not reported). The tie-in agreement variables are highly

22 negative relationship between past buy-and-hold and IPO allocations. Investors are not allocated shares because they repeatedly hold their shares in the long-run. Investors are, however, punished for flipping shares in the past. Flipping investors are kept out of future hot IPOs. These findings also show that investment banks keep records of how investors trade in IPOs. The banks use these records in their future IPO allocations. This is consistent with the SEC releases where it is claimed that banks track investor trading and use this in their future allocation decisions. From Table 6 it can be seen that investors are able to earn a profit from IPO laddering. For allocated shares the monetary return is calculated as the number of allocated IPO shares times the first day and first month return. For shares purchased after the listing the monetary return is calculated as after-listing shares times the first month return. It is clear that the profit earned from hot IPO allocations outweighs any loss from the after-listing purchases. Table 6A show that the average return made by the 357 investors who ladder in IPOs with a positive realized underpricing have a positive return overall. This is also true for the 195 investors who ladder in IPOs with a positive drift after the listing (Table 6B). The 70 investors (427 laddering investors - 357 laddering investors in hot IPOs) that ladder in cold IPOs are earning a profit in their overall IPO participation. This indicates that these investors are rewarded in future IPOs for their cold IPO laddering (Table 6C). The 23 IPO sample is also split into high and low laddering IPOs based on the 427 investors who sell all shares within six months of the listing. Non-allocated investors that buy shares after the listing in the high laddering IPOs are losing money on average (Table 6D). This is not true in the IPOs with low laddering (Table 6E). Overall, this shows that IPO laddering is profitable for laddering investors. However, IPO laddering is very bad for non-allocated IPO investors that buy shares after the listing. From Table 7 it can be seen that there is a positive relationship between aggregate after-listing buying in each IPO (by the investors who sell all shares within six months of the listing) and the average commission generated by other allocated investors before the IPO. This is an important condition for IPO laddering to take place. A main reason why an investment bank would engage in IPO laddering is to increase revenue by sharing in on the money left on the table. Investment banks combine IPO allocations to laddering investors and stock-trading commission investors, and thus create a positive relation between after-listing buying and commissions generated by the allocated investors before the IPO. Laddering investors increase prices after the listing, and commission investors pay more stock-trading commission for shares that will increase in price for sure. The investment banks earns more money from stock-trading commissions in the IPOs where there are more shares purchased after the listing. The data is consistent with that investment banks combine IPO allocations to high stock-trading commission investors and laddering investors. related to IPO allocations in the IPOs by the most active investment bank (not reported). The results indicate that active IPO investment banks are able to use tie-in agreements. The reason why investors go through with the tie-in agreements, and buy more shares after the listing, is to avoid being blacklisted in future IPOs. An active investment bank will have a more reliable threat than less active banks. There is no relationship between IPO allocations and aftermarket purchases by Norwegian government investors. This is also as expected. The findings are consistent with Pulliam and Smith (2000), Ritter (2003), Aggarwal et al. (2006), Hao (2007) and Griffi n et al. (2007).

23 2.6.1 Optimal holdings We reject the hypotheses that the relation between IPO allocations and after-listing buying is driven by optimal holding of shares. There is a stronger relationship between IPO allocations and after-listing purchases for investors that sell all shares soon after the listing, in IPOs with a positive underpricing, and in IPOs with a positive drift in the share price after the listing. There is also no relationship between IPO allocations and past buy-and-hold. Investment banks do not allocate shares to investors because they are expected to be buy-and-hold based on past trading. Therefore, the after-listing purchases are not simply a result of investors trying to reach their optimal holding levels. We reject HA.

2.6.2 The eff ect of IPO laddering We find indications that laddering is affecting company long-run returns negatively after the listing (not reported).33 The 11 companies with high levels of IPO laddering have a negative price evolvement in the time after the listing on average. Non-allocated IPO investors who buy shares in this period are also losing money on average. This is consistent with both Hao (2007) and Aggarwal et al. (2006). When comparing long-run returns of IPOs with high laddering to a one for one matching listed firm, the underperformance results are very weak with zero or very low explanatory power. The matching firm tech- nique is also biased towards finding long-run underperformance, see Eckbo, Masulis and Norli (2008). We are not able to conclude that high levels of laddering leads to low long- run performance, but the results indicate that laddering is negatively related to long-run performance.

2.6.3 Robustness and aggregate IPO laddering The results are robust to including PCC savings banks and trimming IPO allocations at 0.1% instead of at 1%, see Table 8. The results are also robust to removing all company speciﬁc control variables (Table 9). From Table 10 it can also be seen that IPO laddering involves an economically signiﬁcant amount of the IPO shares. Laddering investors are on average allocated 4% of IPO shares before they buy 6% more shares after the listing. These investors then sell 10% of the IPO shares within six months of the listing date (in the 50% IPOs with the highest IPO laddering).

2.7 Conclusion There is a stronger relationship between IPO allocations and after-listing purchases when investors sell shares soon after the listing, the IPO have a positive realized underpricing and there is a positive drift in the share price after the listing. This ﬁnding is not consistent

33 Long run performance is calculated as the (IPO company holding period return / matching company holding period return) (Ritter, 1991). This long run return measure is regressed on the aggregate level of aftermarket share buying and a set of control variables. Companies are matched on market values and book to market ratios, see Eckbo and Norli (2005). All matching companies with a market value within 30% of each IPO company are grouped together. Only companies that have been listed for more than ﬁve years are included as matching companies. The company with the book to market ratio that is closest to the IPO company is used as the matching company. A delisted matching company is replaced by the company with the second closest book to market ratio for the remaining years etc.

24 with HA and this hypothesis is therefore rejected. We reject that the relationship between IPO allocations and after-listing purchases is driven by share rationing only. This finding is, however, consistent with H0. We are not able to reject that the relationship between IPO allocations and after-listing purchases is driven by IPO laddering. The evidence support IPO laddering. We find that laddering investors who buy more shares after the listing are also allocated more shares in IPOs. This controls for the stock-trading commissions generated by the investors, portfolio value, investor type, past trading characteristics and company specific variables. These investors also sell their shares shortly after the listing and earn a high profit from their IPO participation, which is consistent with IPO laddering. The investors that buy the most shares after the listing are also allocated the most shares. These investors are not expected to hold the shares based on past trading characteristics. There is also a positive relationship between the number of times investors have used laddering after the listing in previous IPOs and after-listing ownership of future IPO shares. There is no relationship between past buy-and-hold and future IPO share ownership, -further indicating that this is IPO laddering. Laddering gives more shares in specific IPOs and more shares in future IPOs. The aggregate laddering in IPOs is also positively related to the average commission generated by the allocated investors before the IPOs, thus demon- strating that there is more laddering when there are more shares allocated to investors that generate high levels of stock-trading commission. Investment banks seems to be able to earn money on IPO laddering by combining allocations to after-listing investors and high commission investors. The evidence is consistent with IPO laddering. We are not able to reject that IPO laddering is being used. There are many implications of this finding. The main practical implication is that investors who are not aware of IPO laddering lose money on trading in IPO shares in comparison to more informed investors. IPO laddering is also likely to increase adverse selection problems as many investors are likely to stay away from the IPO market when they know they must provide kickbacks to acquire the good allocations. In the U.S. there has been a large-scale investigation of IPO allocation practices, and this study shows that more countries should probably start their own investigations as well. A main theoretical implication of this finding is that IPO allocation practices should probably be explained more from a rent seeking perspective since most theoretical papers explain IPO allocations from a pricing information or buy-and-hold perspective. There are some limitations to this study. With regard to the generated stock-trading commission, we cannot see that commission is paid from the allocated investor to the investment bank, and can only observe that the commission has been generated. We also calculate commissions based on monthly data, and this is likely to underestimate commissions. The study does not conduct and in-depth investigation of long-run performance (as we only observe a limited number of companies), and we also do not know the oversubscription numbers of the IPOs. This is proxied for by using the actual first day return as the oversubscription hot/cold IPO dummy. Nevertheless, we do expect this dummy to be very accurate. In terms of future research, it would be very interesting to investigate a sample which included the actual IPO laddering agreements in writing.

25 References

[1] Aggarwal, Rajesh K., Amiyatosh K. Purnanandam and Guojun Wu, 2006, Under- writer Manipulation in Initial Public Offerings. Working paper. University of Virginia and University of Michigan. [2] Benveniste, Lawrence and Paul Spindt, 1989, How investment banks determine the offer price and allocation of new issues, Journal of Financial Economics 24, 343-362. [3] Carter, Richard B. and Steven Manaster, 1990, Initial Public Offerings and the underwriter reputation, Journal of Finance 45, 1045-1067. [4] Cliff, Michael T. and David J. Denis, 2004, Do Initial Public Offering Firms Purchase Analyst Coverage with Underpricing?, Journal of Finance 6, 2871-2901. [5] Deneen, M., and J. Hooghuis, 2001, Tidal Wave of IPO ‘Laddering Litigation’ Swamps D&O Market, The Plus Journal 9. [6] Eckbo, Espen B. and Øyvind Norli, 2005, Liquidity risk, leverage and long-run IPO returns, Journal of Corporate Finance 11, 1-35. [7] Eckbo, Espen B., Ronald W. Masulis and Øyvind Norli, 2000, Seasoned Public Of- ferings: resolution of the ’new issues puzzle’, Journal of Financial Economics 56, 251-291. [8] Ellis, Katrina, Roni Michaely and Maureen O’Hara, 2000, When the Underwriter Is the Market Maker: An Examination of Trading in the IPO Aftermarket, Journal of Finance 3, 1039-1074. [9] Ellis, Katrina, 2006, Who trades IPOs? A close look at the first days of trading, Journal of Financial Economics 79, 339-363. [10] Fjesme, Sturla Lyngnes, Roni Michaely and Øyvind Norli 2011, Using stock-trading commissions to Secure IPO allocations, Working paper, Norwegian Business School (BI). [11] Griffi n, John M., Jeffrey H. Harris and Selim Topaloglu, 2007, Why are IPO investors net buyers through lead underwriters?, Journal of Financial Economics 85, 518-55. [12] Hao, Qing (Grace), 2007, Laddering in initial public offerings, Journal of Financial Economics 85, 102-122. [13] Jenkinson, Tim, and Howard Jones, 2004, Bids and Allocations in European IPO Book-building, Journal of Finance 5, 2309-2338. [14] Kecskes, Ambrus, Roni Michaely and Kent Womack, 2010, What drives the Value of Analysts Recommendations: Earnings Estimates or Discount Rate Estimates?, Working paper Cornell University. [15] Kumar, Alok and Charles M.C. Lee, 2006, Retail Investor Sentiment and Return Comovements, Journal of Finance 5, 2451-2486. [16] Liu Xiaoding and Jay. R. Ritter, 2010, The Economic Consequences of IPO Spinning, Review of Financial Studies 5, 2024-2059.

26 [17] Ljungqvist, Alexander P. and William J. Wilhelm, 2002, IPO allocations: discriminatory or discretionary?, Journal of Financial Economics 65,167-201.

[18] Loughran, Tim and Jay R. Ritter, 1995, The New Issues Puzzle, Journal of Finance 1, 23-51.

[19] Loughran, Tim and Jay R. Ritter, 2004, Why has IPO underpricing changes over time?, Financial Management 33, 5-37.

[20] Megginson, William and Kathleen Weiss, 1991, Venture capitalists certiﬁcation in initial public oﬀerings, Journal of Finance 46, 879-904.

[21] Michaely, Roni and Wayne H. Shaw, 1995, The Choice of Going Public: Spin-oﬀs vs. Carve-Outs, Financial Management 24, 5-21.

[22] Nimalendran, M., Jay Ritter, and Donghang Zhang, 2007, Do today’s trades aﬀect tomorrows IPO allocations? Journal of Financial Economics 84, 87-109.

[23] Pulliam, Susan and Randall Smith, 2000, Trade-oﬀs: Seeking IPO Shares, Investors Oﬀer to Buy More in the After Market, The Wall Street Journal December12, 2000.

[24] Pulliam, Susan and Randall Smith, 2000B, Linux Deal is focus of IPO-Commission Probe, The Wall Street Journal, December 12, 2000.

[25] Reuter, Jonathan, 2006, Are IPO allocations for sale? Evidence from Mutual Funds, Journal of Finance 5, 2289-2324.

[26] Ritter, Jay R.,1984, The ‘hot issue’market of 1980, journal of Business 32, 215-240.

[27] ____, 1987, The cost of going public, Journal of Financial Economics 19, 269-281.

[28] ____, 1991, The long run performance of Initial Public Oﬀerings, Journal of Fi- nance 46, 3-27.

[29] ____, 2003, Diﬀerences between European and American IPO markets, European Financial Management 4, 421-434.

[30] Rogers, W. H.,1993, Regression standard errors in clustered samples, Stata Technical Bulletin 13, 19-23. Reprinted in Stata Technical Bulletin Reprints 3, 88—94.

[31] White, Halbert, 1980, A heteroscedasticity-consistent covariance matrix estimator and a direct test of heteroscedasticity, Econometrica 48, 817-838.

27 Table 1 The number of Initial Public Oﬀerings on the Oslo Stock Exchange

The column labeled "IPOs" lists the number of Initial Public Oﬀerings on the Oslo Stock Exchange in the sample period. The column labeled "Data" indicates the IPOs with allocation data. The column labeled "Prospectus" lists the IPOs where we have been able to locate the listing prospectus. The column labeled "Sample" lists the 23 sample IPOs. The columns labeled "Value of shares" list the annually aggregate million USD values of shares sold in the 153 IPOs with listing prospectus. "All", "New" and "Secondary" indicates the value of all shares, only newly issued shares and shares sold by existing shareholders respectively. The columns labeled "P" and "S" is the annual aggregated USD million value of shares sold in the IPOs with prospectuses and in the 23 IPO sample respectively. Value of shares sold is reported in USD using a USD/NOK exchange rate of 0.1792. The sample period is January 1993 through September 2007.

NumberofIPOs Valueofshares(MillionUSD) All New Secondary Year IPOs Data Prospectus Sample P S P S P S 1993119 7 541 539 2 1994 15 9 8 3 275147 218142 57 5 1995 14 12 9 2 452 49 403 1996 15 11 7 2 13780 56 8180 1997 29 25 19 9 976230 504 21 471209 1998 12 9 8 1 18987 14587 43 199933 3 50 21 29 2000 10 10 10 2 817101 753 89 64 12 2001 44 4 183 166 17 2002 2 2 2 2 7070 6465 515 200322 2 83 78 5 2004 14 14 14 1,605 1,319 287 2005 31 30 30 2 2,041 34 566 23 1,475 11 2006 18 17 16 2,730 2,237 493 2007 15 14 14 912 517 395

Total 195 171 153 23 11,061 749 7,232 427 3,873 322

28 Table 2 Summary Statistics of Firms Going Public on the Oslo Stock Exchange

Panel A reports the company characteristics for the 23 sample IPOs and all 171 IPOs with ownership data. "After-listing/ issued" is the additional shares purchased after the listing divided by the shares issued in the IPOs. "-Sell within 6 months " is the "(After-listing shares/shares issued) %" for only investors that sell all shares within six months of the listing date. "Market V. (M.USD)" is the number of shares outstanding on the listing day times the first day closing price. "Book/Market" is the book value of equity after the IPO divided by the market value on the listing day. "Offer price" is the USD IPO price in the listing prospectuses. "VC backed d." is a dummy variable that takes the value of one if the company has venture capital backing. "High-tech d." takes the values of one for IT -companies. "First day return" is the percentage price change from the offer price to the first day closing price. USD values are calculated from a USD/NOK exchange rate of 0.1792. In Panel B the 23 Sample IPOs are split into IPOs with high and low after-listing purchases by investors that sell all shares within six months of the listing date. T —statistics are calculated as: Difference / (square root [(variance sample 1/ numbers in sample 1) + (variance sample 2/ numbers in sample 2)].

PanelA Sample23IPOs All171IPOs Meandifference N Mean Std.Dev Median N Mean Std.Dev Median Diff. t-stat. After-listing/issued 23 8.7% 7.6% 6.2% 171 5.8% 6.2% 3.8% 2.9% (1.8) -Sellwithin6months 23 3.3% 3.7% 2.3% 171 3.6% 5.0% 1.7% -0.3% (-0.3) MarketV.(M.USD) 23 $149.3 $145.2 $117.3 171 $311.4 $871.9 $108.3 -$162.1 (-2.2) OfferpriceUSD 23 $8.7 $6.9 $7.2 171 $8.2 $6.4 $6.8 $0.5 (0.3) Book/Market 23 0.3 0.29 0.23 171 0.46 0.33 0.4 -0.16 (-2.4) VCbackedd. 23 0.17 0.39 0.0 171 0.18 0.38 0.0 -0.01 (-0.1) High-techd. 23 0.09 0.29 0.0 171 0.12 0.32 0.0 -0.03 (-0.5) Firstdayreturn 23 0.13 0.19 0.09 171 0.08 0.19 0.03 0.05 (1.2)

PanelB 11highladderingIPOs 12lowladderingIPOs Meandifference N Mean Std.Dev Median N Mean Std.Dev Median Diff. t-stat. After-listing/issued 11 12.6% 8.5% 8.4% 12 5.2% 4.7% 4.4% 7.4% (2.6) -Sellwithin6months 11 6.2% 3.5% 5.2% 12 0.7% 0.8% 0.4% 5.5% (5.1) MarketV.(M.USD) 11 $117.3 $67.1 $95.8 12 $178.8 $190 $144.9 -$61.5 (-1.1) Book/Marketratio 11 0.34 0.3 0.26 12 0.26 0.29 0.22 0.08 (0.6) Offerprice(USD) 11 $9.0 $5.4 $8.1 12 $8.5 $8.3 $5.7 $0.5 (0.2) VCbackedd. 11 0.09 0.3 0.0 12 0.25 0.45 0.0 -0.16 (-0.1) High-techd. 11 0.0 0.0 0.0 12 0.17 0.39 0.0 -0.17 (-1.5) Firstdayreturn 11 0.16 0.16 0.18 12 0.1 0.21 0.06 0.06 (0.8)

29 Table 3 Summary Statistics on IPO Allocations and on Investors Trading

Panel A reports the summary statistics for the individual trading prior to the 23 sample and all 171 IPOs on the Oslo Stock Exchange in the period 1993 to 2007. "(Allocated/issued) " is the number of allocated shares to each investor divided by the shares issued in the IPO. "(After-listing/issued) " is the additional shares purchased after the listing divided by the shares issued in the IPOs. "Commission" is the accumulated commission generated in USD by the investors in the two years before the IPO allocation date. "Portfolio value" is the portfolio value in million USD for each allocated investor at 31.12.xx in the year before the IPO allocation date. "Previous IPOs" is the accumulated previous IPO participations by the investors divided by the accumulated IPO number in the sample." Previous Buy-and-hold " is the accumulated previous number of times the allocated investor has been a buy-and-hold investor as a percent of all previous IPO participations. This is the number of times the investor has held some IPO allocated shares for more than six months in previous IPOs. "Previous Flipping" is the accumulated number of times the investor have flipped previous IPOs as a percent of all previous IPO participations before the IPO allocation. Flipping is when all shares are sold within one month of the listing. USD values are calculated from a USD/NOK exchange rate of 0.1792. Panel B reports that investors that buy (and sell) more shares after the listing are allocated significantly more IPO shares than investors that do not. T —statistics are calculated as: Difference / (square root [(variance sample 1/ numbers in sample 1) + (variance sample 2/ numbers in sample 2)].

Panel A Sample23IPOs All171IPOs N Mean Std.Dev Median N Mean Std.Dev Median (Allocated/issued) 16,593 0.053% 0.173% 0.009% 175,382 0.036% 0.14% 0.003% (After-listing/issued) 16,593 0.011% 0.19% 0.0% 175,382 0.006% 0.12% 0.0% CommissionUSD 16,593 $3,544 $46,711 $37.9 175,382 $6,274 $93,395 $30.8 PortfoliovalueM.USD 16,593 $2.6 $44.5 $0.003 175,382 $3.6 $72.6 $0.004 PreviousIPOs 16,593 0.05 0.05 0.04 175,382 0.03 0.05 0.017 PreviousBuy-and-hold 16,593 0.21 0.37 0.0 175,382 0.21 0.37 0.0 PreviousFlipping 16,593 0.15 0.31 0.0 175,382 0.1 0.25 0.0

Panel B: Comparing IPO allocations to after-listing investors and non after-listing investors Ladderinginvestors Allinvestors MeanDiﬀerence N Mean Std.Dev Median N Mean Std.Dev Median Diﬀ. t-stat. *D1 427 0.116% 0.264% 0.018% 16,593 0.053% 0.173% 0.009% 0.06% (4.9) *D1*D2 195 0.097% 0.263% 0.009% 16,593 0.053% 0.173% 0.009% 0.04% (2.3) *D1*D3 357 0.097% 0.239% 0.017% 16,593 0.053% 0.173% 0.009% 0.04% (3.5) *D1*D2*D3 195 0.097% 0.263% 0.009% 16,593 0.053% 0.173% 0.009% 0.04% (2.3)

30 Table 4 Relationship between After-listing Purchases and IPO Allocations

This table reports the coeﬃ cients and heteroscedastic consistent t -statistics (errors adjusted for clus- tering across ﬁrms Rogers, 1993) in parentheses for the regressions with (allocated shares/issued shares) % as the dependent variable. This is a standard OLS model. Only the 23 IPOs with exact allocations are included. All variables are as described in Table 2 and Table 3. In regression 6 and 7 D1 indicates if more than 50% of shares are sold within three and six months.

(Allocated shares/shares issued) % Reg 1 Reg 2 Intercept 0.6366 0.7096 (2.3) (2.8) (After-listingshares/sharesissued)% 0.0738 0.054 (1.6) (1.4) (After-listingshares/sharesissued)%*D1 0.1306 (1.7) D1-Investorssellshareswithinmonthsalisting -0.0283 (-1.3) Log (commission) 0.0111 0.0096 (2.5) (2.2) Log (portfolio value) 0.0014 0.0018 (0.6) (0.9) Previous IPOs 0.2242 0.2085 (0.7) (0.7) Previousbuy-and-hold -0.0076 -0.0046 (-0.3) (-0.2) Previous ﬂipping -0.019 -0.0207 (-0.6) (-0.6) Financialinstitutiondummy 0.0657 0.0743 (0.9) (1.1) Log (market value) -0.0573 -0.0602 (-4.2) (-4.8) BV / MV equity 0437 0.4309 (7.3) (6.7) Oﬀer price -0.0024 -0.0023 (-4.0) (-3.9) VC backed dummy 1.0986 1.1027 (15.5) (14.5) High-tech dummy -0.9268 -0.9488 (-12.0) (-14.5) First day return dropped dropped Companyandyeardummy yes yes Observations(IPOallocations) 1,016 1,016 -of which are laddering investors 427 Adjusted R -squared 33.6% 35.2% Investorssellwithinmonthsoflisting all6m.

31 Continued... (Allocatedshares/sharesissued)% Reg3 Reg4 Reg5 Reg6 Reg7 Intercept 6.0717 1.9466 1.268 1.3036 1.1851 (16.8) (9.4) (10.6) (10.6) (13.7) (After-listingshares/sharesissued)% 0.0486 0.0587 0.0587 0.062 0.0487 (1.2) (1.4) (1.4) (1.5) (1.4) (After-listingshares/sharesissued)%*D1*D2 0.286 (4.2) (After-listing shares/shares issued) %* D1*D3 0.2698 (4.9) (After-listingshares/sharesissued)%*D1*D2*D3 0.286 0.3013 0.2398 (4.2) (2.9) (7.7) D1 -Investorssellshareswithin monthsa listing -0.0384 -0.0249 -0.0249 -0.0434 -0.0093 (-1.6) (-1.2) (-1.2) (-1.7) (-0.5) D2-Positivedriftinsharepriceafterthelisting -0.2606 -0.2075 -0.2138 -0.2053 (-5.3) (-4.5) (-4.5) (-4.5) D3-UnderpricedIPO 0.5297 0.1551 0.1713 0.1267 (20.1) (2.6) (2.7) (2.0) Log(commission) 0.0083 0.0085 0.0085 0.0103 0.0094 (1.8) (1.8) (1.8) (2.4) (2.3) Log(portfoliovalue) 0.0024 0.0023 0.0023 0.0015 0.0019 (1.2) (1.2) (1.2) (0.8) (1.0) PreviousIPOsofpossible 0.1996 0.2351 0.2351 0.2691 0.2274 (0.6) (0.8) (0.8) (0.9) (0.8) Previousbuy-and-holdofpossible -0.002 -0.0072 -0.0072 -0.007 -0.0061 (-0.1) (-0.3) (-0.3) (-0.3) (-0.2) Previousﬂippingofpossible -0.0172 -0.0167 -0.0167 -0.011 -0.0194 (-0.5) (-0.5) (-0.5) (-0.4) (-0.6) Financialinstitutiondummy 0.0703 0.0641 0.0641 0.0644 0.0611 (1.1) (0.9) (0.9) (0.9) (0.9) Log(marketvalue) -0.3003 -0.0983 -0.0636 -0.0656 -0.0599 (-16.8) (-9.6) (-11.5) (-11.4) (-15.9) BV/MVequity dropped dropped dropped dropped dropped

Oﬀerprice -0.009 -0.0016 -0.0023 -0.0025 -0.002 (-12.4) (-1.8) (-2.9) (-3.1) (-2.5) VCbackeddummy -0.4496 0.6022 0.5567 0.5428 0.6072 (-17.2) (-16.6) (7.0) (6.5) (7.4) High-techdummy -1.0434 -1.0831 -0.8529 -0.8437 -0.8859 (-12.7) (-16.8) (-8.8) (-8.5) (-9.6) Firstdayreturn dropped dropped dropped dropped dropped Companyandyeardummy yes yes yes yes yes Observations(IPOallocations) 1,016 1,016 1,016 1,016 1,016 -ofwhichareladderinginvestors 357 195 195 145 217 AdjustedR-squared 38.3% 37.0% 34.5% 36.2% 37.9% Investorssellwithinmonthsoflisting all6m. all6m. all6m. 50%3m. 50%6m.

32 Table 5 After-listing Purchases in Past IPOs Give More Future IPO Ownership

This table reports the coeffi cients and heteroscedastic consistent t -statistics (errors adjusted for clus- tering across firms Rogers, 1993) in parentheses for the regressions with (allocated shares/issued shares) % as the dependent variable. This is a standard OLS model. All variables are as described in Table 2 and Table 3. Regression 1 includes all IPOs. Regression 2 and 3 includes hot and cold IPOs respectively. There are 171, 105 and 45 IPOs in regression 1, 2 and 3. Regression 4 to 6 drop all company specific control variables. Past laddering includes only investors who have purchased more shares right after the listing and then sold some of the shares within six months of the listing date in past IPOs.

(Allocated shares/shares issued) % Reg1 Reg2 Reg3 Reg4 Reg5 Reg6 Intercept -0.12984 0.1096 0.7882 -0.0278 2.0078 0.0587 (-25.9) (6.9) (100.0) (-7.0) (90.8) (10.6) Previousladdering 0.1114 0.1137 0.0877 0.1114 0.1137 0.0877 (9.3) (8.2) (3.4) (9.3) (8.2) (3.4) Log(commission) 0.006 0.0051 0.012 0.006 0.0051 0.012 (4.1) (3.6) (8.9) (4.1) (3.6) (8.8) Log(portfoliovalue) 0.0085 0.0007 0.0013 0.0009 0.0007 0.0013 (2.6) (2.3) (2.2) (2.6) (2.3) (2.2) PreviousIPOs 0.1535 0.1619 0.1865 0.1535 0.1619 0.1865 (3.8) (3.6) (3.0) (3.8) (3.6) (3.0) Previousbuy-and-hold -0.0156 -0.0153 -0.0177 -0.01558 -0.0153 -0.0177 (-7.8) (-7.0) (-3.0) (-7.8) (-7.0) (-3.0) Previousﬂipping -0.0022 -0.0051 0.0061 -0.0022 -0.0051 0.0061 (-0.9) (-1.9) (0.9) (-0.9) (-1.9) (0.9) Financialinstitutiondummy 0.1779 0.1653 0.1807 0.1779 0.1653 0.1807 (9.5) (6.9) (5.3) (9.5) (6.9) (5.3) Log(marketvalue) 0.0121 -0.008 -0.0357 (21.4) (-9.8) (-109.1) BV/MVequity -0.0108 0.0474 -0.2327 (-2.1) (10.3) (-90.3) Oﬀerprice -0.0011 -0.001 0.0008 (-63.6) (-32.3) (64.4) VCbackeddummy 0.1196 0.0493 -0.2708 (29.7) (5.0) (-85.9) High-techdummy -0.1668 -0.44 0.0599 (-33.4) (-42.8) (27.3) Firstdayreturn 0.4163 0.3627 dropped (16.3) (12.5) Companyandyeardummy yes yes yes yes yes yes Observations 175,382 145,392 22,114 175,382 145,392 25,891 AdjustedR-squared 22.0% 21.7% 20.1% 22.0% 21.7% 20.1% IPOs all hot cold all hot cold

33 Table 6 Actual Return from After-listing Purchases

This table reports the average USD return for the investors that buy more shares after the listing. Only IPOs with exact IPO allocations are included in the analysis (23 IPOs). First day return $ is calculated as: the number of shares allocated in the IPO * (first day closing price - offer price) * 0.1792 (The NOK/USD exchange rate). First month return $ is calculated as: (The number of shares allocated in the IPO + The shares purchased after the listing) * ( Price one month after the listing - first day closing price) * 0.1792 (The NOK/USD exchange rate). Panel A investigate only IPO allocated investors with after-listing buying that sell early in hot IPOs. Panel B investigate only IPO allocated investors with after-listing buying that sell early in IPOs with a positive drift after the listing. Panel C investigate only IPO allocated investors with after-listing buying that sell early in cold IPOs. Panel C includes all IPO trading for the 70 investors that buy more shares after the listing in the cold IPOs. These 70 investors lose money on their cold IPO after-listing purchases, but they earn money in total. Together these investors receive 447 allocations in the sample. Panel D and E investigates non-allocated IPO investors who buy shares after the listing. Panel D investigates the 11 IPOs with high laddering. Panel E investigates the 12 IPOs with low laddering.

Panel A: (IPOs=14) First day return $ First month return $ Total return $ Std.Dev. Investors Allinvestors $6,526 $9,197 $15,723 $59,730 357 Institutions $16,400 $21,866 $38,265 $106,181 92

Panel B: (IPOs=6) First day return $ First month return $ Total return $ Std.Dev. Investors Allinvestors $6,712 $15,592 $22,303 $61,822 195 Institutionsonly $17,217 $50,733 $67,949 $117,353 40

Panel C: First day return $ First month return $ Total return $ Std.Dev. Allocations Allinvestors $17,705 $3,744 $21,449 $167,687 447 Institutionsonly $46,297 $13,428 $59,725 $266,893 169

Panel D: The 11 IPOs with high laddering Sixmonthreturn$ Std.Dev. Investors Allinvestors -$5,611 -$139,736 10,748 Institutionsonly -$22,189 -$339,564 1,806

Panel E: The 12 low laddering IPOs Sixmonthreturn$ Std.Dev. Investors Allinvestors -$324 -$327,450 6,554 Institutionsonly $1,276 $711,068 1,388

34 Table 7 After-listing Purchases and Generated stock-trading Commissions

This table reports the coeﬃ cients and White (1980) heteroscedasticity consistent t-statistics in parentheses for the regressions with the aggregate (after-listing shares/shares issued) % as the dependent variable. All variables are as described in Table 2 and Table 3. All regressions are standard OLS models, and the sample period is from January 1993 to September 2007. Only investors that sell some shares within six months of the listing are included in Aggregate (After-listing shares/shares issued) %. Only investors that do not buy shares after the listing are included in Log (average commission per share). Regression 2 and 4 drop the variables that Hao (2007) and Aggarwal et al. (2006) predict increase laddering. Regression 3 and 4 use average commission by shares instead of the sum of commission scaled by commission in all IPOs.

Log (aggregate after-listing shares/shares issued) % Reg1 Reg2 Reg3 Reg4 Intercept 1.5289 1.5296 0.6737 0.3876 (1.5) (1.6) (0.7) (0.4) (Combinedcommission)% 51.5725 44.7781 (3.9) (4.4) Log(averagecommissionpershare) 0.2172 0.2116 (1.8) (1.7) Log(marketvalue) -0.0289 -0.03153 0.01 0.0306 (-0.5) (-0.5) (0.2) (0.5) BV/MVequity -0.147 -0.1551 -0.2652 -0.2938 (-0.5) (-0.5) (-0.9) (-0.9) VCbackeddummy -0.4901 -0.6445 -0.505 -0.5339 (-1.5) (-2.3) (-1.5) (-1.7) High-techdummy -0.2395 -0.2797 -0.1333 -0.1139 (-0.7) (-0.9) (-0.4) (-0.3) Absolutepricerevision -0.0305 -0.0044 (-0.2) (-0.2) Sentimentinvestors(million) 0.0000 0.0001 (-0.3) (2.2) Observations 171 171 171 171 AdjustedR-squared 7.8% 7.4% 4.7% 4.1%

35 Table 8 Relationship between After-listing Purchases and IPO Allocations -Robustness

This table reports the coeﬃ cients and heteroscedastic consistent t -statistics (errors adjusted for clus- tering across ﬁrms Rogers, 1993) in parentheses for the regressions with (allocated shares/issued shares) % as the dependent variable. This is a standard OLS model. Only the 23 IPOs with exact allocations are included. All variables are as described in Table 2 and Table 3. In regression 1 all PCC (savings banks) are included. In regression 2 IPO allocations are trimmed at 0.1%.

Intercept 1.5432 1.0574 (14.3) (2.8) (After-listingshares/sharesissued)% 0.0772 0.249 (1.9) (4.6) (After-listingshares/sharesissued)%*D1*D2*D3 0.2674 0.5255 (3.7) (1.7) D1-Investorssellshareswithinsixmonthsalisting -0.0201 -0.0326 (-1.1) (-0.6) D2-Positivedriftinsharepriceafterthelisting 0.1242 -0.1165 (7.6) (-1.6) D3-UnderpricedIPO -0.1832 -0.4437 (-15.0) (-5.5) Log (commission) 0.0093 0.029 (2.5) (2.1) Log(portfoliovalue) 0.0035 0.0005 (2.0) (0.1) PreviousIPOsofpossible 0.0171 0.0135 (0.1) (0.0) Previousbuy-and-holdofpossible -0.004 0.0311 (-0.2) (0.6) Previousﬂippingofpossible -0.0346 -0.0602 (-1.1) (-1.3) Financialinstitutiondummy 0.1086 0.4664 (1.5) (1.7) Log (market value) -0.0779 -0.0727 (-19.2) (-4,2) BV / MV equity 0.0331 dropped (6.3) Oﬀer price 0.0002 0.0152 (1.4) (11.4) VC backed dummy 0.7644 -0.4935 (23.7) (-4.8) High-tech dummy 0.3245 0.4879 (12.8) (7.2) First day return dropped dropped Companyandyeardummy yes yes Observations(IPOallocations) 1,251 1,064 -ofwhichareladderinginvestors 209 200 Adjusted R -squared 34.2% 31.6% Investorssellwithinmonthsoflisting all6m. all6m. 36 Table 9 IPO Allocations and After-listing Purchases -Robustness 2

This table reports the coeffi cients and heteroscedastic consistent t -statistics (errors adjusted for clus- tering across firms Rogers, 1993) in parentheses for the regressions with (allocated shares/issued shares) % as the dependent variable. This is a standard OLS model. Only the 23 IPOs with exact allocations in the sample period are included. All variables are as described in Table 2 and Table 3. All IPO specific control variables are removed.

(Allocated shares/shares issued) % Reg1 Reg2 Reg3 Reg4 Intercept -0.3023 0.3133 0.2943 -0.2871 (-2.2) (1.5) (4.2) (-1.4) (After-listingshares/sharesissued)% 0.054 0.0486 0.0587 0.0587 (1.4) (1.2) (1.4) (1.4) (After-listingshares/sharesissued)%*D1 0.1306 (1.7) (After-listingshares/sharesissued)%*D1*D2 0.286 (4.2) (After-listingshares/sharesissued)%*D1*D3 0.2698 (4.9) (After-listingshares/sharesissued)%*D1*D2*D3 0.286 (4.2) D1-Investorssellshareswithin6m. a. listing -0.0283 -0.0384 -0.0249 -0.0249 (-1.3 (-1.4) (-1.2) (-1.2) D2-Positivedriftinsharepriceafterthelisting -0.6103 0.5593 (-6.4) (24.9) D3-UnderpricedIPO -0.6049 -0.5883 (-9.0) (-8.9) Log(commission) 0.0096 0.0083 0.0085 0.0085 (2.2) (1.8) (1.8) (1.8) Log(portfoliovalue) 0.0018 0.0024 0.0023 0.0023 (0.9) (1.2) (1.2) (1.2) PreviousIPOsofpossible 0.2085 0.1996 0.2351 0.2351 (0.7) (0.6) (0.8) (0.8) Previousbuy-and-holdofpossible -0.0046 -0.002 -0.0072 -0.0072 (-0.2) (-0.1) (-0.3) (-0.3) Previousﬂippingofpossible -0.0207 -0.0172 -0.0167 -0.0167 (-0.6) (-0.5) (-0.5) (-0.5) Financialinstitutiondummy 0.0743 0.0703 0.0641 0.0641 (1.1) (1.1) (0.9) (0.9) Observations(IPOallocations) 1,016 1,016 1,016 1,016 -ofwhichareladderinginvestors 427 357 195 195 AdjustedR-squared 35.2% 38.3% 37.0% 37.0%

37 Table 10 Aggregate IPO Laddering and Allocations

This table reports the aggregate allocation and laddering at the IPO level. Panel A includes the 11 high laddering IPOs. Panel B includes also the 12 low laddering IPOs.

Panel A Group Investors Allocation Laddering Total IPOs D1 363 3.6% 6.2% 9.8% 11 D1,D3 317 3.5% 6.3% 9.8% 9 D1,D2 174 4.5% 8.7% 13.2% 4

Panel B D1 427 2.5% 3.5% 6.0% 20 D1,D3 357 2.5% 4.5% 7.0% 14 D1,D2 195 5.9% 3.2% 9.1% 6

38 Figure 1 The Factors that Create the Incentives to Engage in IPO Laddering

Laddering investors are allocated some shares in the IPO and then they buy more shares after the listing before they sell all shares. Commission investors are investors that generate high levels of stock- trading commission to the investment bank through trading in other shares. Investment bank is the lead manager in the IPO. Hot and cold IPOs are high and low oversubscribed IPOs. Hot and cold IPOs are proxied for by positive and non-positive ﬁrst day return.

Hot IPOs Laddering Agreetobuymoresharesafterthelistingtoincrease Pulliam and Smith(2000) investors currenthotIPOallocations andtheSEClitigation releases

Commission Pay increased stock-trading commissions to the investment Reuter (2006) and investors bank,throughtradesinothershares,toincrease Nimalendran,Ritter currentandfuturehotIPOallocations andZhang(2006)

Investment banks 1) Increase received commissions by allocating IPO Hao (2007) shares to laddering investors and commission investors

2) Ensure a successfulIPO by allocating shares to Hao (2007) laddering investors that increase prices after the listing

Cold IPOs Laddering Agreetobuymoresharesafterthelistingtoincrease Griﬃ netal. (2007) investors futurehotIPOallocations

Investment banks 1) Ensure a more successful IPO by allocating shares to Griﬃ n et al. (2007) laddering investors that increase prices after the listing and Hao (2007)

2) Reduces after-listing price uncertainty by allocating Griﬃ n et al. (2007) shares to laddering investors

3) Reduces the risk of damaged reputation from IPOs Griﬃ n et al. (2007) that fall in price by allocating shares to laddering investors

39 Figure 2 Theoretical Predictions By Hao (2007) and Aggarwal et al. (2006)

Predictions made by Hao (2007) Laddering will increase the following variables: 1) Laddering results in a higher offer price if investors are not expected to sell shares in the six m. after-listing 2) Laddering is positively related to money left on the table. 3) Laddering in itself does not necessarily increase underpricing. 4) Laddering contributes to long-run underperformance. The following variables will increase laddering: 5) More expected underpricing (without laddering) leads to more laddering 6) When there are information momentum effects, there is more laddering. 7) When underwriters shares in on the profits from underpriced IPOs, there is more laddering

Predictions made by Aggarwal et al (2006) Laddering will increase the following variables: 1) Returns should be higher for IPOs with laddering over the six months after the listing 2) The long-run return should be lower for IPOs with laddering than for IPOs with no laddering 3) The number of sentiment investors increases IPO underpricing for IPOs with laddering. 4) Turnover and volume (shares traded) are greater for IPOs with laddering than for IPOs with no laddering The following variables will increase laddering: 5) Underpricing is higher for IPOs with laddering than for IPOs with no laddering 6) When there are more sentiment investors there is a bigger likelihood of laddering

Major Diﬀerences 1) Hao (2007) predict intentional underpricing, and Aggarwal et al. (2006) predict price run ups that are corrected 2) Hao (2007) and Aggarwal et al. (2006) predicts that laddering increases oﬀer/closing and underpricing respectively

40 Figure 3 Timeline of the IPO Allocations for the Diﬀerent Groups

Listing in database is when the company list ownership records in the ownership database. This is when the ownership records are observed in the data the ﬁrst time. IPO allocation is when the companies distribute the allocated shares in the ownership database. Listing is when the company is listed publicly. After-listing purchases is when the laddering trades are calculated. Group 1 to 3 is the ordering of the group of detail in the allocations. Group 1 is 100% accurate IPO allocations. Group 2 IPO allocations includes one to 30 days of after-listing trading. Group 3 IPO allocations includes existing owners who have not sold all of their shares in the IPO. There are 23, 143 and 5 companies in group 1, 2 and 3 respectively.

Timeline Sixmonthsbefore Onemonthbefore Listingmonth Onemonthafter thelisting thelisting thelisting

Group1 Listingindatabase IPOallocation Listing After-listingpurchases

Group2 Listingindatabase IPOallocation After-listingpurchases Listing

Group3 Listingindatabase Listing After-listingpurchases IPO allocation

41 .

42 3 Using Stock-trading Commissions to Secure IPO Allocations

Sturla Lyngnes Fjesme34 BI Norwegian Business School

Roni Michaely Cornell University and the Interdisciplinary Center

Øyvind Norli BI Norwegian Business School

. JEL classiﬁcation: G24; G28 Keywords: IPO allocations; Equity issue; Commission; Rent seeking

34 We are grateful to Jay Ritter, Øyvind Bøhren, François Derrien, seminar participants at BI Norwegian Business School for valuable suggestions, “The Center for Corporate Governance Research (CCGR)”at BI Norwegian Business School for ﬁnancial support, the Oslo Stock Exchange VPS for providing the data and the investment banks and companies that helped us locate the listing prospectuses. All errors are our own. Contact: BI Norwegian Business School, Nydalsveien 37, 0484 Oslo, Norway. E-mail address: [email protected] Telephone: +47-957-722-43.

43 Abstract

Using data, at the investor level, on the allocations of shares in initial public offerings (IPOs), we document a strong positive relationship between the amount of stock-trading commission and the number of shares an investor receives in a subsequent IPO. We find no evidence to support the idea that investment banks allocate shares to investors that are perceived to be long-term investors. Our findings are consistent with the view that investment banks are able to capture some of the profits earned by investors when participating in underpriced IPOs.

44 3.1 Introduction The Securities and Exchange Commission (SEC) has since the early 2000s investigated the initial public offering (IPO) allocation practices of several major investment banks. One concern is that IPO allocations are tied to excessively large stock-trading commissions and that such a practice would constitute illegal kickbacks from investors to investment banks. Reuter (2006) points out that such kickbacks would allow the underwriter to share more of the benefits of underpriced IPOs– and, therefore, exacerbate the agency conflict that exists between the issuing firm and the lead underwriter of the IPO. This paper investigates whether or not investors that has generated large stock-trading commissions in the past receives a preferential treatment in future IPO allocations. Using data on the stock-holdings for every single investor that owned common shares that was listed or became listed on the Oslo Stock Exchange during the period 1993 through 2007, we are able to link stock-trading commission and IPO allocation at the investor level. The main finding of the paper is a strong and robust positive relationship between the level of stock-trading commission generated by an investor prior to the IPO and the number of shares the same investor receives through the IPO allocation. It can be argued that large investors that generate more commissions are likely to apply for more IPO shares. However, the economic and statistical significance of the relationship between commission and allocation is robust to controlling for the market value of the investors portfolio, as well as to other investor characteristics. We conclude that investors generating large stock-trading commission receives the most IPO shares because of the commission they generate. Other investor characteristics are of less or no importance for IPO allocations. The empirical research on the allocation practices of investment banks have been ham- pered by the lack of data on IPO allocations.35 Since information about stock-trading commissions are equally hard to come by, there is little empirical research on the relationship between commissions and allocations. One exception, and the paper closest to ours, is Reuter (2006) who finds a positive correlation between stock-trading commission paid by mutual funds to lead investment banks and the holdings in IPOs underwritten by the same banks. This suggests, in general, that having a business relationship with the lead underwriter increases the chance of getting shares in underpriced IPOs. In particular, it suggests that investors can “buy” allocations by channeling their trades through the brokerage arm of the lead underwriter. In another related paper, Nimalendran, Ritter and Zhang (2006) show that there is a positive relationship between money left on the table in IPOs and trading volume in liquid shares around IPO allocation dates. This is indirect evidence of a positive relationship between trading commission and IPO allocations. The strength of our paper, compared to the existing literature, is that we are able to analyze exact allocations at the investor level. In the main part of the paper, we study 24,308 IPO allocations.36 The existing literature has suggested at least three potential explanations for what determines investment banks’decision of which investors are getting shares in oversubscribed IPOs. First, Benveniste and Spindt (1989) suggest that investment banks allocate IPO shares to informed investors in return for a truthful reve- lation of their valuation of the issuer. Second, investment banks themselves tend to argue

35 See Ritter (2003) and Jenkinson and Jones (2004) for papers that study IPO allocations and summarizes IPO allocation studies. 36 These exact allocations are from 30 diﬀerent IPOs. In other words, there are 24,308 unique investor- IPO combinations in our data. In robustness tests, we also study 162,384 investor-IPO pairs where the IPO allocation data might be contaminated with some post-IPO trading.

45 that they are looking for long-term investors. Third, investment banks allocate shares to investors that can provide some form of kickback. The empirical literature provides mixed results in terms of understanding IPO allocations in the light of the above three potential explanations.37 An important contribution of our paper is that we examine and contrast all three potential explanations simultane- ously. As already mentioned, our data strongly support the view that investors can secure themselves IPO allocations through large stock-trading commissions. We find no evidence of a preferential treatment of buy-and-hold investors. Neither do we find any support for the idea that investors get allocation in return for revealing private information about issuing firm value. The rest of paper is organized as follows: Section 3.2 describes the related literature. Section 3.3 describes theoretical predictions and the testable implications. Section 3.4 describes the data set. Section 3.5 gives the empirical results, and section 3.6 concludes.

3.2 Related literature Ritter (2003) and Jenkinson and Jones (2004) argue that there are three views on how IPOs are allocated. First, is the academic view based on Benveniste and Spindt (1989). In this view, investment banks allocate IPO shares to informed investors in return for true valuation and demand information. Second, is the pitchbook view where investment banks allocate shares to institutional investors that are likely to be buy-and-hold. Finally, is the rent seeking view where investment banks allocate shares to investors in return for kickbacks. Existing research have found support for each of these views and against the academic view and the pitchbook view. The existing research have investigated these views one by one. The exception is Jenkinson and Jones (2004) that compares the academic view to the pitchbook view. There are many papers that investigates the academic view alone, as described by Benveniste and Spindt (1989). Cornelli and Goldreich (2001) investigate the order book of 23 and 16 international IPOs and SEOs. They find that regularly participating, large bid and domestic participants are favored in allocations. They also find that bidders that participate in both hot and cold issues are given larger allocations in hot issues. Cornelli and Goldreich (2003) investigate the order book of 37 and 26 international IPOs and SEOs. They find that bids from large, frequent bidders that include a limit price affect the issue price. It is concluded that book-building is designed to extract information from investors. Ljungqvist and Wilhelm (2002) look at allocations between institutional and retail investors for 1,032 international IPOs. They find that institutional investors are favored over retail investors. They find that an increased institutional allocation is linked to a higher deviation from the midpoint of the book-building pricing range to the final offer price. It is concluded that underwriters use institutional bids to set the offer prices in IPOs. Binay, Gatchev and Pirinsky (2007) investigate 4,668 U.S. IPOs and find that underwriters favor institutions they have previously worked with. A relationship with the underwriter is more important in IPOs with strong demand, IPOs of less liquid firms and IPOs by less famous underwriters. It is argued that favoring regular investors is done to price IPOs more correctly. Regular investors have incentives to report their true value in the book-building so that they will be favored in future IPOs. Bubna and Prabhala (2007) investigate 137 Indian IPO allocations. They find that book-building and discretion in

37 Table 1 summarizes related papers.

46 allocation enhances pre-market price discovery. All these papers are consistent with the academic view of IPO allocations The two main papers that investigate the pitchbook view of IPO allocations are Jenk- inson and Jones (2004) and Aggarwal (2002). Jenkinson and Jones (2004) study 27 European IPO order books to compare the pitchbook view to the academic view. They find that there is limited information gathering in the book-building procedure. This is inconsistent with the academic view. Jenkinson and Jones (2004) do, however, find evidence in favor of the pitchbook view and concludes that IPO allocations are made to buy-and-hold investors. Aggarwal (2002) investigate the pitchbook view by looking at flipping activity of institutional and retail investors in 193 U.S. IPOs. It is found that institutional investors flip a larger part of their IPO allocations than retail investors. This is taken as evidence against the pitchbook view. This view argues that institutions are allocated more IPO shares because institutions are more likely to be buy-and-hold than retail investors. There are four types of IPO rent seeking that have led to investigations (and settlements) between the SEC or NASD and different investment banks, see Liu and Rit- ter (2010).38 IPO allocations can be tied to future corporate business (IPO spinning), after-listing purchases of the IPO shares (IPO laddering) and stock-trading commissions. Issuing companies can also agree to heavy underpricing in return for after-listing coverage from star analysts provided by the investment bank (analyst conflict of interest). The underpriced shares are then allocated to clients that generate high commissions so that the investment bank is able to recapture some of the underpricing. All of these allocation practices have been looked at in different studies. Liu and Ritter (2010) investigate IPO spinning, Fjesme (2011) investigate IPO laddering, Cliff and Denis (2004) investigate analyst conflict of interest and Reuter (2006) and Nimalendran, Ritter and Zhang (2006) investigate IPO allocations for commission trading. Reuter (2006) investigate if IPO allocations are tied to stock-trading commissions by studying 1,868 IPOs on NYSE, AMEX and Nasdaq in the period 1996 to 1999. Reuter (2006) find a positive relationship between stock-trading commissions paid by mutual funds to IPO lead underwriters and mutual fund holdings of IPO shares after the listing. It is concluded that commission generation is a likely reason behind IPO allocations for mutual funds. Reuter (2006) establish a link between IPO allocations and stock-trading commission for mutual funds, but it is not investigated if commission is important for IPO allocations for other investor groups. Nimalendran, Ritter and Zhang (2006) study investor trades in the 50 most liquid stocks in the U.S. during the days surrounding IPO allocations. They find that trading volume is positively related to money left on the table in IPOs. It is suggested that this increased trading is done purely to increase stock-trading commission as payment for IPO share allocations. Both of these papers support the rent seeking view.

3.3 Theoretical predictions and testable implications Investors are placed on A, B and C lists by investment banks before any IPO.39 Investors from the A list, that applies for shares, are more likely to receive more IPO allocations

38 Figure 1 describes the four types of IPO rent seeking. 39 The information about IPO allocation practices is obtained from meetings with former Norwegian investment bankers.

47 than investors on the B lists etc.40 We do not know how investors are placed on the A, B and C lists, but we expect that this is related to the pitchbook, the academic and the rent seeking view of IPO allocations. The investment bank prepares a list with proposed allocations after the book building/pricing of the IPOs that is given to the board of the issuing company. The board then decides IPO allocations based on this list. Anecdotal evidence suggests that most boards approve the proposed list without adjustments. Being on the A list of the lead investment bank is therefore very important when applying for IPO shares. In this paper we test if information gathering, allocation to long term buy-and-hold investors or rent seeking are likely reasons behind IPO allocations. We measure if providing pricing information, price stability or stock-trading commission will place investors on the A, B and C lists of the investment banks. The three allocation views are not mutually exclusive within or between IPOs. It is therefore possible that different IPOs are allocated based on different views. It is also possible that different investors are allocated shares based on different views within one IPO.

3.3.1 The rent seeking view of IPO allocations Rent seeking is an area that have received allot of attention in the media, but there is limited empirical research on the rent seeking view of IPO allocations. A likely reason for this is that it is very hard to obtain data to test for rent seeking. Testing for money transfers from one bank account to another obviously requires very detailed data. The cover-up activities needed to hide transfers as legitimate activities can be very creative. In this paper we focus on the rent seeking suggested in Reuter (2006) and Nimalendran, Ritter and Zhang (2006). We study if IPO allocations are related to generated stock- trading commission. In the Robert Stevenson settlement, an investment bank settled to pay a ﬁne for alledgedly tying IPO allocations to stock-trading commission, it was argued that clients both increased trading and increased commission rates per trade to receive IPO shares.41 In the Norwegian data it is only possible to test if there is a relationship between trading and allocations. It is not possible to detect changes in payments per trade in the data. The data also only let us test for commission generation on monthly trading.42 This means that it is possible that commission trading takes place even if it does not show up in the data. The rent seeking view is tested by regressing IPO allocations on stock-trading commission generated by the allocated investors and a set of control variables. Generated stock- trading commission is accumulated, per investors ID, over monthly portfolio changes in the past 24 months prior to any IPO. Only buy generated commission is included to avoid any issues related to portfolio rebalancing (to make room for the new shares). If investors that generate more commission are also allocated more IPO shares, we conclude that the rent seeking view is an accurate view of IPO allocations. If there is no relationship between commission and IPO allocations, we are not able to reject the rent seeking view because it is possible to hide this type of trading. It is also tested if there is a link between

40 Figure 2 describes all the steps in the IPO allocation process. 41 See January 9, 2003 NASD settlement http://www.ﬁnra.org/Newsroom/NewsReleases/2003/P002957 42 Commissions are generated from monthly data and not daily data. Because of this it is possible that commission trading takes place even if we are not able to ﬁnd it in our data set. If commissions are generated from daily buy and sell orders in the same shares we are not able to detect this. It is also possible that some investors pay higher commission rates to get allocations. This should, however, be discovered in auditing.

48 the number of IPO participations and commission. If rent seeking is an active strategy, there will be a strong relation between stock-trading commission and the number of IPO participations by each investor. It is possible that investors that get repeated allocations do so because they generate high commissions. IPO investors with single time IPO allocation may be kept out of future IPOs because they do not generate suﬃ cient levels of stock-trading commission.

3.3.2 The pitchbook view of IPO allocations The pitchbook view of IPO allocations comes from the sales pitch slides of the investment banks (Ritter, 2003). In these slides it is usually argued that investments banks will use their power to allocate shares to long term buy-and-hold investors. It is argued, by the investment banks, that buy-and-hold investors will create price stability that is good for the issuing company. If buy-and-hold is an accurate view of IPO allocations, investors that buy-and-hold IPO shares must also receive future IPO allocations. There must also be a punishment in terms of no (or at least less) future IPO allocations for investors that sell shares early (flipping investors).43An investment bank that underwrite many IPOs will have a more reliable reward/threat system than less active investment banks. Therefore, it should also be more buy-and-hold investors in IPOs by active investment banks. It should also not be possible for investors to repeat a flipping strategy in IPOs by the same bank over time.44 If buy-and-hold investors do not have the potential threat of not receiving future allocations, there is no point of being buy-and-hold. If investors that continue to flip their shares still get IPO allocations, this is also support against the pitchbook view. The pitchbook view of IPO allocations is tested by splitting allocated IPO investors into three groups. Group one investors flip their shares (sell all shares within the first month after the listing), group two investors hold their shares in the long term (hold some shares longer than six months after the listing) and group three investors are all remaining investors. To control for past buy-and-hold and flipping we add the number of times each investor has been, out of all past IPO participations, placed in group one or two. This will control for past buy-and-hold and past flipping activity in all previous IPO participations. The past buy-and-hold and the past flipping variables are calculated both on the total sample and on a bank by bank basis. The pitchbook view is then tested by regressing IPO allocations on the past level of buy-and-hold and flipping. If investors are allocated shares because they are buy-and-hold, the buy-and-hold variable will be positively related to IPO allocations and the flipping variable will be negatively related to IPO allocations. Since the views are not mutually exclusive, it is possible that both buy-and-hold and stock-trading commission are important for IPO allocations. Because of this, past buy-and-hold and past flipping are included as control variables when testing the other views as well.

43 The rent seeking view and the academic view can be tested in one specific or several unrelated IPOs. The buy-and-hold view should, however, be tested in several related IPOs. The key of the buy-and-hold view is that investors that provide this service over time are allocated shares over time. Investors that hold shares in the long run are rewarded with more shares in the future. Investors that sell shares early are punished by no future allocations. 44 Flipping investors are normally meant to be investors that sell their allocated shares during the first day of trading (Krigman, Shaw and Womack, 1999). We really want to test how selling shares early affects future allocations. It is therefore more accurate for our analysis to include all shareholders that sell their shares within the first month as flipping investors.

49 3.3.3 The academic view of IPO allocations The idea of Benveniste and Spindt (1989) is that investment banks meet with informed investors to price the IPO shares. Investment banks use investor bids to build a demand curve of the company shares. Investors are rewarded for their pricing service with IPO allocations that are underpriced on average. The academic view is controlled for by including a dummy variable that takes the value of one for all professional investors (financial institutions). It is possible that other investors, like non-financial institutions or retail investors, are participating in pricing of the shares, but it is not expected that this is very common. Investment banks are more likely to meet with financial institutions when they price IPO shares. To test the academic view more directly we proceed in the direction of Ljungqvist and Wilhelm (2002). The absolute percentage change from the midpoint in the initial pricing range to the actual offer price is used as the measure of pricing information. This measure should change when more pricing information is collected. The percentage change from the midpoint in the pricing range to the offer price is regressed on the combined allocation percentage to financial institutions and a set of control variables. If there is pricing information, the financial institution allocation percentage will be related to the percentage change in the pricing range. When financial institutions are allocated IPO shares, there should be a significant effect on the price. This analysis can, however, only be performed at the company level on the 71 IPOs that are priced through book-building.

3.4 Data There are 403 new listings on the OSE in the period January 1993 to September 2007. In total, 193 of the 403 companies listed through private placements, cross listings, spin-offs to existing shareholders or directly without any offerings. There are 89 companies with no offerings to new shareholders. The remaining 210 companies listed through IPOs. Table 2 gives the annual distribution of IPOs on the OSE in the period 1993 to 2007. In 30 of the 210 IPOs we have obtained 24,308 IPO allocations. (In 155 additional IPOs we have obtained 162,384 IPO allocations that might be contaminated with after- listing trading. The 162,384 IPO allocations are used in robustness testing). One listing requirement on the OSE is that all shareholders must be registered in the Norwegian Central Depository (the VPS) before the listing. The number of shares owned by each investor must be given to the VPS before any company can list publicly. This database is 100% accurate, as it is not possible to list otherwise. The VPS database includes all shareholders in all companies that are publicly listed or intend to list publicly. This database is used to obtain the IPO allocations by taking the difference in company ownership before and after allocation dates.45 Only IPO allocations to new shareholders are investigated.

45 In 16 of the 210 IPOs it has not been possible to calculate IPO allocations from the ownership data. These companies are listed in the database (VPS) in the same month as the listing month. These companies are therefore removed from the sample. In three companies there is missing allocation data, and in four companies it has not been possible to locate the pricing information (no oﬀer price). These IPOs are therefore not included in the analysis. There are three privatizations in the period that are removed. The ﬁnal sample is 185 IPOs with allocation and pricing data. 210 IPO companies - 16 companies that list in both VPS, OSE and IPO in the same month - 3 privatizations - 2 missing VPS data - 4 missing prospectus and newspaper articles on pricing = 185 companies.

50 More allocations to existing shareholders are removed. Allocation dates are collected from the IPO listing prospectuses. Some companies list in the VPS database in years before the listing. Other companies list in the VPS as part of the listing process. The number of shares by each investor ID is observed at the end of each month. All companies list in the VPS, sell shares in the IPO and list on the OSE. Allocations, by investor ID, are calculated as the difference in company share holdings before and after allocation dates. In terms of IPO allocations there are three dates that are important in the listing process. When companies list in the VPS ownership database, when companies distribute shares in the IPO and when companies list on the OSE influence IPO allocations. Companies do this process in different orders. This leads to three different levels of detail in the obtained IPO allocations. All ownership is observed on a monthly level, so if more than one listing event is performed within the same calendar month we are not able to distinguish between the events. Figure 3 gives a detailed description of how the IPO allocations are obtained for the different company groups. There are 15 savings banks (PCC list) out of the 210 IPOs on the OSE in the sample period. In total, 14 and seven of these savings banks are in the 155 IPOs with allocation data and in the 30 exact sample respectively. These banks are owned by the bank guarantee fund before they are publicly listed. All results remain unchanged if the banks are included or not.

3.4.1 IPO allocations Group one companies list their ownership records in the VPS database in good time before the IPO. These companies also list on the OSE in the calendar month after the IPO. For these companies the IPO allocations are completely accurate. There are 24,308 IPO allocations in these 30 IPOs. Some of these allocations are the same investors that are allocated shares in more than one IPO. IPO allocations for group one companies are obtained as the end of IPO month company ownership minus the ownership prior to the IPO month ownership. The owners that are left are the IPO allocated investors.46

3.4.2 After-listing ownership Group two companies list in the VPS database in good time before the IPO, but these companies list on the OSE in the same calendar month as the IPO allocation month. These companies have IPO allocations that include the actual IPO allocations and some after-listing trading (150 companies out of 185). The IPO allocations for these companies include the actual IPO allocations and between one and 30 days of after-listing trading. IPO allocations for group two companies are calculated as the listing month (and IPO allocation month) end of month company ownership minus the company ownership prior to the listing month. The investor holdings that are left are the allocated investors and the investors that have purchased shares in the period between the listing day and the end of the listing month.47 46 Over the counter (OTC) trading in the IPO allocation month will be treated as IPO allocations. It is, however, not expected that OTC trading will be a big issue because few investors are likely to trade shares in this period. Few IPO allocated investors are likely to sell their allocation and potentially lose out on the ﬁrst day return. The average time between the IPO allocation and listing is less than two weeks in the 30 IPOs. 47 Group two company IPO allocations includes some after-listing trading, but it is expected that most of these allocations are actual IPO allocations. If past trading activity is important for current allocations,

51 Group three companies list in the VPS database in the same month as the IPO allocation month. These IPO allocations does not include any after-listing trading, but they include existing owners who have not sold their shares (5 companies). IPO allocations for these companies are calculated as the end of listing in the VPS month ownership (and IPO allocation month ownership). Previous owners for these companies are not removed. Group two and three companies are used in robustness testing.

3.4.3 Variable description Company characteristics and the aggregate distribution of allocations between the different investor groups are given in Table 3. Market value is the total market value in USD at the listing date of the IPO company. This is calculated as the number of outstanding shares times the first day closing price. Book/Market is the book to market ratio of the IPO company at the listing date. This is calculated as the book value of equity, after the IPO, divided by the market value. Offer price is the actual offer prices in USD reported in the listing prospectuses or in the newspapers after the listing. VC dummy is a dummy variable that takes the value of one for companies with venture capital baking. High-tech dummy is a dummy variable that takes the value of one for IT -companies. Year dummy are dummy variables for each of the 15 years in the sample period. Company dummy are dummy variables for each of the 185 companies in the sample. Lead manager IPOs is the number of times the lead manager has been lead in the sample period. There are 32 different mangers in the 185 IPOs. There is one big manager that underwrites 23 out of the 185 IPOs. The ten biggest managers underwrite 144 of the 185 IPOs. There are 14 different managers that underwrite the 30 sample IPOs. Lead manager market share is the market share of the lead manager. This is calculated by the percentage market capitalization of the companies taken public out of total in the sample. Investor characteristics, for the individual investors on the OSE in the period 1993 to 2007, are described in Table 4. The dependent variable (Allocated shares/shares issued) is allocated shares to each investor divided by the total number of shares issued in the IPO. This is the same dependent variable as in Reuter (2006). The all IPO sample of 190,504 IPO allocations (in 185 IPOs) is trimmed at 1% to 186,694 allocations. This has no effect on results. This is done to remove the most extreme IPO allocations. Commission is the accumulated stock-trading commission generated by the investors in the two years before the IPO allocation dates.48 Commission is calculated as the monthly portfolio turnover times the share prices and a fixed percentage commission rate.49 Commission this will be reflected in the data even if the data includes some after-listing trading. This is especially true for the past buy-and-hold trading variables. Buy-and-hold investors will not sell their allocated shares, so if past buy-and-hold behavior is important for future IPO allocations this will be observed in the group two IPO allocations also. Group two IPO allocations can, however, not be used to reject that flipping investors are not punished for selling shares early. This is because some of the flipping investors are lost in the way the group two IPO allocations are obtained. 48 Commissions are generated from monthly data and not daily data. Because of this it is possible that commission trading takes place even if we are not able to find it in our data set. If commissions are generated from daily buy and sell orders in the same shares, then we are not able to detect this. It is also possible that some investors pay higher commission rates to get allocations. This should, however, be discovered in auditing. 49 We construct two separate data sets. In the first data set we obtain the allocated shares in the IPOs. The second data set is constructed by using the allocated shares in the first data set. For all allocated investors we collect the portfolio of publicly traded shares on OSE. We collect the change in the monthly portfolio ownership for each investor and this is multiplied with the correct market stock prices and the standard commission rates. The average commission rate offered by the 11 biggest internet share trading

52 is calculated as buy generated commissions only.50 Only commission by investors that do at least one trade in each of the four six months periods before the IPO is included. This has no effect on results as most investors trade in all four periods. This is done to remove investors that buy a large block of a company in one period without trading in the other periods. Generated commission below the minimum rate is replaced by the fixed minimum fee for one transaction ($15). The non-negative underpricing dummy is a dummy variable that takes the value of one for all IPOs with zero or positive underpricing. The variable commission*D is commission times the non-negative underpricing dummy. This variable is used to test if generated commission is more important for allocations in IPOs with a non-negative underpricing. Portfolio value is the portfolio value, in million USD, for each allocated investor at 31.12.xx in the year before the IPO allocation date. Previous IPOs is the accumulated number of past IPO participations by investors divided by the accumulated IPO number in the sample. This is used to measure how many IPOs, out of all possible, each investor has participated in. Previous buy-and-hold is the accumulated previous number of times the allocated investor has been a buy-and- hold investor divided by all previous IPO participations. This is the number of times, out of all previous IPO participations, an investor has held some IPO allocated shares for more than six months. Previous flipping is the accumulated number of times an investor has flipped previous IPOs divided by all previous IPO participations. Flipping is when all shares are sold within one month after a listing. This is the number of times, out of all previous IPO participations, the investor has held all IPO allocated shares for less than one month. Held cold IPO dummy is a dummy variable that takes the value of one if the IPO has a positive underpricing and the investor is allocated shares in a previous IPO with a negative underpricing. This variable is used to test if investors receive shares in hot IPOs because they accepted allocations in past cold IPOs. The Previous IPOs, Previous buy-and-hold, Previous flipping and Held cold IPO dummy are calculated on all 185 IPOs when allocations of the 30 exact IPOs are studied separately. The variables are also recalculated on a bank by bank basis when the most active bank is studied separately. Financial institution dummy is a dummy variable that takes the value of one for investors that are either Norwegian or foreign financial institutions.

3.5 Empirical results The main empirical result is that there is a strong and robust relationship between stock- trading commission generated in the period before IPO allocations and the number of shares allocated in IPOs. This is true for all investor types (retail and institutions). There is no consistent relationship between previous IPO share holding periods and current IPO allocations. There is also not more change in the pricing range of book-built IPOs when more shares are allocated to financial institutions (or institutions in general). It is concluded that IPO shares are allocated to the investors that generate the most stock- trading commission before the IPO allocations. companies in Norway is 0.075%. Some investors are likely to buy shares directly from their broker at a higher commission rate. We use the commission rate of 0.075% for all investors. The final number is the monthly commissions paid by each investor. The commission generated in each specific IPO is removed. 50 Sell generated commission variables are also related to IPO allocations. Only buy generated commission in the 24 month period before the IPOs is used as commission This is to avoid any issues related to sell generated commissions from rebalancing portfolios before IPOs. The results are the same when different measures of commission are used.

53 3.5.1 The rent seeking view of IPO allocations From Table 5 it can be seen that there is a positive relationship between generated stock- trading commission and IPO allocations in the 30 IPO sample (group one companies). IPO allocated shares, scaled by total shares issued in the IPOs, is regressed on the accumulated stock-trading commission in the 24 month period before the IPO allocation and a set of control variables. The level of generated commission is highly related to the number of allocated shares. This result control for investor size (measured by investor portfolio value), investor past trading behavior (past buy-and-hold, past flipping, past IPO participations and past accepted cold IPO allocations), investor type (financial institution or not), company fixed effects, year fixed effects and company specific variables. The results are statistically significant even if the sample size in all regressions is very large. All significance levels are correspondingly large to the sample sizes.51 The results are also economically significant. The point estimate for stock-trading commission is about 0.1 for retail investors (and 0.05 for institutional investors). If stock-trading commission is increased by 10% for retail investors, the allocation percentage is increased with one percent, see Wooldridge (2003). Stock-trading commission is also calculated for only IPOs with a non-negative underpricing. E.g. Stock-trading commission is multiplied with an interaction dummy variable that takes the value of one for IPOs with a non-negative underpricing. This new variable, Commission*D, is used to test if IPO shares are mainly allocated to investors with a high level of commission when IPOs are underpriced. The Commission*D is not always significant. This means that high commission investors are allocated more shares in general and not only in underpriced IPOs.52 Investors that generate high commission rates are likely to be preferred when they apply for IPO shares regardless of the expected underpricing. It is likely that high stock-trading commission will place investors on the A, B and C lists of the banks. When investors from these lists apply for shares, they are likely to be allocated shares. It seems unlikely that investment banks will discourage investors from accepting IPO allocations even if the issues may fall in price after the listing (even if the investors are from the A, B or C list). IPO allocations are also studied on the sub groups only retail investors and only institutional investors. The results remain unchanged. In Table 6 IPO allocations in group two and three are also included in the analysis. From Table 6 it can be seen that there is a positive relationship between generated stock- trading commission and IPO allocations for all IPOs. The results remain unchanged when the IPOs that might be contaminated by after-listing trading are included in the analysis. The number of IPO participations by each investor is also regressed on the stock-trading commission. The results are highly significant and explanatory. There is a strong relationship between generated stock-trading commission and the number of investor IPO participations. This means that investors that generate more commission are also participating in more IPOs than investors that generate less commission (not reported).

51 Even if the sample sizes are reduced by a large factor and the corresponding t —statisticsare reduced by the square root of this factor, the ﬁndings are still signiﬁcant; see Kecskes, Michaely and Womack, 2010. 52 Investors are, however, not likely to always know if IPOs will be under or overpriced. Investors are likely to apply for the shares they want. It is not certain that it is always the expected underpricing that drives the IPO application. If this was the case, there would be no IPO applicants in overpriced issues. Investors are likely to apply for shares in some issues that will fall in price after the listing. It is also likely that investment banks will allocate to investors that generate high commission rates when they apply for IPO shares.

54 3.5.2 The pitchbook view of IPO allocations The pitchbook view of IPO allocations is controlled for by including the past number of times investors have been buy-and-hold or flipping, out of past IPO participations, in all regressions. The pitchbook view argue that IPO shares are allocated to investors that are expected to be long term buy-and-hold investors. Buy-and-hold investors will create long term price stabilization of the IPO shares. Long term buy-and-hold investors can develop a relationship with investment banks and then be rewarded with future IPO allocations in return for previous buy-and-hold services. The long term investors will hold their shares to avoid being blacklisted in future IPOs. From Table 5 (the exact IPO allocations) it can be seen that the number of times an investor has been buy-and-hold in the past is negatively related or unrelated to current IPO allocations. For retail investors there is actually a positive relationship between past flipping activity and current IPO allocations. In Table 6 (all IPO allocations) the exact same results appear. This indicates that there is no or limited IPO allocations to buy-and-hold investors. The 30 exact IPOs are underwritten by several different investment banks. The same is true when all 185 IPOs are studied together. It is possible that this is causing the results. In many listing prospectuses there are two to three participating investment banks. It is then assumed that the bank that is mentioned first on the left side on the cover page of the listing prospectus is the lead investment bank. The single most active bank is the lead underwriter in 23 (out of 185) IPOs in the sample period. To study the pitchbook view it is necessary to also study this sample separately.53 From Table 7 it can be seen that there is not allocations to buy-and-hold investors for the most active bank either. There is no significant relationship between previous holding periods and future IPO allocations when only the single most active bank is studied separately. This is the exact same result as in Table 5 and Table 6. In Table 7 (regression 2 and 4) investors are also classified as only buy-and-hold investors if they have never been flipping investors in the past. (E.g. An investors that has a positive value for flipping in the past will take a zero value for the buy-and-hold by definition). The results remain unchanged. We are not able to detect a positive relationship between long holding periods and IPO allocations. It is concluded that the pitchbook view is not a likely reason behind IPO allocations.

3.5.3 The academic view of IPO allocations The academic view of IPO allocations is controlled for by including a dummy variable that takes the value of one for the expected pricing investors (ﬁnancial institutions) in all regressions. In Table 8 the academic view is tested more directly. In Table 8 the percentage price revision in book-built IPOs is regressed on the allocation percentage to ﬁnancial institutions and a set of control variables. This is similar to Ljungqvist and

53 Most of the IPOs of the very active banks are of the group two IPO allocations. This means that these IPO allocations includes from one to 30 days of aftermarket trading. We argue that this is of smaller importance when we study the pitchbook view, as this view argues that the investors hold their shares in the long run. The IPO allocations from group two will include the long term buy-and-hold investors if they are really buy-and-hold investors. In the sample it is observed if investors that hold shares in the long run are allocated more shares in future IPOs. The only problem is that some investors may buy shares after the listing and then hold these shares in the long run. These investors will be treated as buy-and-hold investors in the data, but they will not be awarded with future shares. It is expected that there will be less of this type of investors than actual buy-and-hold if buy-and-hold is an accurate view. Group two IPOs allocations should therefore detect any IPO allocations in return for past buy-and-hold.

55 Wilhelm (2002) that regress the percentage price revision on the percentage IPO allocation between institutional and retail investors. Ljungqvist and Wilhelm (2002) show that IPO allocations to institutional investors have a significant impact on price revisions. From Table 8 it can be seen that IPO allocations to financial institutions have no impact on the percentage price revision in our sample. IPO allocations to financial institutions is actually negatively related to changes in the offer price in the book-building period. This indicates that there is no price information collected from financial institutions. The same result is found when total allocations to institutional investors is investigated separately. The sample size is, however, very low with only 71 book-built IPOs in the sample.

3.5.4 Robustness As robustness we also test if there is a relationship between share ownership right after new listings and generated stock-trading commission for companies with no IPO. There are 89 companies with a suffi cient share spread and equity value to list directly at the OSE without conducting an IPO first. These 89 straight listings are used as a comparable sample. From Table 9 it can be seen that there is a relationship between stock-trading commission and share holdings after the listing in non-IPO companies also, but that this relationship is weaker than for IPOs. Investor stock-trading commission is multiplied with a dummy variable that takes the value of one for the IPO companies. After-listing ownership for both IPO companies and non-IPO companies is regressed on stock-trading commission. The coeffi cient for the relation between stock-trading commission and IPO allocations is many times greater in IPOs than in non-IPOs. It is concluded that the relationship between after-listing share ownership and stock-trading commission is driven by IPO allocations. In Table 9 the IPO allocations are also trimmed at 1%. In Table 10 the allocated investors are matched one for one with a non allocated investor in a Tobit regression. These IPO allocations are also trimmed at the 1% level. In total 9,498 (out of 24,308 investors) are first observed in the data with their IPO allocation. These investors have no previous stock-trading commission, past trading or portfolio size. The remaining investors are matched one for one with a non-allocated investor with no IPO participations in the last 12 months on investor type, investor country and number of shares in the portfolio. Among these investors the investor with the closest portfolio size is selected as the matching investor. The allocation percentage is then regressed on stock- trading commission and the control variables in a Tobit regression. The matching investors take the value of zero for (Allocated shares/shares issued) because they are not allocated IPO shares. We do not know if the matching investors applied for shares or not, but the regressions in Table 10 show that the matching investors generated less stock-trading commission than the allocated investors before the allocations. This further indicates that the stock-trading commission was generated to receive IPO allocations. The level of stock-trading commission is highly related to IPO allocations. This show that investors that are allocated IPO shares generate more stock-trading commission than investors that are not allocated IPO shares (matching on investor type, investor country and portfolio value).

3.6 Conclusion The main ﬁnding of the paper is that there is a strong and robust relationship between stock-trading commission generated by investors before IPO allocations and the number

56 of shares allocated in IPOs. The investors that generate the most stock-trading commission are also allocated the most IPO shares. This result is consistent for all investor types, in all IPOs and in all sample years. This result control for the portfolio value of the allocated investors, past trading behavior (the pitchbook view) and investor types (information gathering view). The result is also robust to companies that do not conduct IPOs and investors who are not allocated IPO shares. The meaning of this result is that there is a strong indication that investors are able to buy IPO allocations with stock-trading commission. The investors that are the most profitable clients, for the investment banks, are rewarded with the most IPO allocations. It can be argued that investors that trade more are also likely to apply for more IPO shares. The IPOs are, however, on average highly oversubscribed, so some investors are given more IPO allocations than other investors. We show that the investors that generate the most stock-trading commission are allocated the IPO shares. There is no evidence that support the information gathering view. There is not a bigger change in the percentage price revision from the midpoint in the pricing range to the offer price when financial institutions, or institutions in general, are allocated more shares. The sample size for the information gathering view is, however, too small to make any meaningful inferences. There is also no support for the pitchbook view. There is no detectable relationship between past IPO share holding periods and current IPO allocations. Investors that hold shares in the long run are not allocated more future IPO shares. This is also true when IPOs are studied on a bank by bank basis. Some investors are also able to obtain IPO allocations even if they repeatedly flip their shares. The conclusion is that more IPO shares are allocated to investors that generate more stock-trading commission. IPO shares are allocated based on the rent seeking view of IPO allocations. This finding is consistent with Reuter (2006) and Nimalendran, Ritter and Zhang (2007) in that IPO shares are allocated in return for stock-trading commission. A main contribution to the previous literature is that we are able to combine all existing views on IPO allocations in the same data set. We rank the views as the most to the least important view. This has not been possible to do before. There is strong evidence supporting the rent seeking view. There is no evidence supporting the academic view or the pitchbook view when controlling for the rent seeking view. There are also some practical implications of the study. Investors should be able to increase IPO allocations by increasing their stock-trading commission before IPOs. Investors can also be able to increase IPO allocations by directing trades to investment banks that underwrite many IPOs. There should also be more regulatory investigations into IPO allocation practices. It seems like the exchange of IPO allocations with stock-trading commission is a widespread practice. There are some limitations to the study. It is not observed that stock-trading commission is paid from the allocated investor to the investment bank. It is only observed that the commission is generated. Commission is also calculated based on monthly data. This is likely to underestimates commission. For future research it would be interesting to study stock-trading commission that is paid directly to the investment bank for all the allocated investors on a daily basis.

57 References

[1] Aggarwal, Reena, 2002, Allocation of initial public oﬀerings and ﬂipping activity, Journal of Financial Economics 68, 111-135

[2] Benveniste, Lawrence M. and Paul A. Spindt, 1989, How investment bankers determine the oﬀer price and allocation of new issues, Journal of Financial Economics 24, 343-361.

[3] Binay, Murat M., Vladimir A. Gatchev and Christo A. Pirinsky, 2006, The role of Underwriter-Investor Relationships in the IPO Process, The Journal of Financial and Quantitative analysis 42, 785-809.

[4] Bubna, Amit and Nagpurnanand R. Prabhala, 2007, When Bookbuilding Meets IPOs, AFA 2008 New Orleans Meetings Paper Available at SSRN: http://ssrn.com/abstract=972757

[5] Cliﬀ, Michael T. and David J. Denis, 2004, Do Initial Public Oﬀering Firms Purchase Analyst Coverage with Underpricing?, Journal of Finance 6, 2871-2901.

[6] Cornelli, Francesca and David Goldreich, 2001, Bookbuilding and strategic allocation, Journal of Finance 56, 2337-2369.

[7] Cornelli, Francesca and David Goldreich, 2003, Bookbuilding: How informative is the order book?, Journal of Finance 58, 1415-1443.

[8] Derrien, François and Kent L. Womack, 2003, Auctions vs. Bookbuilding and the Control of Underpricing in Hot IPO Markets, The Review of Financial Studies 1, 31- 61.

[9] Eckbo, Espen B. and Øyvind Norli, 2005, Liquidity risk, leverage and long-run IPO returns, Journal of Corporate Finance 11, 1-35.

[10] Fjesme, Sturla Lyngnes, 2011, Laddering in Initial Public Oﬀering Allocations, Work- ing paper, Norwegian Business School (BI).

[11] Hao, Qing (Grace), 2007, Laddering in Initial Public Oﬀerings, Journal of Financial Economics, 85, 102-122

[12] Jenkinson, Tim and Howard Jones, 2004, Bids and Allocations in European IPO Bookbuilding, Journal of Finance 59, 2309-2338.

[13] Kecskes, Ambrus, Roni Michaely and Kent Womack, 2010, What drives the Value of Analysts Recommendations: Earnings Estimates or Discount Rate Estimates?, Work- ing paper Cornell University.

[14] Krigman, Laurie, Wayne H. Shaw and Kent L. Womack, 1999, The persistence of IPO mispricing and the predictive power of ﬂipping, Journal of Finance 54, 1015-1044.

[15] Liu, Xiaoding and Jay R. Ritter, 2009, The economic consequences of IPO spinning, Forthcoming in Review of Financial Studies.

[16] Ljungqvist, Alexander P. and William J. Wilhelm, 2002, IPO allocations: discriminatory or discretionary?, Journal of Financial Economics 65,167-201.

58 [17] Loughran, Tim and, Jay Ritter, 2004, Why has IPO underpricing changed over time?, Financial Management 33, 5-37.

[18] Nimalendran, M., Jay R. Ritter, and Donghang Zhang, 2007, Do today’strades aﬀect tomorrows IPO allocations?, Journal of Financial Economics 84, 87-109.

[19] Pulliam, Susan and Randall Smith, Linux Deal is focus of IPO-Commission Probe, The Wall Street Journal, December 12, 2000.

[20] Reuter, Jonathan, 2006, Are IPO allocations for sale? Evidence from Mutual Funds, Journal of Finance 61, 2289-2324.

[21] Ritter, Jay, 1991, The long run performance of Initial Public Oﬀerings, Journal of Finance 46, 3-27.

[22] – – , Jay R. and Ivo Welch, 2002, A review of IPO activity, pricing and allocations, Journal of Finance 57, 1795-1828.

[23] – – , Jay R., 2003, Diﬀerences between European and American IPO markets, Eu- ropean Financial Management 9, 421-434.

[24] Rogers, W. H.,1993, Regression standard errors in clustered samples, Stata Technical Bulletin 13, 19-23. Reprinted in Stata Technical Bulletin Reprints 3, 88—94.

[25] White, Halbert, 1980, A heteroskedastic-consistent covariance matrix estimator and a direct test of heteroscedasticity, Econometrica 48, 817-838

[26] Wooldridge, Jeﬀrey M., 2003, Introductory Econometrics A Modern Approach (Tsinghua, Tsinghua University Press, South-Western College Publishing Thompson Learning).

59 Table 1 Related Empirical Papers

Rent seeking view Reuter(2006) Positiverelationshipbetweencommissionand holdings of IPO shares after new listings

Nimalendran,Ritter and Zhang (2006) Positive relationship between money on table and trading volume in liquid shares

LiuandRitter(2010) FindevidenceofIPOspinning

CliﬀandDenis(2004) Findevidenceofanalystconﬂictofinterest

Academic view

CornelliandGoldreich(2001) Regularlyparticipating,largebidanddomestic participants are favored in allocations

CornelliandGoldreich(2003) Bidsfromlarge,frequentbiddersthat include a limit price aﬀect the issue price

LjungqvistandWilhelm(2002) Increasedinstitutionalallocationsresultsin higher oﬀer price deviations from the midpoints of the book-building pricing ranges

Binay,GatchevandPirinsky(2007) Underwritersfavorinstitutionsthey have previously worked with

BubnaandPrabhala(2007) Book-buildinganddiscretioninallocation enhances pre-market price discovery

JenkinsonandJones(2004) Findevidenceagainstthe academic view Pitchbook view Aggarwal(2002) Institutionsﬂipmorethanretailinvestors (Evidence against the pitchbook view)

JenkinsonandJones(2004) Findevidenceinfavor of the pitchbook view

60 Table 2 The Number of Initial Public Oﬀerings on the Oslo Stock Exchange

The column labeled "IPOs" lists the number of Initial Public Oﬀerings on the Oslo Stock Exchange in the sample period. The column labeled "Data" indicates the IPOs with allocation data. The column labeled "Prospectus" lists the IPOs where we have been able to locate the listing prospectus. The column labeled "Sample" lists the 30 sample IPOs. The columns labeled "Value of shares" list the annually aggregate million USD values of shares sold in the IPOs with listing prospectus. "All", "New" and "Secondary" indicates the value of all shares, only newly issued shares and shares sold by existing shareholders respectively. The columns labeled "P" and "S" are the annual aggregated USD million value of shares sold in the IPOs with prospectuses and in the 30 IPO sample respectively. Value of shares sold is reported in USD using a USD/NOK exchange rate of 0.1792. The sample period is January 1993 through September 2007.

NumberofIPOs ValueofsharesUSD All New Secondary Year IPOs Data Prospectus Sample P S P S P S 1993119 7 541 539 2 1994 18 12 11 5 626 392 218142 409250 1995 18 14 11 3 516 47 113 47 403 1996 14 12 8 3 146 89 65 9 8180 1997 30 26 20 9 988 229 516 20 472208 1998 15 11 11 2 233 95 19094 43 1 199944 4 60 31 29 2000 12 12 11 2 839 112 765 90 74 22 2001 44 4 183 166 17 2002 2 2 2 2 70706464 66 200322 2 83 78 5 2004 14 14 14 1,605 1,319 287 2005 32 31 31 3 2,069 61 594 511,475 11 2006 18 17 17 2,730 2,237 493 2007 16 15 15 1 931 20 537 20 395

Total 210 185 168 30 11,621 1,130 7,431 550 4,190 580

61 Table 3 Summary Statistics of Firms Going Public on the Oslo Stock Exchange

Panel A reports the average percentage distributions of the IPO allocations. The exact sample includes the 30 IPOs with no after-listing trading. The total sample includes all 185 IPOs with IPO allocations. Panel B reports the IPO company characteristics for the 185 and the 30 companies. "Market value (Mill USD)" is the number of shares outstanding on the listing day times the first day closing price. "Offer price" is the USD IPO price in the listing prospectuses. "Book/Market" is the book value of equity after the IPO divided by the market value on the listing day. "VC backed dummy" is a dummy variable that takes the value of one if the company has venture capital backing. "High-tech dummy" is a dummy variable that takes the value of one for IT companies. "% change in pricing range" is change from the midpoint in the pricing range to the offer price in book-building IPOs. "Lead manager IPOs" is the average number of times the lead manager has been lead in the total 185 sample period. "Lead manager market share" is the market share of the lead manager. This is calculated by the percentage market capitalization of the companies taken public out of total in the sample. USD values are calculated from a USD/NOK exchange rate of 0.1792. IPO allocations are trimmed at 1%.

Exactallocations FullSample Variable N Mean Std.Dev Median N Mean Std.Dev Median A. Average Percent Allocation Retail% 30 39.9% 19.5% 38.3% 185 41.9% 23.6% 40.4% Norwegiannonﬁnancial% 30 30.3% 16.6% 30.2% 185 24.5% 14.8% 23.2% Norwegianﬁnancial% 30 16.4% 16.6% 10.9% 185 13.9% 12.2% 12.3% Foreign% 30 7.9% 8.7% 3.7% 185 15.3% 17.5% 6.7% Other% 30 5.5% 5.5% 3.2% 185 4.4% 5.2% 1.2%

B. IPO Characteristics Marketvalue(MillUSD) 30 $128 $135 $98.2 185 $291.2 $841.2 $101.4 OﬀerpriceUSD 30 $11.4 $7.8 $8.2 185 $9.1 $6.9 $7.2 Book/Market 30 0.77 1.4 0.29 185 0.63 0.82 0.42 VCbackeddummy 30 0.13 0.35 0 185 0.16 0.37 0 High-techdummy 30 0.07 0.25 0 185 0.11 0.32 0 %changepricingrange 30 0 0 0 71 8.3% 8.2% 7.3% LeadmanagerIPOs 30 9.8 7.5 7 185 6.3 7.6 3 Leadmanagermarketshare 30 4.6% 11.9% 1.3% 185 3.1% 8.2% 0.5%

62 Table 4 Summary Statistics on IPO Allocations and on Investors Trading

This table reports the summary statistics for the individual trading prior to the 30 sample IPOs and all 185 IPOs on the Oslo Stock Exchange in the period 1993 to 2007. Panel A reports the percentage share distribution between the investor groups. Panel B reports the investor characteristics. "Commission" is the accumulated commission generated in USD by the investors in the two years before the IPO allocation date. "Non negative underpricing dummy" takes the value of one for all IPOs with zero or positive underpricing. "Commission *D" is commission times the Non-negative underpricing dummy. "Portfolio value" is the portfolio value in million USD for each allocated investor at 31.12.xx in the year before the IPO allocation date. "Previous IPOs" is the accumulated previous IPO participations by the investors divided by the accumulated IPO number in the sample. "Previous buy-and-hold" is the accumulated previous number of times the allocated investor has been a buy and hold investor as a percent of all previous IPO participations. This is the number of times the investor has held some IPO allocated shares for more then six months in previous IPOs. "Previous ﬂipping" is the accumulated number of times the investor have ﬂipped previous IPOs as a percent of all previous IPO participations before the IPO allocation. Flipping is when all shares are sold within one month of the listing. "Held cold IPO dummy" takes the value of one if the IPO has a positive underpricing and the investor is allocated shares in a previous IPO with negative underpricing. USD values are calculated from a USD/NOK exchange rate of 0.1792. IPO allocations are trimmed at 1%.

ExactSample30IPOs FullSample185IPOs N Mean Std.Dev Median N Mean Std.Dev Median A. Average Allocation All% 24,308 0.06% 0.17% 0.01% 186,692 0.04% 0.14% 0.004% Retail% 19,999 0.03% 0.098% 0.01% 157,942 0.02% 0.08% 0.003% Norwegiannonﬁnancial% 2,336 0.18% 0.33% 0.04% 14,310 0.12% 0.26% 0.02% Norwegianﬁnancial% 524 0.39% 0.42% 0.21% 3,377 0.3% 0.41% 0.11% Foreigners% 937 0.1% 0.25% 0.03% 7,073 0.15% 0.32% 0.02% Others% 512 0.12% 0.25% 0.03% 3,990 0.07% 0.18% 0.01%

B. Investor Characteristics CommissionUSD 24,308 $2,328 $38,680 0 186,692 $5,406 $87,915 0 Non-neg.underpricingd. 24,308 0.73 0.44 1 186,692 0.86 0.35 1 Commission*D 24,308 $2,030 $37,974 0 186,692 $4,133 $74,132 0 PortfoliovaluemillionUSD 24,308 $2.01 $37.32 0 186,692 $3.43 $70.44 $0.003 PreviousIPOs 24,308 0.05 0.05 0.04 186,692 0.04 0.05 0.02 PreviousBuy-and-hold 24,308 0.19 0.36 0 186,692 0.21 0.37 0 PreviousFlipping 24,308 0.12 0.28 0 186,692 0.09 0.25 0 HeldcoldIPOdummy 24,308 0.1 0.3 0 186,692 0.12 0.33 0

63 Table 5 IPO Allocations and Generated Commission for the 30 Sample IPOs

This table reports the coeﬃ cients and heteroscedastic consistent t -statistics (errors adjusted for clus- tering across ﬁrms Rogers, 1993) in parentheses for the regressions with the number of allocated shares divided by the total number of shares issued in the IPO as the dependent variable. This is a standard OLS model. Only the 30 IPOs with exact allocations in the sample period September 1993 to January 2007 are included. All variables are as described in Table 3 and Table 4. Regression 1 includes all investors. Regression 2 includes only retail investors. Regression 3 includes only institutional investors. In Regression 4 and 5 the investors with zero in commission in the 24 month period before the new listings are dropped. In Regression 6 the savings banks (7) are removed. IPO allocations are trimmed at 1%.

Log (Allocated shares/shares issued) % Reg1 Reg2 Reg3 Reg4 Reg5 Reg6 Intercept 4.356 -16.1345 2.5336 -0.9762 -3.2502 5.0669 (19.3) (-48.0) (8.6) (-2.5) (-8.3) (48.5) Log(commission) 0.0949 0.0961 0.0539 0.081 0.0487 0.0962 (5.7) (5.8) (3.2) (7.9) (3.4) (7.4) Log(commission)*D -0.0714 -0.077 -0.0277 -0.0561 -0.0237 -0.0522 (-3.6) (-3.5) (-1.5) (-4.0) (-1.4) (-2.8) Non-negativeunderpricingD. 0.6562 2.7586 -1.1431 0.3261 -0.1382 -3.3122 (18.5) (3.5) (-21.7) (5.9) (-1.6) (-75.9) Log(portfoliovalue) 0.0513 0.0327 0.0668 0.035 0.0768 0.0404 (4.3) (3.5) (5.8) (4.1) (5.9) (3.5) PreviousIPOs 0.8032 -0.321 0.8488 -0.6907 0.4453 -0.1461 (1.0) (-0.7) (0.9) (-1.8) (0.4) (-0.2) Previousbuy-and-hold -0.1286 -0.0798 -0.03 -0.0711 -0.0119 -0.1544 (-2.7) (-2.3) (-0.4) (-1.6) (-0.1) (-3.1) Previousﬂipping 0.1446 0.2016 -0.2193 0.2568 -0.0388 0.1295 (2.4) (3.5) (-1.7) (4.3) (-0.4) (2.1) HeldcoldIPOdummy 0.0788 0.0087 -0.0106 -0.0005 0.0418 0.0506 (1.2) (0.2) (-0.1) (-0.0) (0.5) (0.6) Financialinstitutiondummy 1.8786 dropped 0.5319 dropped 0.502 1.1784 (10.2) (3.9) (3.6) (6.3) Log(marketvalue) -0.4328 0.4991 -0.3657 -0.2573 -0.0861 -0.5397 (-29.8) (30.2) (-20.5) (-12.0) (-4.8) (-59.1) BV/MVequity 0.316 0.5386 0.3979 0.3877 0.4329 dropped (28.1) (62.6) (51.7) (83.1) (32.3) Oﬀerprice 0.0069 -0.0103 0.0001 -0.0034 0.0063 0.0084 (14.0) (-35.3) (0.3) (-33.8) (13.5) (24.1) VCbackeddummy 1.6586 0.69 0.7532 -0.4467 0.3223 5.5484 (16.9) (9.6) (12.7) (-8.5) (2.7) (39.3) High-techdummy dropped 4.433 -1.7555 0.1653 dropped -7.8126 (201.3) (-17.5) (2.5) (-77.9) YearandCompanydummy yes yes yes yes yes yes Observations 24,308 19,999 4,309 10,207 2,876 16,593 AdjustedR-squared 42.4% 49.1% 34.6% 42.5% 31.5% 42% Investorgroup All Retail Institution Retail Institution All

64 Table 6 IPO Allocations and Generated Commission for All 185 IPOs

This table reports the coeﬃ cients and heteroscedastic consistent t -statistics (errors adjusted for clus- tering across ﬁrms Rogers, 1993) in parentheses for the regressions with the number of allocated shares divided by the total number of shares issued in the IPO as the dependent variable. This is a standard OLS model. All 185 IPOs in the sample period September 1993 to January 2007 are included. All variables are as described in Table 3 and Table 4. Regression 1 includes all investors. Regression 2 includes only retail investors. Regression 3 includes only institutional investors. In Regression 4 and 5 the investors with zero in commission in the 24 month period before the new listings are dropped. In Regression 6 the savings banks (14) are removed. IPO allocations are trimmed at 1%.

Log (Allocated shares/shares issued) % Reg1 Reg2 Reg3 Reg4 Reg5 Reg6 Intercept -4.185 -14.8485 -4.9007 -12.9069 -0.0259 -8.818 (-29.8) (-85.7) (-29.1) (-141.4) (-0.1) (-40.7) Log(commission) 0.0679 0.0384 0.0608 0.0392 0.0612 0.0674 (6.1) (4.3) (5.3) (6.3) (5.7) (5.9) Log(commission)*D 0.014 0.0079 0.02359 0.0031 0.0164 0.0187 (0.6) (0.5) (1.1) (0.3) (1.0) (0.8) Non-negativeunderpricingD. -4.1829 -1.2985 -0.9574 -1.2387 -0.4123 -1.14076 (-70.8) (-24.6) (-8.0) (-36.4) (-4.2) (-12.5) Log(portfoliovalue) 0.042 0.0285 0.057 0.0296 0.0678 0.0403 (13.4) (9.7) (13.3) (9.7) (11.6) (13.8) PreviousIPOs -1.1269 -1.067 -0.6539 -1.1515 -0.6968 -1.3911 (-1.6) (-2.5) (-0.9) (-3.7) (-1.5) (-2.0) Previousbuy-and-hold 0.0396 0.115 -0.1148 0.0445 -0.1463 0.0424 (0.3) (1.0) (-2.0) (0.5) (-3.9) (0.3) Previousﬂipping 0.2679 0.3299 -0.1543 0.3323 -0.1023 0.272 (3.0) (4.2) (-1.7) (5.4) (-1.3) (2.9) HeldcoldIPOdummy 0.0805 0.04 0.0486 0.0518 0.0673 0.0877 (2.4) (1.3) (1.0) (2.0) (1.4) (2.4) Financialinstitutiondummy 1.1915 dropped 0.6573 dropped 0.6003 1.8942 (18.2) (8.0) (8.3) (16.2) Log(marketvalue) 0.0987 0.4584 -0.0064 0.4066 -0.0336 0.1266 (21.0) (48.0) (-2.6) (126.7) (-12.6) (15.3) BV/MVequity -0.0094 0.0979 0.1429 0.0984 0.1518 2.1588 (-0.9) (8.2) (22.1) (10.5) (12.2) (31.2) Oﬀerprice 0.0055 -0.0138 0.0085 -0.0125 -0.0025 -0.001 (51.6) (-39.3) (40.7) (-76.0) (-6.5) (-5.8) VCbackeddummy 1.6097 -1.9087 -0.2361 -2.2017 -0.6031 -0.3591 (18.5) (-322.9) (-6.0) (-258.2) (-56.1) (-23.4) High-techdummy -0.2622 2.2149 -1.0701 2.4697 -0.0991 0.2558 (-3.5) (69.0) (-11.4) (66.8) (-2.9) (8.1) YearandCompanydummy yes yes yes yes yes yes Observations 186,692 157,942 28,750 94,362 21,195 175,382 AdjustedR-squared 79.5% 84.2% 53.4% 83.3% 48.6% 79.5% Investorgroup All Retail Institution Retail Institution All

65 Table 7 IPO Allocations to Buy-And-Hold Investors

This table reports the coeﬃ cients and Clustered (Rogers, 1993) heteroscedasticity consistent t-statistics in parentheses for the regressions with the number of allocated shares divided by the total number of shares issued in the IPO as the dependent variable. All variables are as described in Table 3 and Table 4. All regressions are standard OLS models, and the sample period is from January 1993 to September 2007. Only the 23 IPOs by the most active bank in the sample period is investigated. Previous trading variables are only in past IPOs by the most active bank. Regression 1 and 2 includes all IPO allocations. Regression 3 and 4 includes only allocation in the 13 underpriced (hot) IPOs. Regression 2 and 4 includes only past buy-and-hold for investors who have never been ﬂipping investors before. IPO allocations are trimmed at 1%.

(Allocated shares/shares issued) % Reg1 Reg2 Reg3 Reg4 Intercept 0.0668 0.0668 0.0317 0.0315 (58.3) (57.7) (26.9) (26.1) Log(commission) 0.0038 0.0038 0.0007 0.0007 (8.5) (8.9) (3.5) (3.4) Log(commission)*D -0.0031 -0.0031 (-6.1) (-6.3) Non-negativeunderpricingD. -0.0036 -0.0004 (-0.2) (-0.2) Log(portfoliovalue) 0.0002 0.0002 0.0002 0.0002 (3.3) (3.4) (3.1) (3.2) PreviousIPOs -0.0016 0.0013 -0.0004 0.001 (-0.6) (0.4) (-0.2) (0.3) Previousbuy-and-hold 0.0003 -0.0004 00005 -0.0002 (0.4) (-0.4) (0.7) (-0.2) Previousﬂipping 0.0034 0.001 (1.3) (0.7) HeldcoldIPOdummy 0.001 0.0012 0,002 0.0022 (0.5) (0.7) (1.1) (1.2) Financialinstitutiondummy 0.0315 0.0314 0.0285 0.0285 (4.2) (4.3) (4.0) (4.0) Log(marketvalue) -0.0025 -0.0025 -0.0005 -0.0005 (-15.0) (-14.6) (-8.3) (-8.4) BV/MVequity 0.0002 0.0001 0.0002 0.0001 (0.6) (0.3) (0.3) (0.2) Oﬀerprice -0.002 -0.0002 -0.0003 -0.0003 (-16.8) (-16.3) (-65.2) (-63.6) VCbackeddummy -0.0283 -0.0284 dropped dropped (-40.8) (-43.9) High-techdummy 0.0101 0.0101 0.0068 0.007 (20.7) (20.1) (10.6) (10.6) YeardummyandCompanydummy yes yes yes yes Observations 67,795 67,795 63,539 63,539 AdjustedR-squared 33% 33% 20.4% 20.4% IncludedIPOs 22 22 13 13 66 Table 8 IPO Allocations in Return for Pricing Information

This table reports the coeﬃ cients and White (1980) heteroscedasticity consistent t -statistics in parentheses for the regressions with the absolute percentage change in the price revision as the dependent variable. This is a standard OLS model, and all book-built IPOs in the sample period from January 1993 to Septem- ber 2007 are included. All variables are described in Table 3 and Table 4. Regression 1 includes "Financial institution allocation %" in all book-built IPOs. Regression 2 includes "Institutional allocation %" in all book-built IPOs.

Absolute % price revision Reg 1 Reg 2 Intercept 17.7888 11.9459 (2.3) (1.1) Financialinstitutionallocation% -0.1046 (-2.0) Institutional allocation % -0.0022 (-0.0) Log (market value) -0.472 -0.2403 (-1.1) (-0.5) BV / MV equity -1.3576 -1.4011 (-0.5) (-0.5) Oﬀer price 0.016 0.0008 (0.5) (0.0) VC backed dummy 3.6374 4.1113 (1.5) (1.6) High-tech dummy 1.2246 0.2923 (0.5) (0.1) Observations 71 71 Adjusted R -squared 5.8% 0.4%

67 Table 9 Commission and Share holdings of Newly Listed Companies with No IPO

This table report the coeﬃ cients and Rogers (1993) clustered (on company) heteroscedasticity consistent t-statistics in parentheses for the regressions with the shares owned per investor at the end of the listing month divided by outstanding shares in the listed company as the dependent variable. There are 89 companies that list with no oﬀering to new shareholders. The regression is a standard OLS model, and the sample period is from January 1993 to September 2007. Regression 1 includes allocations and after-listing ownership in all 185 IPO companies and all 89 companies with no IPO. The dummy IPO takes that value of one for all investors in the 185 IPOs and zero for all the investors in the 89 non-IPO companies. "Log (commission) * Dummy IPO" is investor commission in all IPOs and zero commission in all 89 non-IPOs. IPO allocations and after-listing ownership are trimmed at 1%.

Variables (Sharesholdings/sharesoutstanding)% Reg 1 Intercept 0.1697 (7.0) Commission 0.00000 (7.3) Commission * Dummy IPO 0.0048 (4.7) Dummy IPO 0.1679 (370.1) Portfolio value 0.0000 (3.5) Previous IPOs 0.1208 (4.3) Previous buy-and-hold -0.00062 (-5.5) Previous ﬂipping 0.005 (2.6) Financial institution dummy 0.127 (11.4) Year and Company dummy yes Observations 374,584 Adjusted R -squared 22.6%

68 Table 10 Matching Allocated Investors with Non-Allocated Investors

This table reports the coeﬃ cients and Rogers (1993) heteroscedasticity consistent t -statistics in parentheses for the regressions with the number of allocated shares divided by the total number of shares issued in the IPO as the dependent variable. This is a standard Tobit regression. Allocated investors with previous trading are matched one for one with a non-allocated investor. The non-allocated investors takes a value of zero for "(Allocated shares/shares issued) %". IPO allocations are trimmed at 1%. Regression 1 includes all exact IPO allocations and matched investors that did not receive IPO allocations. Regression 2 includes only investors with a positive level of commission.

(Allocated shares/shares issued) % Reg 1 Reg 2 Intercept -0.1971 -0.2331 (-12.9) (-20.1) Log (commission) 0.0079 0.0106 (6.7) (7.8) Log(commission)*D -0.0038 -0.0037 (-2.1) (-1.6) Non-negativeunderpricingD. 0.0465 0.0174 (7.1) (1.2) Log(portfoliovalue) -0,0001 0.0003 (-0.4) (0.9) Previous IPOs 0.5768 0.5341 (5.5) (4.7) Previousbuy-and-hold -0.006 -0.0055 (5.5) (-1.2) Previous ﬂipping 0.0076 0.0124 (1.3) (1.8) HeldcoldIPOdummy -0.0152 -0.0138 (-3.4) (-3.2) Financialinstitutiondummy 0.0914 0.0747 (8.4) (6.6) Log (market value) 0.0077 0.0079 (10.4) (14.2) BV / MV equity 0.0136 0.0134 (14.8) (20.1) Oﬀer price -0.0001 0.0004 (-3.8) (15.9) VC backed dummy 0.0461 0.0334 (7.6) (24.1) High-tech dummy 0.2516 0.0268 (79.8) (12.4) Yeardummy,Companydummy yes yes Observations 38,973 27,774 Pseudo R -squared 16.9% 10.5%

69 Figure 1 The diﬀerent IPO Allocation Views

Ritter (2003) and Jenkinson and Jones (2004) argue that there are three views on how IPOs are allocated. First, is the academic view based on Benveniste and Spindt (1989). In this view investment banks allocate IPO shares to informed investors in return for true valuation and demand information. Second, is the pitchbook view where investment banks allocate shares to institutional investors that are likely to be buy-and-hold. Finally, is the rent seeking view where investment banks allocate shares to investors in return for some form of kickback.

The Rent seeking view of IPO allocations

Commission Sharesareallocatedtoinvestorsthatgenerate high levels of stock-trading commissions.

IPOspinning Sharesareallocatedtocompanyexecutivestoattract corporate business.

IPOladdering Sharesareallocatedtoinvestorsthatwillprovide after-listing share price support. Underpriced shares are allocated to investors that generate high stock-trading commissions.

Analystoverage Companiesacceptunderpricinginexchangeforfuture. research coverage. Underpriced shares are allocated to investors that generate high stock-trading commissions.

The pitchbook view of IPO allocations Shares are allocated to investors that are expected to be buy-and-hold. This will create long run price stability.

The academic view (information gathering) of IPO allocations Shares are allocated to investors that report true share values. Shares are exchanged with price information.

70 Figure 2 Timeline of the Listings on the Oslo Stock Exchange

Listing in the VPS is when the company list ownership records in the ownership database. This is when the ownership records are observed in the data the first time. Public Offering is when the companies distribute the allocated shares in the ownership database. The public offering is in most cases in the month before (30 exact IPOs) or the month of the listing (150 IPOs).

Timeline of the listing Company list shares in the VPS database Six months before the listing The company selects an investment bank

The initial meeting between company, investment bank and the OSE

Compliance report is ﬁnalized by the investment bank

The legal and accounting due diligence is performed

The formal application is submitted to the OSE

Prospectus is ﬁnalized and distributed

IPO shares are priced through meetings with investors

Onemonthbeforethelisting SharesaretransferredinthePublicOﬀering

Listing month Listing

71 Figure 3 Timeline of the IPO allocations for the diﬀerent groups

Listing in database is when the company list ownership records in the ownership database. This is when the ownership records are observed in the data the ﬁrst time. IPO allocation is when the companies distribute the allocated shares in the ownership database. Group 1 to 3 is the ordering of the group of detail in the allocations. Group 1 is 100% accurate IPO allocations. Group 2 IPO allocations includes one to 30 days of after-listing trading. Group 3 IPO allocations includes existing owners who have not sold all of their shares in the IPO. There are 30, 150 and 5 companies in group 1, 2 and 3 respectively.

Timelineofthelisting Sixmonthsbefore Onemonthbefore Listingmonth thelisting thelisting

Group1 Listingindatabase IPOallocation Listing

Group2 Listingindatabase IPOallocation Listing

Group3 Listingindatabase Listing IPO allocation

72 .

4 Initial Public Oﬀering or Initial Private Placement?

Sturla Lyngnes Fjesme54 BI Norwegian Business School

Øyvind Norli BI Norwegian Business School

Abstract

This paper studies the choice between an auction and a negotiation when selling a large fraction of a company. Using detailed data on ownership structure in 123 public offerings and 88 negotiated private placements, we show that negotiated private placements are much more common when there are significant private benefits of control. This finding supports the idea that a negotiated transaction allow the seller to extract more of the gains from trade when the gains from trade include private benefits.

JEL classiﬁcation: G24

Keywords: Private Placements; Public Oﬀerings; IPOs; Equity oﬀerings

54 We are grateful to "The Center for Corporate Governance Research (CCGR)" at BI Norwegian Business School for ﬁnancial support, to Øyvind Bøhren, François Derrien, and seminar participants at BI Norwegian Business School for valuable suggestions, and the Oslo Stock Exchange VPS for providing the data. All errors are our own. Corresponding author: BI Norwegian Business School, Nydalsveien 37, 0484 Oslo, Norway, E-mail address: [email protected], Telephone: +47-957-722-43

73 4.1 Introduction Stock exchanges have stringent rules on minimum equity levels and the minimum number of shareholders that are required to list publicly. Most private companies must issue equity to be able to meet these minimum requirements. Shares can either be sold in an IPO to a large group of dispersed investors or in a negotiated private placement to a small group of specialized investors. Most theoretical papers on equity offerings show that public offerings will almost always be preferred by the seller, so why some companies use private placements has been the focus of many empirical studies in finance. The research question addressed in this paper is whether private placements are used to transfer private benefits of control from the seller to the buyer. The new and unique data in this paper includes investor level ownership and audited financial statements in 88 private placements and 123 public offerings during their listing on the Oslo Stock Exchange (OSE) in the period 1993 to 2007. The main contribution of the paper is that we show a strong and robust relationship between private benefits of control before the initial equity offering and the use of private placements55. This suggests that sellers of a company use private placements to transfer private benefits of control to buyers. Private placements are used by family firms and firms with controlling owners before the offerings. Public offerings are used by companies with more dispersed ownership before the offerings. Companies that use private placements also have more block ownership after the listings. Public offerings reduce block ownership. The main implication of this finding is that companies with low private benefits of control should be sold in public offerings and companies with higher private benefits of control should be sold in private placements. The finding also have implications for research on auctions and negotiations. When auctions are structured like IPOs and there are large private benefits of control, the seller is likely to prefer a negotiation over an auction. Several papers have proposed explanations to the private placement choice made by some companies. Some papers argue that private placements are used to attract value creating investors such as monitoring or certification investors (Wruck, 1989; Hertzel and Smith, 1993). These investors ensure that companies are run optimal or put their stamp of approval on company valuations. Other papers suggests that private placements are used when buyers value private benefits of control (Zingales, 1995; Zingales, 1994; Zwiebel, 1995 and Damodaran, 2005). Most existing research on private equity offerings are on SEOs by publicly listed companies. The reason for this is likely to be that there are more available data on publicly listed companies. Only investigating public companies is problematic for this research question because this leaves out the major equity offerings taken place before the actual listing. Many of the companies that list on the OSE through private placements have follow-on public and employee offerings before the listing. This shows that private placements must often be used in connection with a follow-on offering to meet listing requirements. This also show that the private or public choice is not dictated by the minimum size listing requirements.

55 The agency problem investigated is between large owners and small owners. Large owners have a controlling benefit at small owners expense. Throughout the article, we mean the private benefit of controlling the firm enjoyed by the controlling/big shareholders at the expense of smaller owners when the term private benefit of control is used. Other agency problems, that we do not study, can for instance be between owners and managers in the firm.

74 Derrien and Kecskés (2007) show that many U.K. companies lists publicly without issuing equity and that these companies issue equity in a SEO after the listing. This two-stage listing is cheaper than the normal IPO. On the OSE there are only a limited number of companies that are allowed to use this two-stage process. In most listings on the OSE the offering is a requirement to list. The choice faced by most companies is not if there should be an offering before or after the listing. The choice is if the required offering should be public, private or to existing shareholders. Few companies have an existing shareholder base that can cover the offering in full. Listing rules require that there must be at least 500 owners to list on the main list of the OSE (100 at the Small and Medium Sized SMB/Axess list). Only 21 out of 403 companies have listings with only an offer to existing shareholders. Therefore, the main choice at the OSE is between a negotiated private placement and an IPO. This makes the OSE an ideal market to study the choice between IPOs and private placements. The remaining paper is organized as follows. Section 4.2 describes related literature. Section 4.3 describes the road to the listing. Section 4.4 describes predictions and testable implications. Section 4.5 and 4.6 describes the data set and the empirical results. Section 4.7 concludes.

4.2 Literature review There are many theoretical papers that study the equity sales process.56 Bulow and Klem- perer (1996, 2009) compare auctions to negotiations and sequential sales mechanisms.57 Bulow and Klemperer (1996) show that for a seller it is better to sell in an auction with (N+1) bidders than in a negotiation with N bidders. The seller should therefore focus on maximizing the number of bidders and not focus on finding a single bidder to nego- tiate with. The exception to this rule is when more information must be disclosed in the auction. When more information, that can possible reduce the future asset value for the final owner, is disclosed in the auction, it is possible that the negotiation is more profitable for the seller than the auction. Bulow and Klemperer (2009) show that buyers (usually) prefer to buy in a sequential sale (negotiation), and sellers (usually) prefer to sell in an auction. The exception to this finding is when the marginal revenue curve of the winner is very flat, there are many potential bidders and the bidder cost of obtaining value information is neither too high nor to low. French and McCormick (1984) find that negotiations should be used instead of auctions when there is an ongoing relationship between bidder and seller, there is a low asset value difference between bidder and seller, there is a low asset value difference between different bidders and the actual negotiation cost is low compared to auctions. Zingales (1995) propose that the buyer of a company can have a higher company value than the current owner from either an increase in the private benefits of control or an increase in the cash flow. By selling to dispersed shareholders the proceeds from the sale of cash flow rights are maximized. Through bargaining with a buyer the seller maximizes proceeds from the sale of control rights. Zingales (1994) argue that one of

56 Table 1 summarizes all related papers. 57 IPOs are not really open auctions, and private placements are not really negotiations in the exact same sense as used in all of the literature. There are, however, large similarities between IPOs and auctions and private placements and negotiations, and we therefore include a literature review on the auctions and negotiations literature. We also expect that our ﬁndings may have implications for research on auctions.

75 the most common areas of private benefits of control is dilution of minority property rights. This shows that there should be some smaller investors in the companies that use private placements. It is also argued that control is more valuable during proxy contests. Damodaran (2005) argues that the value of a block of shares comes from the ability to influence control by changing the way the business is currently run. Damodaran (2005) argues that block shares are sold at a premium compared to dispersed shares. Value of control can be calculated as the value of the firm assuming that it is optimally run minus the status quo value of the firm. Control of a firm does not necessarily require 51% of shares if the remaining shares are sold to a dispersed group of shareholders. Zwiebel (1995) investigates smaller block shares. It is argued that there are benefits of having blocks that are smaller than controlling stakes from partial benefits of control. Smaller block holders can join together and get control if desired. Private benefits of control can be the ability of owners, management or directors to dilute corporate funds for private benefits.58 Private benefits can also be synergies obtainable through mergers (during takeover contests opposing sides actively recruit block shareholders), favors by firms, access to inside information, perquisites of control and utility derived directly from power of control. Some firms, such as sports and communication firms, are likely to yield private benefits from the nature of their business. Stoughton and Zechner (1998) argue that IPOs are allocated to institutions to increase monitoring. Several empirical papers propose explanations to the private placement choice. Wruck (1989), later referred to as the monitoring hypothesis, show that active investors buy shares privately and monitor management. It is argued that monitoring will increase value by ensuring effi ciency and openness to value creating takeovers. The article investigates 128 private placements made by companies listed on NYSE and AMEX in the period 1979 to 1985. Hertzel and Smith (1993), later referred to as the certification hypotheses, argues that an informed investor buy large blocks of shares in private placements to put their stamp of approval on company valuations. The paper investigates 106 private placements made by smaller companies listed on NASDAQ in the period 1980 to 1987. It is concluded that certification is a likely reason behind private placements. Barclay et al. (2007) investigate if monitoring (Wruck, 1989) and certification (Hertzel and Smith, 1993) explains private placements by investigating 594 U.S. publicly traded firms in the period 1979 to 1997. The main finding is that private placements are often allocated to passive investors that help management keep control of the companies. This is proposed as the entrenchment hypothesis, and it is concluded that entrenchment is a more likely reason for private placements than monitoring or certification. Anshuman et al. (2010) propose the undervaluation hypothesis as appose to the monitoring, certification and entrenchment hypotheses. The undervaluation hypothesis is an extension of Myers and Majluf (1984), and the hypothesis propose that company management and insiders buy shares in their own company, through private placements, when they believe that the company is undervalued. The hypothesis is tested on a sample of 164 private placements in the Indian capital market in the period 2001 to 2009. It is concluded that private placements (to company insiders) can eliminate underinvestment, and the underinvestment hypothesis can explain the private placement choice after controlling for monitoring, certification and entrenchment. Wu (2003) investigates how information asymmetry and monitoring affects the company choice between public offerings and private placements. The data investigated is 728 public offerings and 360 private placements

58 In this paper we study private beneﬁts of control enjoyed by big owners through dilution of corporate funds.

76 made by high technology companies that have recently been publicly listed on NYSE, Nasdaq or AMEX. The main finding is that private placement companies have a higher information asymmetry than public offering companies. Private placement investors also do not monitor more than public offerings investors. Wu (2003) concludes that monitoring is not a likely reason behind private placements. Brennan and Franks (1997) investigate 67 U.K. IPOs and find that underpricing is used to ensure suffi cient oversubscription and rationing of shares. This is done by IPO company insiders to discriminate between shareholders and reduce block sizes. Brennan and Franks (1997) argue that underpricing is used to avoid block holder formations. Arugaslan,˘ Cook and Kieschnick (2004) investigate 3,441 U.S. IPOs. They find that determinants of initial returns, institutional share holdings and post- IPO likelihood of acquisition are not consistent with either Brennan and Franks (1997) or Stoughton and Zechner (1998). Arugaslan˘ et al. (2004) conclude that monitoring considerations are not important determinants of IPO underpricing. Cron- qvist and Nilsson (2005) investigate how Swedish publicly traded companies in the period 1986 to 1999 choose between rights offerings and private placements in SEOs. It is found that companies with much asymmetric information will choose private placements over rights offerings. Companies will choose private placements to current shareholders when asymmetric information is extreme. Companies also do private placements to new business partners. It is concluded that private placements can be used to reduce moral hazard, adverse selection costs and offset high issue cost. Boone and Mulherin (2007) investigate why not all firms are sold in competitive auctions. The investigated data includes 202 auctioned and 198 negotiated takeovers of U.S. public firms in the period 1989 to 1999. The main finding is that there is no difference in wealth effects of the target firms after negotiations and an auctions. Auctions does not increase revenue for the sellers. Boone and Mulherin (2008) investigate 145 auctioned and 163 negotiated takeovers by U.S. publicly traded bidders in the period 1989 to 1999. The paper test if the return to the winning bidder is related to the level of competition in the takeover market. It is assumed that there is a negative relationship between the number of bidders and the level of value uncertainty and the bidder return if the winners curse is true. The main finding is that there is no relationship between bidder return and competition. It is concluded that there is no winners curse in the corporate takeover market.

4.3 The road to the listing The listing process includes many formal requirements. These are dictated changes the private company must make to be allowed to list publicly. The private company must also make many decisions that are not formal requirements. The most notable, for this article, is if equity should be raised through an IPO or in a negotiated private placement.

4.3.1 The formal listing process The listing process at the OSE takes between eight and 14 weeks to complete.59 The private company must ﬁrst select an investment bank to help with the listing process. The company and the chosen investment bank then have a meeting with the board of the OSE to initiate the listing process. After this initial meeting the investment bank hires

59 The information about the listing process is obtained from the seminar “The road to the listing” November 3, 2009 by Deloitte Public Accountants and the Oslo Stock Exchange.

77 an accounting firm and a law firm to complete a financial and a legal due diligence of the private company. The investment bank then, assuming everything is in order, makes a compliance report that shows that the private company meet all formal requirements to list on the OSE. Four weeks after the initial meeting with the OSE, there is a meeting between the accounting firm, the law firm and the OSE. At this time, the formal application is handed in to the OSE by the investment bank. During the next four weeks, the investment bank completes the formal listing prospectus. The OSE use this time to go through the application. The company is then accepted or rejected to list on the OSE. About 80 to 90% of all companies are accepted. Most companies are, however, accepted to list with conditions. Most companies have to adjust before they are allowed to list publicly. There are two very common conditions to list. The first common condition is that the equity level must be increased. Companies must show that they have suffi cient equity to keep the company running for at least 12 months after the listing. It is not necessary with a positive cash flow as long as the company can run on equity for at least 12 months. Many companies on the OSE are shipping companies with high cash outflows around the listing date and high cash inflows at a later point in time. The second common condition is that one or two members of the board must be replaced with more independent board members. Many private companies have boards consisting of representatives that are related to the company in some way. Public companies must have more independent boards. When a company is accepted or accepted with conditions, the investment bank starts the roadshow (the marketing and sale of new stock). This is the main reason why a private company needs to use an investment bank. Distribution of shares is potentially hard to accomplish without the sales force of the bank. The company has 45 days to list after it has been accepted or accepted with conditions. If the company is not listed in this period, the process must be repeated. Most of the companies that list on the OSE are forced to issue equity as a part of the listing process. Out of the 403 listings at the OSE in the period 1993 to 2007 only 90 companies are able to list without increasing their equity level in some way.60

4.3.2 A public or a private off ering? Due to oversubscription and share rationing it is diffi cult for investors to buy large blocks of shares in IPOs. In the traditional IPO setting investors submit bids for a given number of shares at a specified offer price (book-building). (In a fixed price offering the investment bank determine the price first and then investors submit bids for shares at the given price). It is common that IPOs are oversubscribed. This means that there are normally bids for more shares than the company is planning to sell. Investment banks usually set the offer price where demand is above supply. Sometimes demand is many times greater than the supply of shares (this is the oversubscription fraction reported in the newspapers after the offering). When IPOs are oversubscribed, shares are often rationed to the applicants at the price decided. An investor that bid for a high number of shares with a high bid price is likely to only be awarded a fraction of the applied for shares. The price is likely to be lower than the bid price because there is only one offer price to all investors. Rationing means that investors are likely to not be allocated blocks of shares. In negotiated private placements, on the other hand, shares are normally sold in blocks. The investors that are willing to pay the most for blocks of shares are awarded the blocks. This means that negotiated private placements are more suitable to transfer

60 Figure 1 list the timeline in the listing process.

78 blocks of shares. It is easier for an investors to obtain company blocks following private placements. An investor that wants to sell company control rights should therefore issue shares in a private placement. It is possible to stage the equity sales by ﬁrst selling blocks and then selling the remaining shares. This is also what is observed in the data. Many companies that use private placements also sell shares publicly afterwards. Interestingly, this is the opposite order of what is predicted by Zingales (1995).61

4.4 Theoretical predictions and testable implications The value of owning company shares can come from two sources. The first source is the residual claim to cash (cash flow rights). When all debtholders and other claimants to company cash flow has been paid, the remaining cash is the property of shareholders. The second source is the ability to enforce control (control rights). An owner with a high ownership percentage can influence more control and dilute more corporate resources away from smaller owners. This is private benefits of control that comes from owning a big stake in a company. The private benefits of control only goes to the controlling owner(s). Private benefits of control is enjoyed by the single biggest owner, or a group that together makes a controlling stake, at the expense of smaller shareholders (Zwiebel, 1995). Zwiebel (1995) explains that smaller block holders can join together and get control if desired. Transfer of control is therefore not necessarily from one big shareholder to another big shareholder. Transfer of control can also be from one big shareholder to a small group of block shareholders. Value of control can come from influencing how a company is run, but value of control can also come from the ability to misuse corporate resources. In some companies it is likely that it is easier to use control to move resources than in other companies. According to Zingales (1995) the seller of a company can maximize proceeds from cash flow rights by selling in an IPO to dispersed shareholders. The seller can maximize proceeds from control rights by directly bargaining with the buyer. Zingales (1995) explains that companies should optimally be sold in a two-stage process. Sellers should first sell a part of the company to dispersed shareholders. Then, the control rights should be sold in a private negotiation. In our data set there are no companies that follow this two-stage strategy, so we can not test this model directly. We can, however, test if companies with more value from control rights (higher private benefits of control) are more likely to be sold in negotiations (private placements). A company with a high value of control should be sold in a private placement because it is easier to transfer control this way.62 The testable prediction from Zingales (1995) is that there should be a relation between private benefits of control and the use of private placements. We label this the private benefits of control hypothesis based on Zingales (1995).

61 Zingales (1995) predicts that companies with high private benefits of control will sell shares in a public offering first. Remaining shares will be sold in a private placement at a later stage. We observe that the private placement takes place before the public offering every time this two stage process is used. This is opposite of what is predicted by Zingales (1995). 62 If there are high private benefits of controlling a firm, the firm could potentially stay private so that the owner can continue to enjoy the private benefits of control. If owners still want to go public, it can be argued that it will be better for the seller to sell control rights separately. There are many benefits of being publicly listed. The most notable is access to capital. It is therefore safe to assume that also companies with high private benefits of control benefit of being publicly listed.

79 4.4.1 The private benefits of control hypothesis To test the relationship between private benefits of control and the use of private placements it is necessary to measure private benefits of control. It is not possible to know the exact level of private benefits of control because it is an unobservable variable. It is, however, possible to observe some sources of private benefits of control. We use these sources as estimates of the private benefits of control for the controlling owners. It is mainly expected that companies with block ownership before the initial offering have higher private benefits of control. Zwiebel (1995) argue that the main reason why there are block owners is because of private benefits of control from taking advantage of smaller owners. Accordingly, there should be more private benefits of control in a company when there are more and bigger block owners. Private benefits of control are therefore estimated on the basis of bock ownership before the offerings.63 The ownership fraction of the largest owner before the offering is used as one measure of private benefits of control.64 The combined ownership fraction of all block holders is used as another measure of private benefits of control. Other measures that also indicate the level of private benefits of control are the tim- ing of the offering, company industry, dividend payout, family ownership and positions, minority power and CEO/board compositions. In 2006 there was introduced a new law that increased tax on dividends in Norway. It is expected that this new tax will reduce the level of dividend paid out after 2006. It is expected that private benefits of control will increase after 2006 because more money is left in the companies. Total dividends paid in the year before the listing is also included based on the same argument. It is also expected that firms in certain industries give higher private benefits of control. Especially, it is expected that firms in the sports and communications industry have more benefits of control, see Zwiebel (1995). Unfortunately, there are no sports companies and very few communications companies listed in Norway. This variable is therefore dropped. It is also expected that family firms have higher benefits of private control than non- family firms. It can be argued that family firms have already used their benefits of control by placing family members in management positions. Family firms are defined, in this paper, as firms where members of one family together hold the largest fraction of the company and more than one member of the family is in the senior management. It is also expected that minority power is decreasing in private benefits of control. It is expected that the founder is the minority owner in the company. New owners can group together and gain control. It is therefore expected that minority (founder) power should decrease in private benefits of control. Minority power is measured by founder position in the

63 It is likely that tunneling is one of the major sources of private benefits of control. In tunneling, the biggest owner owns a large stake (e.g. 51%) in one firm and 100% of another firm. The biggest owner then tunnels resources from the firm with 51% ownership to the firm with 100% ownership. Tunneling can for instance be in the form of selling assets below actual value. Tunneling lets the big owner steal resources from the shareholders that own the remaining 49% of the shares in the first company. We are not able to detect tunneling in the data. 64 All variables, unless otherwise specified, are obtained in the VPS ownership database prior to the offering or in the listing prospectus made before the offering. This means that all independent variables are known and observed before the private placement/public offering choice is made. The listing prospectus is mainly based on annual accounting data, so it is reasonably assumed that all information in the prospectus is available before the public offering/private placement choice is made. Even the level of capital raised should be known before the public offering/private placement choice is made. Capital raised is in most cases dictated by OSE as a requirement to list. We argue that there are no simultaneous decisions in our data, and there is no endogeneity issues in the analysis.

80 companies (E.g. The founder is the CEO or on the board of directors). The ownership concentration of the owners besides the single biggest owner is also a measure of minority power. This is measured by the Herfindahl index of the 50 biggest owners besides the single biggest owner. Finally, it is expected that there are more private benefits of control in companies where the largest owner use control in an observable manner. It is expected that in companies where the largest owner is the CEO or on the board of directors there are more benefits of private control. The dummy variable private placements (0) or public offerings (1) is regressed on the private benefits of control measures in a standard probit model to test if companies use private placements when there are more private benefits of control.65

4.4.2 Alternative explanations Private placements have, in addition to private benefits of control, also been explained with the monitoring (Wruck, 1989), the certification (Hertzel and Smith, 1993), the entrenchment (Barclay et al.,2007), the undervaluation (Anshuman et al., 2010) and the asymmetric information (Cronqvist and Nilsson, 2005) hypotheses. The monitoring hypothesis is that investors buy shares in private placements to increase company valuation through increased monitoring of management. It is likely that companies with high ownership concentration, before the initial offering, already have more monitoring of management than companies with lower ownership concentration. Block owners are more likely to monitor management because they have more at stake in the companies. The monitoring hypothesis therefore predicts (indirectly) that companies with lower ownership concentration should be more likely to use private placements. This is the opposite prediction of the private benefits of control hypothesis. The monitoring hypothesis is therefore controlled for by testing the relationship between ownership concentration, before the initial offering, and the use of private placements. The certification hypothesis is that informed investors buy shares in private placements to put their stamp of approval on company valuations. This does not give the same implications as the private benefits of control hypothesis. There is no reason why a company with more concentrated ownership would need more certification than a company with less concentrated ownership. It is, however, likely that smaller and younger companies would be more likely to want certification, as there is less information publicly available for these companies. The certification hypothesis is therefore controlled for by including the number of employees (size) and company age in all regressions. The entrenchment hypothesis is that private placements are used by company management to keep their positions (even if they perform poorly). Entrenchment is a highly unlikely explanation for the companies in our sample. All companies are eventually listed publicly and this indicates that these companies are doing very well. It is very unlikely that the companies in our sample have management that consistently need ownership manipulation to keep their positions. It can also be seen in Table 3 that most of the companies in the sample have the largest owner as the CEO or on the board of directors. This indicates that these owners are active and not passive investors that help keep poor management in their positions. The entrenchment hypothesis will also not explain why companies with more concentrated ownership before the initial offering are more likely to

65 It is argued that value of control does not require 51% of the shares (Damodaran, 2005). We do not know how much ownership that is needed to enjoy private beneﬁts of control, so the ownership percentage of the largest owner or the combined block ownership is included in all regressions.

81 use private placements. If private placements are used by companies with poor management, it is, however, likely that company results before the offering are negatively related to the use of private placements. The entrenchment hypothesis is therefore controlled for by including company results before the offering in all regressions. The undervaluation hypothesis is that insiders buy shares through private placements when they perceive the company to be undervalued. In the capital history section in the listing prospectus there are clear distinctions between employee offerings and private placements. Company insiders buy shares in employee offerings and not through private placements. The ownership level for all company insiders is also disclosed before and after the equity offerings, so we know that the private placements are not made towards company insiders. The undervaluation hypothesis is therefore not relevant for our data set and question. The asymmetric information hypothesis is that companies with high information dis- crepancies, between company insiders and outsiders, use private placements to reduce the cost of conveying information to investors. It is likely that certain (harder to value) industries are more likely to have more information asymmetry. Especially, it is expected that companies in the information technology (IT) sector have more information asymmetry than other companies that list on the OSE. It is also expected that younger and smaller companies have more information asymmetry because less information is publicly available for these companies. IT, younger and smaller companies should use more private placements if this hypothesis is true. It is tested if asymmetric information drives the private placement choice by including a dummy variable for all companies in the IT sector, the company age and the number of employees in all regressions.

4.4.3 Other control measures The reasons why companies issue equity is to have suffi cient levels of equity and number of owners before the listings. The OSE requires a minimum of 500 investors to list on the main list of the OSE (and 100 to list in the small and medium sized list). Therefore, it is necessary to control that the number of investors prior to the offering and the capital raised do not decide the method chosen. These variables are therefore included in all regressions. Carpentier and Suret (2009) show that Canadian firms that use private placements have lower book to market rations, are in special industries, are financially distressed or constrained, are in the development stage and in general raise less capital. Barclay et al. (2007) show that private placements are made at a discount to certain investors. Boone and Mulherin (2007) show that market value is related to the use of private placements. The problem with these variables is that they are observed only after the listing. Most of these variables are observed the first time about six months after the initial private placement or public offering choice has been made. The variables book to market ratio, first day return and market value are observed the first time on the day of the listing. These variables are not available for the companies in our sample because they are privately held. All companies in the sample are also eventually listed on the stock exchange, so there are no financially distressed or constrained firms in the sample. (This is, however, controlled for by including the last annual net result reported in the listing prospectus).

82 4.4.4 Private benefits of control also after the listing It can be argued that companies with high private benefits of control should stay private. The reason for this is that some of the private benefits of control is likely to disappear when companies become public. We therefore test if there are private benefits of control after the new listings. If control rights are sold in private placements, there should be greater values of control also after the listings following private placements. To test for benefits of control after the listing it is necessary to regress private benefits of control after the listing on the public offering or private placement choice. Private benefits of control is an unobservable variable that is estimated by a portfolio of measures. Most of these measures are very persistent. E.g. Few companies change the CEO or board members right after the listing and company specific variables such as age, number of employees, family firm, result and dividend do not change. These variables are not suitable as single measures of private benefits of control. A more suitable measure of private benefits of control is the ownership fraction of the biggest owner(s) after the listing. If there is a more concentrated ownership also after the listing, it can be argued that there is persistence in the control. This is tested by regressing the ownership percentage of the biggest owner(s) one month after the listing on the private placement or public offering choice (before the offering) and a set of control variables.

4.5 Data and descriptive statistics There are 403 companies the list publicly on the OSE in the period January 1993 to September 2007. Table 2 gives the yearly distribution of IPOs and negotiated private placements in this period. All companies must list their ownership records in the Nor- wegian central depository (VPS) database as a part of the listing procedure. From this database the pre offering ownership in all listed companies is observed. Accounting variables are collected from the listing prospectuses. It is assumed that private placements in the six month period before the listing date are part of the listing procedure. Private placements before this are assumed to not be part of the listing procedure.66 Company ownership at the end of month six prior to the listing date is the measure of ownership concentration prior to the offering. Most public offerings are in the calendar month before or in the same calendar month as the listing date. Private placements are spread out over the six months prior to the listing date. From Table 2 it can be seen that there is a proportionate number of private placements and IPOs over the sample period. There is a slight increase in the number of private placements compared to IPOs in the end of the sample period. It is argued that the reason for this is an increase in the Norwegian tax rates in 2006 that increased overall private benefits of control from more retained cash. There are 210 public offerings and 106 private placements by companies listing on the OSE in the period 1993 to September 2007.67 For 19 public offerings and 6 private

66 Companies define the equity offering to be private or public in the capital history section in the listing prospectuses. Data on all historical equity offerings are provided in these prospectuses.

67 In total, 44 companies used a private placement before a public offering, and 131 companies did not offer shares to new investors in the lead up period to the listing (21 of these companies were spinoffs to existing shareholders). Private placements are made at different points in time in the six months period before the listings. Private placements before this is not included in the sample. The public offerings are

83 placements it has not been possible to identify the ownership before the offering from the VPS ownership database. These companies are removed from the sample.68 A total of 44 companies made a private placement before the public offering. These companies are regarded as only private placement companies as they made this offering first.69 The final sample is 88 companies that used a private placement and 123 companies that used an IPO.

4.5.1 Descriptive statistics From Table 3 it can be seen that companies that use private placements and public offerings are very similar. Private placement companies do, however, have on average more large owners on the boards, higher ownership fractions of the largest owners after the listings, more founders on the boards or as the CEOs, are more likely to be family firms before the offerings and have lower age. The average capital raised in the 88 private placements is $57.3 million. This is just below the average size of the public offerings. For private placements the combined sale of new and existing shares averages about 22% of total outstanding shares at the listing date. For public offerings this number is 41%. There are no significant differences between companies that use private placements and public offerings on total assets, dividends, results, number of owners before the offering, capital raised and number of employees.

4.5.2 Variable description The dependent variable in most regressions is a dummy variable for public offering (1) and private placements (0).70 Combined block ownership is the combined ownership fraction of all investors that owns more than 5% of the company before any offering is made.71 Holding of largest owner b. offer is the holding fraction of the single biggest owner before the offering. Holding of largest owner a. listing is the holding fraction of the single biggest owner one month after the listing. Largest owner is the CEO and Largest owner is on the board are dummy variables that takes the value of one for companies where the largest usually performed in the month before the listing or in the listing month itself. Some private placements have a follow on offering to the public or to employees of the company. By using follow on offerings the minimum number of investors regulation, set by stock exchanges, has no influence on the equity offering method chosen. The remaining 110 listings are results of mergers with an already listed company, cross listings or companies traded actively at the Norwegian over the counter list (OTC list) before the OSE listing. 68 For 27 public offerings and 12 private placements it has not been possible to obtain all company specific information (i.e. listing prospectuses). These companies are therefore removed from the sample. 69 When there is both a public and a private sale it is common that the investors in the private placement sell a small fixed percentage of their allocated shares in the public offering. It is likely that the private placement is made to increase the capital for the company through the issue of new shares. It is also likely that the public offering is made to increase the number of shareholders. It is common that there is one fixed resell percentage that applies to all investors in the private placement. This percentage is usually very low (less than 10%). The issuing company have then sold shares with the condition that the investors must sell some of their allocated shares before the listing. It is likely that this condition is included to meet minimum spread requirements set by the OSE. The final sample is 123 public offerings and 88 private placements. 70 All ownership variables are obtained from the VPS database. All other pre listing variables are obtained from the listing prospectuses that are made in connection with the listings. 71 In Norway, all shareholders that own more than 5% of the outstanding shares must be reported in the listing prospectus. In the remainder of the article we refer to shareholders that own more than 5% of outstanding shares as block holders.

84 owner is the CEO or on the board of directors. The founder is the CEO and The founder is on the board are dummy variables that take the value of one if the founder is the CEO or on the board of directors. Herfindahl index is the squared ownership fraction of the sum of the 50 biggest owners besides the largest owner.72 The 2006 dummy takes the value of one for all companies listed after 2005. (Dividend / Total Assets) is the total dividend payment made in the year before the listing year scaled by total assets. The Family firm dummy takes the value of one for family firms. Family firms are identified in the listing prospectuses as firms where members of one family together hold the largest fraction of the company and more than one member of the family is in the senior management of the firm. Age is the difference between listing year and the year of incorporation of the companies. Number of employees is the number of annual accumulated full time employees in the issuing company. Capital raised is the total number of shares sold in the offering times the offer price. N. owners before offering is the number of investors that own shares in the company before the offering. Capital raised and N. owners before offering are weakly negatively correlated. (Net result / Total Assets) is the last annual end of year result, scaled by total assets, listed in the listing prospectus. IT dummy takes the value of one for companies in the information technology (IT) sector. Year fixed dummy is included as dummy variables for the different years in the sample period (1993 to 2007).

4.6 Empirical Results The main empirical finding of the paper is that companies with more block ownership, before the offerings, are more likely to use private placements instead of public offerings. The companies that did use private placements also have more block ownership after the listings. There is also a bigger reduction in block ownership following public offerings than following private placements.

4.6.1 The private benefits of control hypothesis The dummy dependent variable private placement (0) or public offering (1) is regressed on the estimated private values of control in a probit regression.73 From Table 4 it can be seen that companies with one large owner prior to the initial offering are more likely to use private placements. The coeffi cient for holding fraction of the largest owner on the issue choice is positive and significant at the 0.01 level. The positive coeffi cient supports the private benefits of control hypothesis that private placements are used to transfer private benefits of control. Companies where there is one controlling owner prior to the offering

72 In general, it is expected that private benefits of control should decrease in minority power. There are, however, some sample characteristics that may alter this expectation. In many companies there are a small group of investors that jointly owns a controlling stake in the company together (E.g. a family or a group of friends). It is expected that all of these investors will enjoy the private benefits of control even if one investor have a slightly larger stake than the others. Zwiebel (1995) also argues that there are private benefits of control from block holders that are not the single biggest owner. 73 We do not expect there to be any problems with endogeneity in the analysis. All independent variables are observed in the listing prospectus before the public offering. We assume that these variables are also publicly available before the private placements even if these may be up to five moths before the listing prospectus is available. We argue that the used independent variables are determined before the private placement/public offering choice, and any endogeneity due to simultaneity will therefore not be an issue. The variables in the listing prospectuses are also available in annual (and quarterly) reports. It is reasonably assumed that investors are able to locate this information before the private offering.

85 are more likely to use private placements instead of public offerings. Companies that issue equity in periods where there is likely to be more private benefits of controlling firms (after 2006) also issue more in private placements. Companies that use private placements are also more likely to be family firms. Most control variables are unrelated to the issue choice. The level of capital raised is highly related to the use of public offerings. From Table 5 it can be seen that the exact same results are obtained when the block ownership fraction of the biggest owners is used instead of the ownership fraction of the single biggest owner. Companies with more block ownership are more likely to use private placements whereas companies with less block ownership are more likely to use IPOs. These results control for the level of capital raised, the number of investors that own shares in the companies before the offerings and the alternative explanations for private placements. The results are also robust to the removal savings banks (13). From Table 6 it can be seen that the relationship between private benefits of control and private placements is robust to including year fixed effects. It is not possible to reject the hypothesis that private placements are used to transfer private benefits of control.

4.6.2 Alternative explanations The private placement choice has in the previous literature, in addition to the private benefits of control hypothesis, been explained with monitoring, certification, entrenchment, undervaluation and asymmetric information. There is a positive relationship between the use of private placements and the holding fraction of the largest owner(s) before the offering. This is the opposite finding of what is predicted by the monitoring hypothesis and this hypothesis is therefore rejected. There is not a consistent relationship between the use of private placements and company age and number of employees. It is likely that younger and smaller companies have more need for value certification from informed investors than other companies. The certification hypothesis is therefore also rejected. If company management use private placements to keep their control even if they perform poorly, it is expected that there will be a negative relationship between company results and the use of private placements . There is, however, not a consistent relationship between company results before the offerings and the use of private placements. The entrenchment hypothesis is therefore also rejected. There is also not more private placements by younger and smaller companies in the IT industry. If private placements are used to reduce the problems associated with information asymmetry, it is expected that there will be a relationship between companies with more expected information asymmetry (e.g. smaller, younger and IT companies) and the use of private placements. This relationship does not exist and the asymmetric information hypothesis is therefore also rejected.

4.6.3 Private benefits of control also after the listing If control rights are sold in private placements, there should be greater values of control also after the listing in companies that used private placements. This is tested by regressing private benefits of control after the listing on the IPO or private placement choice. In Table 7 the combined ownership percentage of block owners one month after the listing is regressed on the public offering or private placement choice (and the control variables for the alternative explanations). There is more block ownership one month after the listing following private placements than following public offerings. Public offerings are related to smaller block ownership one month after the listing. In Table 8 it can be seen that

86 the exact same results are found when only the ownership of the single largest owner is studied separately. This show that there is more block ownership in companies that used private placements also after the listings.

4.7 Conclusion There is a strong and robust relationship between the ownership fraction of the largest owner(s), before the initial equity offering, and the use of private placements. The biggest owner(s) also have a higher ownership fraction following private placements than following public offerings. If it is assumed that the main reason that investors are willing to hold blocks of shares is to enjoy private benefits of control, it can be concluded that private placements are used to transfer private benefits of control. Zwiebel (1995) argue that the only reason investors hold blocks of shares is to enjoy private benefits of control. We reject that private placements are used because of monitoring, certification, entrenchment, undervaluation or asymmetric information considerations. We conclude that private placements are used to transfer private benefits of control between the seller and the buyer. The main theoretical implication of this finding is that Zingales (1995) is correct in that company control rights are better sold separately. Companies are sold based on the value of control rights when they are higher than the stand alone cash flow rights. The finding also have implications for auction theory. When the auction makes it hard to obtain blocks of shares, as in the case of the IPO, the negotiation may be preferred by the seller if there are private benefits of control. The main practical implication of this finding is that companies should use private placements when the value of control rights are higher than stand alone cash flow rights. If there are larger private benefits of controlling a firm, the firm should be sold in a private placement. There are some limitations to the study. Private benefit of control is an unobservable variable that can come from an unlimited number of sources. Private benefit of control is estimated based on existing ownership and company specific variables. A more directly observable measure of private benefits of control would have been preferable. It is also not possible to detect tunnelling in the data. Tunnelling is likely to be a major source of private benefits of control. For future research it would be interesting to study a bigger sample that includes more firms with obvious private benefits of control such as sports companies. It would also be very interesting to study cross company ownership and related business deals. Business deals by companies with the same ownership would allow us to study tunneling.

87 References

[1] Anshuman, V. Ravi, Vijaya B. Marisetty and Marti G. Subrahmanyam, 2010, Private Placements, Regulatory Restrictions and Firm Value: Theory and Evidence from the Indian Market, NYU working paper.

[2] Arugaslan,˘ Onur, Douglas O. Cook and Robert Kieschnick, 2004, Monitoring as a Motivation for IPO Underpricing, Journal of Finance 59, 2403-2420. [3] Barclay, Michael J., Clifford G. Holderness and Dennis P. Sheehan, 2007, Private Placements and Managerial Entrenchment, Journal of Corporate Finance 13, 461-484. [4] Boone, Audra L. and J. Harold Mulherin, 2007, How are Firms Sold?, Journal of Finance 2, 847-875. [5] Boone, Audra L. and J. Harold Mulherin, 2008, Do Auctions Induce a Winner’sCurse? New Evidence from the Corporate Takeover Market, Journal of Financial Economics 89, 1-19. [6] Bulow, Jeremy and Paul Klemperer, 1996, Auctions Versus Negotiations, The Amer- ican Economic Review 86, 180-194. [7] Bulow, Jeremy and Paul Klemperer, 2009, Why Do Sellers (Usually) Prefer Auctions?, The American Economic Review, forthcoming. [8] Brennan, Michael and Julian Franks, 1997, Underpricing, Ownership and Control in Initial Public Offerings of Equity Securities in the U.K., Journal of Financial Economics 45, 391-413. [9] Carpentier, Cecile and Jean-Marc Sure, 2009, Private Placements by Small Pub- lic Entities: Canadian Experience, Douglas J. Cumming, Ed., Companion to Private Equity, Wiley, Chapter 8, forthcoming. [10] Cronqvist, Henrik and Mattias Nilsson, 2005, The Choice Between Rights Offerings and Private Equity Placements, Journal of Financial Economics 78, 357-407. [11] Damodaran, Aswath, 2005, The value of Control: Implications for Control Premia, Minority Discounts and Voting Share Differentials, NYU working paper. [12] Derrien, François and Ambrus Kecskés, 2007, The Initial Public Offerings of Listed Firms, Journal of Finance 1, 447-479. [13] French, Kenneth R. and Robert E. McCormick, 1984, Sealed Bids, Sunk Cost, and the Process of Competition, The Journal of Business 4, 417-441. [14] Hertzel, Michael and Richard L. Smith, 1993, Market Discounts and shareholder Gains for placing Equity Privately, Journal of Finance 48, 459-485. [15] Jenkinson, Tim and Howard Jones, 2004, Bids and Allocations in European IPO Bookbuilding, Journal of Finance 59, 2309-2338. [16] Megginson, William L., Robert C. Nash, Jeffry M. Netter and Annette B. Poulsen, 2004, The Choice of Private Versus Public Capital Markets: Evidence from Privatiza- tions, Journal of Finance 59, 2835-2870.

88 [17] Myers, S.C., Majluf, N., 1984, Corporate Financing and the Investment Decision When Firms Have Information that Investors do not have, Journal of Financial Eco- nomics 13, 187-221.

[18] Ritter Jay R., 2003, Diﬀerences between European and American IPO markets, Eu- ropean Financial Management 9, 421-434.

[19] Ritter Jay R., 1984, The Hot Issue Market of 1980, The Journal of Business 57, 215-240.

[20] Stoughton Neal and Josef Zechner, 1998, IPO —Mechanisms, Monitoring and Own- ership Structure, Journal of Financial Economics 49, 45-77.

[21] White, Halbert, 1980, A heteroskedastic-consistent covariance matrix estimator and a direct test of heteroscedasticity, Econometrica 48, 817-838

[22] Wu, YiLin, 2004, The Choice of Equity Selling Mechanisms, Journal of Financial Economics 2004, 3-28.

[23] Wruck, Karen Hopper, 1989, Equity Ownership Concentration and Firm Value: Ev- idence from private equity ﬁnancing, Journal of Financial Economics 23, 3-28.

[24] Zingales, Luigi,1995, Insider Ownership and the Decision to Go Public, The Review of Economic Studies 62, 425-448.

[25] Zingales, Luigi, 1994, The value of Voting Right: A study of the Milan Stock Ex- change Experience, The Review of Financial Studies 1, 125-148.

[26] Zwiebel, Jeﬀrey, 1995, Block Investment and Partial Beneﬁts of Corporate Control, The Review of Economic Studies 62, 161-185

89 Table 1 Related Studies

Auction (theory) BulowandKlemperer(1996) Sellerprefertosellinanauction

BulowandKlemperer(2009) Buyersprefertobuyinnegotiation and sellers prefer to sell in an auction.

FrenchandMcCormick(1984) Auctionsareusuallypreferred

Equity oﬀerings (theory)

Zingales(1995) Controlrightsareoptimallysoldprivate

Zingales(1994) Privatebeneﬁtsofcontrolisdilution of minority property rights

Zwiebel(1995) Therearebeneﬁtsofblockssmallerthancontrol

StoughtonandZechner(1998) Privateplacementsincreasemonitoring

Attract certain types of investors (empirical)

Wruck(1989) Monitoringhypothesis

HertzelandSmith(1993) Certiﬁcationhypotheses

Barclayetal.(2007) Entrenchmenthypothesis

Anshumanetal.(2010) Undervaluationhypothesis

BrennanandFranks(1997) Underpricingusedto avoid block holder formations

Arugaslan,CookandKieschnick(2004)˘ Monitoringnotimportant

Wu(2003) Monitoringnotimportant

CronqvistandNilsson(2005) Privateplacementsreduce moral hazard and adverse selection

BooneandMulherin(2007) Auctionsdoesnotincreaserevenue for the seller

BooneandMulherin(2008) Thereisnorelationbetweenbidder return and competition

90 Table 2 IPOs and Private Placements on the Oslo Stock Exchange

This table gives the annual distribution of initial offerings: Column 1 is the sample years. Column 2 is the number of public offerings per year. Column 3 is the average underpricing of the public offerings per year. Column 4 is the total capital raised in all public offerings combined per year in USD. Column 5 is the number of private placements per year. Column 6 is the average underpricing of the private placements per year. Column 7 is the total capital raised in all private placements combined per year in USD. Underpricing is calculated as: (offer price in the listing prospectus —first day closing price) / offer price in the listing prospectus. Value of shares sold is reported in USD using a USD/NOK exchange rate of 0.1792. The sample period is January 1993 through September 2007.

PublicOﬀerings PrivatePlacements Distribution Capitalraised Distribution Capitalraised Year N Underpricing% MUSD N Underpricing% MUSD 1993 5 -1.8% $474 4 27.4% $81 1994 10 4.2% $609 2 4.8% $20 1995 6 6.7% $467 5 8.1% $49 1996 4 24% $99 5 10.6% $49 1997 15 16.6% $972 11 34.6% $139 1998 8 1.9% $185 6 -6.1% $108 19994 18.7% $185 0 0 0 2000 9 -0.9% $517 6 36% $527 2001 4 -7.4% $183 2 6.5% $483 2002 2 -9.8% $70 1 2.5% $210 20032 -2.3% $83 0 0 0 2004 13 5.6% $1,602 1 5.5% $3.6 2005 20 3.3% $1,709 18 6.6% $1,711 2006 12 3.2% $1,417 9 9.2% $584 2007 9 3.3% $793 18 6.9% $1,077

Total 123 5.3% $9,365 88 12.7% $5,074

91 Table 3 Summary Statistics on Firms Going Public

This table show the difference between companies using initial private placements and initial public offerings. "Combined block ownership" is the combined ownership of all investors that owns more than 5% of the company before the offering. "Holding of largest owner b. offer" is the holding fraction of the single biggest owner before the offering "Holding % of largest owner a. listing" is the holding fraction of the single biggest owner one month after the listing. "Reduced % of largest owner" is the difference in the ownership fraction of the largest owner from before the offering to one month after the listing. "Largest owner is the CEO dummy", "Largest owner is on the board dummy", "The founder is the CEO dummy" and "The founder is on the board dummy" are dummy variables that take the value of one if the biggest owner or founder are the CEO or on the board. Herfindahl index is the sum of the squared ownership fraction of the 50 biggest owners besides the largest owner. "Age of company" and "Number of employees" is the age and the number of employees of the issuing company. "2006 dummy" and "Family firm dummy" takes the value of one for issues after 2006 and family firms respectively. "IT dummy" takes the value of one for companies in the information technology (IT) sector. "Capital raised " is the offer price times the number of shares sold in the offering. "N. owners before offering" and "First day return %" are the number of owners in the company before the offering and the first day return from offer price to first day closing price respectively. "Market value" is the number of outstanding shares at the listing day times the first day closing price. "Fraction of company sold" is the fraction of sold shares to outstanding shares in the offering. "Net result", "Dividends" and "Total assets" are the last annual result, dividend paid and total assets reported in the listing prospectus before the offering. The t —statistic is calculated as: (Mean private placements - mean public offerings) / (square root [ (variance private placements / numbers of private placements) + (variance public offerings/ numbers of public offerings)].

Privateplacement Publicoffering Difference Variables Obs. Mean Std.Dev Obs. Mean Std.Dev Diff. t-stat. Combinedblockownership 88 0.78 0.23 123 0.76 0.26 0.02 (0.6) -withnosavingsbanks 88 0.78 0.23 110 0.74 0.26 0.04 (1.1) Holdinglargestownerb.offer 88 0.5 0.31 123 0.5 0.34 -0.01 (-0.2) -withnosavingsbanks 88 0.5 0.31 110 0.47 0.32 0.02 (0.7) Holdinglargestownera.listing 85 0.3 0.16 123 0.26 0.18 0.04 (1.7) Reduced%oflargestowner 85 0.2 0.23 123 0.25 0.32 -0.05 (-1.3) LargestowneristheCEOD 88 0.24 0.43 123 0.16 0.37 0.08 (1.4) LargestownerisontheboardD 88 0.52 0.5 123 0.31 0.46 0.21 (3.1) Herfindahlindex 88 0.05 0.05 123 0.04 0.05 0.01 (1.4) ThefounderistheCEOD 88 0.27 0.45 123 0.18 0.38 0.09 (1.5) ThefounderisontheboardD 88 0.36 0.48 123 0.23 0.42 0.13 (2.0) Ageofcompanyinyears 88 19.5 28.4 123 36.2 47 -16.7 (-3.2)

92 Table3continued. Privateplacement Publicoffering Difference Variables Obs. Mean Std.Dev Obs. Mean Std.Dev Diff. t-stat. Numberofemployees 88 507 1,343 123 735 2,220 -228 (-0.9) 2006dummy 88 0.31 0.46 123 0.17 0.38 0.14 (2.3) Familyfirmdummy 88 0.27 0.45 123 0.12 0.32 0.15 (2.7) ITdummy 88 0.15 0.36 123 0.2 0.4 -0.05 (-0.9) Capitalraised(MillUSD) 88 57.3 93.1 123 75.1 121 -17.8 (-1.2) N.ownersbeforeoffering 88 233 654 123 135 265 98 (1.3) Firstdayreturn 88 0.13 0.334 123 0.05 0.14 0.08 (2.0) MarketvalueE.(MillUSD) 88 351.8 525.2 123 236.6 418.7 115.2 (1.7) Fractionofcompanysold 88 0.22 0.24 123 0.41 0.26 -0.19 (-5.5) Netresult(MillUSD) 88 5.6 74.5 123 4.2 30.8 1.4 (0.2) Dividends(MillUSD) 88 0.31 0.96 123 1.4 9.3 -1.1 (-1.3) Totalassets(MillUSD) 88 912 4,926 123 408 968 504 (0.9)

93 Table 4 Private Placements and Private Beneﬁts of Control of the Single Biggest Owner

This table reports the coeffi cients and t -statistics in parentheses for the regressions with the dummy variable that takes the value of one for IPOs and zero for private placements as the dependent variable. All regressions are standard Probit models. The sample period is September 1993 to January 2007. All variables are as described in Table 3. Age, employees, capital raised and number of owners are in log in all regressions. In all Regressions the ownership fraction of the single biggest owner before the (first) offering is included. In Regression 1 and 2 savings banks (13) are dropped. Regression 2 includes White (1980) robust standard errors. In regression 3 all savings banks (13) are included. No independent variables have a correlation above 0.5.

Dummy IPO (1) or Private Placement (0) Reg1 Reg2 Reg3 Intercept -4.6415 -4.6415 -4.0439 (-2.9) (-2.8) (-2.6) Holdingfractionoflargestownerbeforeoffering -1.5283 -1.5283 -1.5313 (-2.5) (-2.4) (-2.6) LargestowneristheCEOdummy 0.2489 0.2489 0.1852 (0.9) (0.9) (0.6) Largestownerisontheboarddummy -0.257 -0.257 -0.303 (-1.0) (-1.0) (-1.2) Herfindahlindex -3.9762 -3.9762 -5.0464 (-1.6) (-1.5) (-2.1) ThefounderistheCEOdummy -0.3744 -0.3744 -0.3821 (-1.3) (-1.3) (-1.3) Thefounderisontheboarddummy 0.2393 0.2393 0.1855 (0.9) (0.9) (0.7) Ageofcompany 0.1346 0.1346 0.203 (1.6) (1.4) (2.6) Numberofemployees 0.077 0.077 0.0556 (1.4) (1.3) (1.0) 2006dummy -0.5095-0.5095-0.5206 (-2.1) (-2.2) (-2.2) Familyfirmdummy -0.4829 -0.4829 -0.5055 (-1.7) (-1.8) (-1.8) Capitalraised 0.2998 0.2998 0.2789 (3.6) (3.4) (3.4) N.Ownersbeforetheoffering -0.1183 -0.1183 -0.145 (-1.6) (-1.6) (-2.0) Netresult/TotalAssets 0.3284 0.3284 0.3076 (1.2) (1.7) (1.1) Dividend/TotalAssets 3.9909 3.9909 3.0634 (0.8) (0.8) (0.6) IT dummy 0.421 0.421 0.4179 (1.5) (1.6) (1.5) Observations 198 198 211 PseudoR-squared 15.8% 15.8% 16.9% 94 Table 5 Private Placements and Private Benefits of Control of Block Owners

This table reports the coeffi cients and t -statistics in parentheses for the regressions with the dummy variable that takes the value of one for IPOs and zero for private placements as the dependent variable. All regressions are standard Probit models. The sample period is September 1993 to January 2007. All variables are as described in Table 3. Age, employees, capital raised and number of owners are in log in all regressions. In all Regressions the combined block ownership fraction of all investors with a holding percentage above 5% before the (first) offering are included. In Regression 1 and 2 savings banks (13) are dropped. Regression 2 includes White (1980) robust standard errors. In regression 3 all savings banks (13) are included. No independent variables have a correlation above 0.5.

Dummy IPO (1) or Private Placement (0) Reg1 Reg2 Reg3 Intercept -3.7758-3.7758-3.1326 (-2.3) (-2.2) (-1.9) Combinedblockownershipfraction -2.0244 -2.0244 -2.0456 (-2.8) (-2.8) (-2.8) LargestowneristheCEOdummy 0.2298 0.2298 0.1654 (0.8) (0.8) (0.6) Largestownerisontheboarddummy -0.2156 -0.2156 -0.2618 (-0.9) (-0.9) (-1.1) Herfindahlindex 1.515 1.515 0.4645 (0.8) (0.8) (0.2) ThefounderistheCEOdummy -0.3396 -0.3396 -0.348 (-1.1) (-1.1) (-1.2) Thefounderisontheboarddummy 0.2199 0.2199 0.1652 (0.8) (0.8) (0.6) Ageofcompany 0.1298 0.1298 0.2004 (1.6) (1.4) (2.6) Numberofemployees 0.0945 0.0945 0.072 (1.7) (1.6) (1.4) 2006dummy -0.4829 -0.4829 -0.4951 (-2.0) (-2.0) (-2.1) Familyfirmdummy -0.5202 -0.5202 -0.5425 (-1.8) (-1.9) (-1.9) Capitalraised 0.2786 0.2786 0.256 (3.4) (3.3) (3.2) N.Ownersbeforetheoffering -0.1256 -0.1256 -0.1521 (-1.7) (-1.7) (-2.1) Netresult/TotalAssets 0.3123 0.3124 0.2913 (1.2) (1.6) (1.1) Dividend/TotalAssets 4.5216 4.5216 3.4597 (0.8) (0.9) (0.7) ITdummy 0.4253 0.4253 0.4199 (1.5) (1.6) (1.4) Observations 198 198 211 PseudoR-squared 16.4% 16.4% 17.6%

95 Table 6 Private Placement and Private Beneﬁts of Control - Year Fixed Eﬀects

This table reports the coeffi cients and standard t -statistics in parentheses for the regressions with the dummy variable that takes the value of one for IPOs and zero for private placements as the dependent variable. All regressions are standard Probit models. The sample period is September 1993 to January 2007. All variables are as described in Table 3. Regression 1 and 3 includes year fixed effects and the combined block ownership fraction before the offering. Regression 2 and 4 includes year fixed effects and the holding fraction of the single largest owner before the offering. In regression 3 and 4 all savings banks (13) are included. No independent variables have a correlation above 0.5.

Dummy IPO (1) or Private Placement (0) Reg1 Reg2 Reg3 Reg4 Intercept -4.3351 -5.2256 -3.6263 -4.5002 (-2.2) (-2.8) (-1.9) (-2.5) Combinedblockownershipfraction -1.9282 -1.8672 (-2.5) (-2.4) Holdingfractionoflargestownerbeforeoffering -1.5167 -1.4764 (-2.4) (-2.3) LargestowneristheCEOdummy 0.1811 0.204 0.1156 0.1377 (0.6) (0.6) (0.4) (0.4) Largestownerisontheboarddummy -0.3166 -0.3572 -0.3425 -0.3803 (-1.1) (-1.3) (-1.2) (-1.4) Herfindahlindex 2.9923 -2.3186 1.305 -3.8551 (1.4) (-0.9) (0.6) (-1.5) ThefounderistheCEOdummy -0.3931 -0.4329 -0.387 -0.425 (-1.2) (-1.3) (-1.2) (-1.3) Thefounderisontheboarddummy 0.2501 0.2547 0.1538 0.1625 (0.8) (0.8) (0.5) (0.5) Ageofcompany 0.1224 0.1342 0.2132 0.2212 (1.4) (1.5) (2.6) (2.6) Numberofemployees 0.0924 0.0727 0.0597 0.0419 (1.6) (1.2) (1.0) (0.7) 2006dummy -0.29 -0.1931 -0.3488 -0.2583 (-0.5) (-0.4) (-0.6) (-0.5) Familyfirmdummy -0.4996 -0.463 -0.5252 -0.4857 (-1.6) (-1.5) (-1.7) (-1.6) Capitalraised 0.3048 0.3271 0.2801 0.303 (3.3) (3.4) (3.2) (3.3) N.Ownersbeforetheoffering -0.1051 -0.1027 -0.1416 -0.1416 (-1.3) (-1.3) (-1.8) (-1.8) Netresult/TotalAssets 0.3574 0.3762 0.3003 0.3239 (1.2) (1.3) (1.1) (1.2) Dividend/TotalAssets 4.7078 3.4868 3.5196 2.4382 (0.8) (0.6) (0.6) (0.4) ITdummy 0.4174 0.3761 0.3965 0.3652 (1.3) (1.1) (1.2) (1.1) Yearfixeddummy yes yes yes yes Observations 193 193 205 205 PseudoR-squared 21.9% 21.6% 22.1% 21.9% 96 Table 7 Block Owners own more of the Company Following Private Placements

This table reports the coeﬃ cients and heteroscedastic consistent t -statistics (errors adjusted for clus- tering across ﬁrms Rogers, 1993) in parentheses for the regressions with the combined ownership percentage of the biggest owners one month after the listing as the dependent variable. All regressions are standard OLS models. The sample period is September 1993 to January 2007. All variables are as described in Table 3. Regression 1 drops savings banks (13). Regression 2 includes savings banks (13). No independent variables have a correlation above 0.5.

Combined block ownership % after the listing Reg 1 Reg 2 Intercept 71.1734 67.771 (2.9) (3.2) DummyIPO(1)orPrivatePlacement(0) -4.5329 -8.4463 (-2.0) (-3.3) Age of company 2.3088 0.2019 (2.5) (0.2) Number of employees 0.435 1.4037 (0.7) (1.7) . Capital raised -0.4381 -0.365 (-0.4) (-0.3) N.Ownersbeforetheoﬀering -1.6963 -0.5708 (-2.7) (-1.0) Netresult/TotalAssets -2.1682 -1.1675 (-1.3) (-0.5) IT dummy -1.1653 -1.003 (-0.3) (-0.3) Year ﬁxed dummy yes yes Observations 195 208 Adjusted R -squared 11.1% 8.1%

97 Table 8 The Biggest Owner have a Larger Ownership % Following Private Placements

This table reports the coeﬃ cients and heteroscedastic consistent t -statistics (errors adjusted for clus- tering across ﬁrms Rogers, 1993) in parentheses for the regressions with the ownership percentage of the biggest owner one month after the listing as the dependent variable. All regressions are standard OLS models. The sample period is September 1993 to January 2007. All variables are as described in Table 3. Regression 1 drops the savings banks (13). Regression 2 includes the savings banks (13). No independent variables have a correlation above 0.5.

Ownership % of the biggest owner after the listing Reg 1 Reg 2 Intercept 10.7614 6.9755 (0.5) (0.3) DummyIPO(1)orPrivatePlacement(0) -3.4035 -6.3423 (-1.8) (-3.1) Age of company 2.9051 1.3201 (2.5) (1.1) Numberofemployees 0.0852 0.8391 (0.2) (1.5) . Capital raised 0.9532 1.0741 (0.8) (1.0) N.Ownersbeforetheoﬀering -1.9657 -1.1421 (-2.6) (-1.6) Netresult/TotalAssets 0.7173 1.4292 (0.3) (0.7) IT dummy -6.1962 -5.6924 (-2.3) (-2.2) Year ﬁxed dummy yes yes Observations 195 208 Adjusted R -squared 15.7% 9.7%

98 Figure 1 Timeline of the Listings on the Oslo Stock Exchange

Listing in database is when the company list ownership records in the ownership database. This is when the ownership records are observed in the data the first time. Public offering or Private placement is when the companies distribute the allocated shares in the ownership database. The private placement can be at any point in time in the six month period leading up to the listing. The public offering is in most cases in the month before or the month of the listing.

Timelineofthelisting Privateplacements Publicoﬀerings Listingindatabase Listingindatabase Six months before the listing MeetingwiththeOSE MeetingwiththeOSE Compliancereport Compliancereport Duediligence Duediligence

Application submitted Application submitted Prospectusismade Prospectusismade Private Placement

One month before the listing (PublicOffering) PublicOffering (Employeeoffering) (Employee offering)

Listingmonth Listing Listing

99 5 Summary

This dissertation consists of the three papers; ’Laddering in Initial Public Offering Allo- cations’,’Using Stock-trading Commissions to Secure IPO Allocations’and ’Initial Public Offering or Initial Private Placement?’ In the paper ’Laddering in Initial Public Offering Allocations’it is found that it is likely that IPO allocations are tied to after-listing purchases of the IPO shares (IPO laddering). In the paper ’Using Stock-trading Commissions to Secure IPO Allocations’it is found that it is likely that IPO allocations are tied to stock-trading commissions before the allocations. In the paper ’Initial Public Offering or Initial Private Placement?’ it is found that it is likely that private placements are used to transfer private benefits of control from the seller to the buyer. The overall contribution of these findings is an extension to our understanding of how companies and investment banks allocate shares in initial/primary equity offerings. There is strong evidence supporting the rent seeking view of IPO allocations. Both in terms of allocating IPO shares based on after-listing purchases (IPO laddering) and allocating IPO shares based on stock-trading commissions. There is no evidence supporting the academic view or the pitchbook view of IPO allocations. The thesis also shows that investors with private benefits of controlling companies are likely to sell their control rights in private placements. The overall conclusion is that both investors and investment banks are likely to optimize their own return in the equity issuance process.

100 101