1 Amsterdam Business School
The Effects of Market-Based Pricing Mechanisms on Underpricing: A Comparison of European IPO’s
Name: Joeri de Wolf Student number: 10003983 Thesis supervisor: Réka Felleg Date: 20 June 2016
Word count: 13,272
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
2 Statement of Originality
This document is written by student Joeri de Wolf who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the
3
Abstract
This thesis examines whether market-based pricing mechanisms result in a lower degree in underpricing, by looking at European IPO’s. Auctioned and bookbuilt offerings are compared to fixed-price offerings. Europe makes for an ideal testing field, due to the shift from underwriter-based pricing techniques to market-underwriter-based pricing techniques. This study finds evidence that bookbuilding positively affects first-day returns due to the bookbuilding paradox. Conversely, no evidence is found that suggests auctioning decreases underpricing due to free riders and aggressive bidders. When pooling both bookbuilding as auctioning as market-based techniques, a significant positive effect on underpricing is found. This implies that market-based pricing mechanisms leave more money on the table for issuing companies, but ensure that all shares will be sold and an IPO failure will not occur.
4
Table of Content
1. Introduction ... 6
1.1. Contribution to current literature ... 8
2. Literature Review ... 8
2.1. Underpricing ... 8
2.1.1. Information Asymmetry ... 9
2.1.2. Underwriter Conflicts ... 9
2.1.3. Managerial conflicts ... 10
2.1.4. Regulation and Litigiousness ... 10
2.1.5. Underwriter Reputation ... 11
2.1.6. Auditor Reputation ... 12
2.2. IPO Strategies ... 12
2.2.1. Initial Public Offerings in Hot and Cold Markets ... 13
2.2.2. Best Effort Offerings vs. Firm Commitment ... 13
2.2.3. Pricing Mechanisms ... 14
2.3. Market-based versus Underwriter-based ... 18
3. Hypothesis Development ... 19
4. Sample and Empirical Design ... 20
4.1. Sample selection and Data Retrieval ... 21
4.2. Empirical Design ... 22
4.3. Measuring Offer Price Accuracy through Initial Returns... 23
4.4. Explanation of Variables ... 24
5. Results ... 27
5.1. Descriptive Results ... 27
5.2. Results of Hypothesis Tests ... 32
6. Conclusion ... 35
5 Appendices ... 41
Appendix A: Distribution of pricing mechanisms between countries in Europe ... 41 Appendix B: Distribution of pricing mechanisms per year in Europe ... 42
6
1. Introduction
On the morning of November 7, 2013, Twitter held its initial public offering on the New York Stock Exchange. The technology company issued 70 million shares, at an initial offering price of $26 per share (Popper, 2013). By the end of the day, share price had risen to $44.90, an increase of 73%, suggesting that the company left approximately $1.25 billion on the table. Variance in initial returns is nothing new and has been studied extensively over the past decade.
Underpricing is one of the most studied topics in the IPO literature. Several explanations exist as to why the initial offer price lies below the market price. For example, investors tend to deliberately underprice offered shares in order to stimulate investor interest (Francis & Hasan, 2001). Conversely, stocks can be over valuated by the market (Swaminathan & Purnandam, 2004). Furthermore, the extent by which underpricing occurred has varied significantly over the past decades. A study done in the U.S. showed a variance of first-day return of 7% in 1980 to a staggering 65% during the internet bubble years of 1999-2000, after which underpricing
decreased to an average 4.3% over the following seven years (Bairagi & Dimovski, 2011). Although the cause of underpricing has been explained by, amongst other things,
deliberateness and overvaluation, the overall theme surrounding this phenomenon is information asymmetry. The issuing company, as well as the underwriter, is significantly more informed on the issuing companies’ health and future prospects. Potential investors, however, are limited to a preliminary prospectus provided by the issuing company. This prospectus contains the
companies’ financial information and strategy plan, the various risk factors, the use of proceeds of the offering and the dilution of other listed securities (Investing in an IPO, 2014). Additionally, it states the intended offer price or offer price range the company uses to go public. This
prospectus somewhat diminishes the risks of investing in a relatively new company and investors’ reluctance to do so. However, as these companies’ tend to have no prior reporting history, the information provided to the investors and the SEC is relatively one-sided and offers little guarantees for future endeavours (Investing in an IPO, 2014).
One of the aspects stated in the prospectus is the price mechanism used in establishing the initial offer price. Several pricing mechanisms exist in order to establish an offer price prior to an IPO. However, previous literature shows a clear preference for three of those methods. Initially, the offer price was determined by either an auction method or fixed price offering. The U.S. expanded on these options by allowing a bookbuilding mechanism. In the 1990s, this method was also introduced in Europe. A plethora of studies researched the implementation of bookbuilding mechanisms compared to auctioning mechanisms. Differences in market-based and underwriter-based pricing techniques are researched to a lesser extent and showed mixed results.
7 Furthermore, these studies did not incorporate other explanatory variables that could affect underpricing.
This study examines whether market-based pricing mechanisms can more accurately determine the fair value of the company by focusing on the initial returns following an IPO. This results in the following research question:
“Does a market-based pricing technique affect underpricing in European IPO’s?”
The questioned is answered by looking at European IPO’s in the period of 1990 to 2010. The emphasis on European IPO’s derives from the recognition that distribution of market-based pricing mechanisms has shifted in the past two decades. This shift from underwriter-based to market-based initial public offerings makes Europe an ideal testing ground. In this regard, I use the period of 1990 to 2010. After the bookbuilding technique was copied from the U.S. stock market, it quickly gained popularity and was implemented by a significantly increasing amount of issuing companies (Sherman, 2000).
This thesis finds that, for a sample of 1,023 IPO’s within the period of 1990 to 2010, bookbuilt IPO’s have an average initial return of 58.09%, compared to 59.94% for auction-priced offerings and 18.36% for fixed-price offerings. By using two ordinary least squares regression models, this study determines the effect of market-based pricing mechanisms on underpricing. In the first model, bookbuilding and auction are regressed separately, while taking into account other explanatory variables. For bookbuilding, this study finds some evidence that a positive relation exists between using a bookbuilding approach and first-day returns. Alternatively, no evidence is found that auctioning decreases underpricing. Model 2 combines both market-based approaches and compares this with fixed-price offerings. A significant positive relation is found. This study contributes to current literature by determining that an issuer benefits by using a fixed-price offer. As underpricing is lower, less money is left on the table. This will result in higher proceeds for the issuer. Alternatively, using auctioned or bookbuilt pricing mechanisms ensures all shares are sold and prevents the failure of the initial public offering.
8
1.1. Contribution to current literature
Underpricing is studied extensively, both in the U.S. and European countries. However, the used pricing mechanism is rarely taken into account. The studies that focus on the correlation between underpricing and the chosen pricing mechanism all focus on differences between market-based pricing mechanisms, i.e., auctions and bookbuilding.
This study initially extends this literature by showing the correlation between market-based pricing mechanisms, fixed-price pricing mechanisms and underpricing within European countries. Next to auctions, bookbuilding has been adopted in Europe and seen a quick growth in the amount of countries that use this technique. This added additional market-based pricing mechanism has led to a shift in the distribution of pricing mechanisms, from underwriter-based to market-based. This makes Europe an ideal testing field. Moreover, this study contributes to current knowledge by taking auditor prestige into account. Several studies have shown the effects of underwriter and auditor reputation on underpricing, but not in relationship with the used price mechanism.
The remainder of this paper is structured as follows. In chapter 2, the literature review, which discusses relevant factors that cause underpricing and the various pricing mechanisms available, as well as previous studies. Chapter 3 elaborates on the hypothesis development. Chapter 4 addresses the data retrieval and methodology. Lastly, chapter 5 and 6 provide the results and conclusion, respectively.
2. Literature Review
2.1. Underpricing
Underpricing is defined as the difference between the opening price and the first-day closing price. This is also known as the initial or first-day return. It is the pricing of an initial public offering below its market value (Su & Fleisher, 1999). If the offer price is lower than the price of the first trade, the stock is considered to be underpriced. This is usually temporarily, as supply and demand by investors will drive the stock to the market value.
The extent of underpricing, the difference between first-day opening and closing price, has varied significantly over the past two decades, especially in Chinese stock markets. Research has shown an average first-day return of 948.6% for Chinese domestic A-shares in the period between January 1, 1987 and December 31, 1995 (Su & Fleisher, 1999). Although the European stock market is not susceptible to numbers of such magnitude, underpricing is also a common characteristic of European IPO’s. The following examples are the more significant average first-day returns among European IPO’s in the past two decades. From 1996 – 2000, Germany
9 showed an average initial return of 43.42% (Goergen & Renneboog, 2003). In 2000, the U.K. stock market underpricing averaged 60.1% (Gajewski & Gresse, 2006) and even the Netherlands has shown significant numbers of 86.07% 1996 – 2000 (Goergen & Renneboog, 2003).
Considering the effect on share price and, by extension, investor participations, underpricing is studied extensively and multiple theories have been formed regarding possible causes. This chapter will reflect on previous literature and discuss the several factors that are known to create underpricing.
2.1.1. Information Asymmetry
The main theme surrounding the presence of underpricing is information asymmetry. According to Rock (1986), information asymmetry exists between the issuing firm and the market. The issuing firm discloses all ‘material information’ to investors through the prospectus. However, the firm reveals additional ‘private’ information to its underwriter. This ensures an informational advantage for the underwriter, making this company best suited for establishing an offer price. And although his information and expertise are superior to the knowledge of a single investor, it still remains lesser than the pooled information and knowledge of all investors combined. For example, some investors may possess insider information regarding a competitor or are more informed about the future discount rates of the capital market (Jenkinson & Jones, 2004). Regardless, it may seem ignorant to assume that the firm and its underwriter are better suited to establish an offer price than the market as a whole. Yet, this is generally the case when a firm is valued prior to a public offering, which results in valuation differences between the underwriter and the market.
2.1.2. Underwriter Conflicts
Aside from information asymmetry between the issuing firm and the market, which is an example of adverse selection rather than moral hazard, several studies also indicate the possibility of deliberately underpricing stock (Ruud, 1993). The less information investors acquired concerning certain IPO’s, the more dependent they are on pricing advice by an investment banker1.
However, underwriters themselves have several incentives to advise an offer price below the market-clearing price. A reduced offer price decreases the effort necessary in order to sell the issued securities (Ruud, 1993). This, by extension, reduces the risk of having to absorb unsold
10 shares. And lastly, deliberately decreasing the offer price enhances the possibility of greater initial returns, which is beneficial to its client relations.
However, based on the study by Beatty and Ritter (1986), which highly relies on the study by Rock (1986), the allowed variance on the amount of underpricing, in either direction, is rather small. Therefore, Beatty and Ritter (1986) state that underwriters are obligated to uphold a certain equilibrium. If the offer price is too high and the amount of underpricing is (relatively) small, investors refrain from buying issued shares. In contrast, if the underwriter sells shares below this equilibrium, it risks the possibility of losing potential issuing clients, due to the relatively little amount of capital gained. It even risks losing investing clients, as the underwriter appears incapable of estimating a reasonable offer price.
2.1.3. Managerial conflicts
Another theory suggests, although little evidence supports this, that management enforces certain extent of underpricing (Loughran & Ritter, 2002). Despite the amount of ‘money left on the table’, underpricing by management tends to focus more on firm control rather than gaining financially. By offering a lower IPO price, more investors are willing to buy. So instead of one relatively big company acquiring large blocks of shares, those shares will be divided over a significant larger amount of buyers. This prevents major corporations from obtaining
considerable decision making possibilities. Managerial risk will therefore be reduced and control of the company remains.
2.1.4. Regulation and Litigiousness
Another less supported theory posits that lawsuit avoidance can be an incentive for underwriters and issuing firms to enforce a decreased offer price. Practically every country has initiated
securities laws for issuers, accountants and underwriters with regards to misleading statements or excluded relevant information in the IPO prospectus. If, for example, underwriters fail or neglect to incorporate material expenses in the offer price, investors will pay more than the fair share price. If this appears to be the case, investors tend to take legal action.
This can either occur in the form of multiple parallel lawsuits or a single class action where one lead plaintiff is responsible for filing suit on behalf of all affected investors. This generally depends on the legal system in place. The legal fees associated with lawsuits are costly and they often result in the obligation to pay settlement payments. These expenses are further increased by a significant amount of time where management involvement is required to finish these matters. This prohibits them from continuing with operational management. In addition,
11 the publicity of such litigations affects the sued companies’ reputation and perceived ability to establish a fair offer price.
In order to decrease such litigation risks, firms could be incentivized to lower the offer price beforehand. If any misstatements or omissions become apparent, the investors do not have a claim since these failures were already taken into account in the IPO price (Lin, Pukthuanthong, & Walker, 2013).
2.1.5. Underwriter Reputation
Aside from the previously mentioned underwriter conflict, in which underwriters deliberately lower the offer price, underwriter reputation also affects the initial returns reported. Although underwriter conflict and auditor reputation are highly correlated, they are different. Where underwriter conflict tends to focus on the exploitation of asymmetry for profit gain and is solely dependent of the choices of the firm itself, the consequences of a poorly perceived market reputation can find their origin in factors beyond the firm’s grasp.
Surely, if underwriters are ‘caught’ underpricing their issued shares in order to decrease selling effort, rather than selling them at a fair price, this will inform investors about the firms’ integrity and investors might be reluctant to maintain their purchasing privileges with the firm. However, investment banking reputation can also suffer damage due to external reasons. If a considerable amount of the underwriters issued firms went bankrupt shortly after the IPO date, this might create the impression that the underwriter is unable to estimate future performances and ensures the loss of both investors as well as issuing clients.
Therefore, despite the firms’ intrinsic motivations with regards to establishing an IPO offer price, the outward reputation also affects the extent of underpricing. Carter, Dark & Singh (1998) provide an overview of three different underwriter ranking measures and the associated average initial returns. Of those three measures, one shows statistical significance and shows returns are less negative for IPO’s brought to the market by more prestigious underwriters. Based on the investment bank reputation, if investors are worried that the prospectus provided is incomplete or contains material misstatements, they refrain from bidding on emission shares. This will be interpreted as a decrease in demand, which will decrease the initial offer price and ultimately increase underpricing. For the sake of completeness, it is worth mentioning that the relevance of underwriter reputation differs per pricing mechanism. In the case of fixed price offerings, where the offering price is determined without consulting investor demand,
underwriter reputation is trivial. The difference between the offer price and first-day closing price remains the same, despite the credibility of the investment banker.
12
2.1.6. Auditor Reputation
With regards to previous literature, some research has been done to conclude whether auditor choice has an impact on first-day returns. Beatty (1989) and Michaely & Shaw (1995) show an inverse relation between auditor’s prestige and underpricing. IPO’s associated with the, back then, Big Eight accounting firms dealt with significantly less underpricing. However, these studies date back to the previous century. This does not imply their immediate obsoleteness, but the date of their publication should not be ignored. Especially as multiple events have occurred in the following period. Both to auditors and issuing IPO’s.
Firstly, auditor reputation is somewhat diminished compared to the 1990’s (NBA, 2012), which has decreased the difference in reputation between larger and smaller accounting firms. This has several causes. In 2001, the Enron scandal significantly raised public awareness of the possibility of auditor dependence. Arthur Andersen, one of the big accounting firms, voluntarily surrendered its licenses to practice as CPA’s in the U.S., after they were found guilty of criminal charges with regards to handling the audit of Enron, due to a conflict of interest. Furthermore, the in 2014 released report by the AFM, also had several remarks to the audit quality of the big 4 accounting firms (Autoriteit Financiële Markten, 2010).
With such events increasing the current focus on auditor credibility and its shortcomings, one can argue that the influence of auditor reputation has severely decreased over the past years. Following the knowledge that the public perception of difference in audit quality between big 4 firms and smaller local firms has declined, its hypothetically possible that first-day returns for both these firm types has levelled out. This would indicate that perceived information asymmetry is just as influential as actual information asymmetry. Auditor reputation is used in order to determine the current influence of pricing mechanisms on initial returns.
2.2. IPO Strategies
Over the past few decades, underwriters have brought several companies to the public market. This has brought about multiple strategies as to how to successfully perform an initial public offering. In this chapter, those strategies are discussed. Firstly, an overview is given of possible differences between hot and cold IPO markets. Consequently, underwriter and issuing company agreements that are possible when performing an IPO are discussed. And lastly, the different pricing methods are discussed and the extent to which they are used in European countries.
13
2.2.1. Initial Public Offerings in Hot and Cold Markets
There are always firms opting for a public issuance in order to gain capital to improve investment and operating activities. However, the amount of firms that actually undergo an initial public offering tends to vary each year. This has given rise to the notion of hot and cold initial public offering markets. Hot IPO markets have been described as having an unusually high volume of offerings, offerings with frequent oversubscription, severe underpricing and (in some cases) concentrations in particular industries (Ritter, 1984). In contrast, cold IPO markets tend to have much lower issuances, a smaller degree of underpricing and fewer cases of oversubscription.
IPO literature offers a variety of opinions whether offerings in hot and cold markets might differ, and in what aspects. By the use of signalling models, hot markets are often characterized as periods when a greater amount of high quality firms opt to go publics (Welch, 1989). According to these models, firms are drawn in hot markets because offer prices are less affected by adverse selection. Adversely, literature also indicates that long-term performance of IPO’s in hot markets are of lower quality and show worse stock returns than IPO’s in cold markets (Loughran & Ritter, 1995). Hot markets are a result of wild bullishness from irrational investors. Managers aim to take advantage of this ‘window of opportunity’ and perform an IPO.
A more recent study by Helwege and Liang (2004) finds evidence that both hot and cold IPO’s are drawn from the same industries, but that cold markets exhibit more industry
concentration. Additionally, they find that IPO’s in hot markets are more likely to have lower earnings, possess lower capital expenditures and R&D ratios, are approximately the same age at the time they go public and do not show faster sales growth of high profits in the five years after their public offering. Moreover, venture capital financing before the IPO appears to be less likely in hot markets (Helwege & Liang, 2004).
2.2.2. Best Effort Offerings vs. Firm Commitment
Practically all firms that are going public use either firm-commitment or best-efforts methods to market initial public offerings. After approval from the Securities Exchange Commission, the investment bank acquires all the issued shares from the issuer and tries to resell those shares to the public. In the case of a firm-commitment approach, the underwriter guarantees to acquire all of the issued shares, regardless of their ability to resell them to the public. Issuing firms therefore prefer firm-commitment, as it will guarantee they receive their money right away. Depending on the offering demand, the underwriter will decide if the IPO will be on a firm-commitment basis. With a higher demand firm-commitment is more likely, as the underwriter puts its own money at risk. This often results in a market out clause, where the investment banker is freed from their
14 obligation to purchase all of the securities in case the quality of the securities is impaired. For example, the market out clause can be invoked if, for a bio tech company, FDA just denied approval of the issuing companies’ new drugs.
If the underwriter chooses to take a best-efforts approach, the underwriter solely pledges to provide its ‘best efforts’. This generally entails that it will attempt to sell all shares, but is in no way obligated to purchase the securities for their own account if shares prove to sell difficultly. The lower the demand for the IPO, the more likely that the underwriter decides do to the emission on a best-efforts basis. Any bonds or shares they were unable to resell will be returned to the issuing company (Dunbar, 1998). A best-efforts approach can be further subdivided in three subcategories. A mini-maxi best efforts approach states that, after the minimum amount of sells has been reached, the investment banker may sell additional securities until the maximum amount. Collected funds will be held in escrow. If they fail to sell the minimum amount of shares, the offering will be cancelled and the collected funds will be returned to the investors. An ‘All or None’-approach (AON) specifies that collected funds will be held in escrow until all securities have been sold. If some securities remain unsold, the IPO is cancelled and the collected funds will also be returned to the investor. And lastly, a standby agreement states that the
underwriter will purchase the remaining shares. This seems highly similar to a firm-commitment basis, but a standby agreement generally gives the right to purchase the remaining shares below the market price. This may appear an advantage, but remaining shares indicate a lack of demand.
2.2.3. Pricing Mechanisms
Both IPO’s on a firm-commitment basis as well as those that are issued on a best-efforts base have several methods to establish an initial offer price. For the remainder of this chapter, those pricing mechanisms are discussed.
In general, issuing firms and their investment bankers often have the option to determine an initial offer price through either a fixed-price offering, an auction or through the relatively new method called bookbuilding.
2.2.3.1. Fixed-Price Offering
In the case of fixed-price offerings, market prices are established prior to the actual sale of the shares. If demand exceeds the offered quantity of shares, they will be prorated among all bidders or randomly rationed. With insufficient demand, issuing companies tend to cancel the IPO or postpone it. The offer price is determined without consulting investors and is therefore solely based on the issuing company and underwriters’ knowledge of the company value. Considering
15 institutional investors are generally more experienced and have a significant informational
advantage, this results in a problem (Zhang, 2008). Investors with low expectations and retail investors shall obtain larger share blocks if the share value is at a lower level. Because a fixed-price approach does not allow fixed-price adjustments in order to conform to demand, underwriters are forced to set the fixed market price below the expected market value to convince the participation of such investors (Rock, 1986). Otherwise, shares are less likely to be fully subscribed, which could result in an IPO failure.
Fixed-price offerings exist in all European countries except Austria, Greece, Finland and Spain. The UK uses a different terminology to refer to fixed-price offerings. ‘Offer for
subscription’ if new funds are raised and ‘offer for sale’ if not (Gajewski & Gresse, 2006). However, a decreasing number of IPO’s still use fixed-price offerings.
2.2.3.2. Auctions
Auctions or tender offerings are pricing mechanisms where the issuing company predetermines a minimum price for which it is willing to sell its securities. Consequently, both individual and institutional investors are invited to place subscription orders. These subscription orders allow them to order a certain amount of shares at a price equal or above the minimum price. Based on these orders, a demand curve is constructed. The auctioneer then uses this curve to set an issue price, equal to all investors after which the shares are allocated on a pro rata basis.
Throughout Europe, several variations on the auction approach exist. They possess the same basic characteristics, but differ in other aspects. For example, some auctions are open exclusively to institutional investors, while others also include the secondary market.
Furthermore, the so-called Dutch auction uses a descending order of price limits. The orders are cleared in decreasing order until the number of offered shares are cleared. This ensures to perfectly equate supply and demand, as investors that consider the values to be more valuable, will pay a higher price. The lowest price becomes the equilibrium price. In contrast, a uniform-price mechanism serves all orders at a uniform-price equal to the equilibrium uniform-price.
Pure Dutch-auctions are rare. Tender offerings generally use an uniform-pricing technique in order to determine a market price. Furthermore, IPO auctions tend to be
maximized, which entails that a maximum price is set and orders above that limit are prohibited. Additionally, the auctioneer has the ability to disregard share orders which lie significantly below the minimum price.
With regards to auction usage in European countries, they have seen a considerable decline. In Germany and Switzerland, auctions were common. However, they have disappeared
16 in the last decades. The few European countries where this pricing technique still remains are the Netherlands, Poland, the UK, Portugal and France (Gajewski & Gresse, 2006).
2.2.3.3. Bookbuilding
Also known as ‘placing’ in the UK and ‘placement’ in France, bookbuilding refers to the process of collecting orders from fund managers in order to establish an offer price. This has seen a significant increase in European countries since 1995 (Sherman, 2005). Shares are offered exclusively to institutional investors. The underwriter determines a price range for the share orders to be placed. Institutional investors are then invited to place orders for prices within this range, usually with a subscription period of one week.
For example, suppose company XXX decides to offer 3,000 shares at a price band of CU 20-24. Bids are offered in the following manner (Table 1):
Naturally, the intent is to sell at the highest price of 24. Yet, at that price only 500 shares will be bought. Consequently, the underwriter lowers the price by one CU, which will remain
unsuccessful in selling al shares. At the price of 22 CU, the underwriter will be able to allocate all shares, and this will therefore be the cut-off price. All bids above this price level are considered legal bids. This is known as the price discovery mechanism of the book building process, and shows how the offer price of most initial public offerings are determined recently.
Where bookbuilding significantly differs from auctions, is the period prior to the subscription week. In preparation for establishing a price range, the issuing companies’ management perform a road-show, which can last for two weeks. Based on research reports stating the value of the company, provided by the underwriters’ analysts, managers approach institutional investors and present them with data surrounding the IPO. This presentation of securities to potential buyers allows them to gauge the market and measure the acceptance of the
Bid Quantity Bid Price Cumulative Quantity Subscription 500 24 500 16.67% 1,000 23 1,500 50.00% 1,500 22 3,000 100.00% 2,000 21 5,000 166.67% 2,500 20 7,500 250.00%
17 new issue. Additionally, it generates excitement and is often critical to the success of the initial public offering (Gajewski & Gresse, 2006). When the subscription period is over, a review of the orders shows a definitive issue price, which is maintained for all interested investors. If the amount of subscriptions exceeds the amount of shares offered, the investment banker is responsible for the allocation of the shares.
But the key distinguishing feature of bookbuilt IPO’s, as opposed to auctions or public offers, is that the investment bank that underwrites the issue, has complete discretion in
allocating the shares. In public auctions, everyone is allowed to participate and the shares will go to the highest bidder. In bookbuilt offerings however, the underwriter, and in particular the bookrunner, has authority in deciding who receives shares. According to a study by Jenkinson & Jones (2004), there are three views on how this discretion is exercised.
Firstly, allocations can be used to incentivize investors to reveal truthful information useful in valuing the price of the security. In these instances, certain investors possess
information, and the investment bank uses its discretion over allocation to essentially to reward investors for revealing this information. Conversely, information about the issuing company can be costly to obtain. Investors face a moral hazard because they know that the offer price will be calculated using all available information. This entails that, rather than purchasing their own signal, investors may to be tempted to free-ride on the information of others (Sherman & Titman, Building the IPO Order Book: Underpricing and Participation Limits with Costly Information, 2002). If the expected level of underpricing exactly compensates for the costs of acquiring information, investors will be indifferent about whether or not to participate in the offering. However, if investors expect excess returns from participating in new issues, they will cooperate.
The second view, which tends to be heavily emphasized by investment bankers, is that share allocations are preferably directed toward to investors who intend to be long-term holders of the stock, as opposed to ‘flippers’. Both formal and informal justifications have been advanced to explain this preference. The most obvious of which, is the tendency for flippers to depress the trading price in the immediate aftermarket. Especially in relatively weak offerings there may be concerns that flipping could lower the trading price below the stock price (this proved true with regards to the Twitter IPO).
Lastly, the final view relates allocation to the generation of subsequent trading commissions for the investment bank. Loughran and Ritter (2002) provide some general evidence that confirms this profit-seeking view.
18 Although bookbuilding is currently present in all European countries, the majority tends to engage in a double stage issue where the secondary market can be included in the
bookbuilding process. The most common method to achieve this, is to offer shares to the secondary market via a fixed price equal to the equilibrium price.
2.3. Market-based versus Underwriter-based
The global increase in bookbuilding usage has given rise to various studies focusing on its effect on underpricing. However, considering its similarity to auction-priced IPO’s, the clear major of these studies are aimed at providing differences between the two market-based approaches. Relatively little literature compares underpricing differences between market-based techniques and the underwriter-based fixed-pricing approach. The studies that focus on these differences are discussed in this paragraph.
Firstly, Benveniste and Busaba (1997) compare fixed priced methods and American bookbuilding from a theoretical perspective. They use a setting where investors obtain the same information and are able to see the subscription decisions of other investors. They find that either method can be optimal, depending on the risk attitude and the size of issuing firm. More specifically, they find that relatively smaller issuances are more likely to utilize a fixed-price approach to avoid the cost of the bookbuilding process. Furthermore, they state that pricing technique regulation can give rise to inefficiencies.
Continuing on this research, Busaba and Chang (2010) focus on the ability of informed IPO investors to trade on their own information in the aftermarket. If informed investors aim to generate profits in the aftermarket, they require an increased extent of underpricing, for both fixed-priced as bookbuilt offerings. With bookbuilding, this creates a paradox. Investors are allocated shares in return for truthful information about the issuing firm, but profit from negative information. This incentivizes investors to deliberately provide bad information, in order to lower the offer price and gain more in the aftermarket. Busaba and Chang (2010) find that only when a small group of investors secure the majority of the issue, underpricing is lower with bookbuilt offerings. This is due to the inability to gain from trading in the aftermarket.
Furthermore, a study by Jagannathan & Sherman (2006) attempts explain why, by looking at the Singapore IPO market, auction-priced offerings are abandoned in favour of other
methods. They find evidence suggesting the presence of free riders that place unrealistically high bids. In the case of a uniform price auction, they will end up paying the lowest price provided by other investors, essentially relying on other bidders to perform due diligence and engage in price discovery. This led to average initial returns of 4.6 percent for auctions, compared to 36.9 percent
19 for fixed offerings. As a result, investors lost money on auction-priced IPO’s in Singapore and issues became undersubscribed. Therefore, issuers have come to prefer fixed-price offerings in order to maximize proceeds.
Also comparing auctioned offerings with those that use a fixed-price, Zhang (2008) finds similar results. They find that the market price, as established in uniform price auctions, lies higher than the price used by underwriters. As a result, compared to fixed-priced offerings, underpricing is less frequent and less severe. This entails that the issuing firm receives more revenue in uniform price auctions. The increased market price is explained by the notion that bidders with a high expected value of the company tend to bid more aggressively, either by using higher prices or attempting to obtain a larger amount of shares.
Lastly, Kaneko & Pettway (2003) performed an ordinary least squares regression to compare investor-priced IPO’s using price-auctions with those using bookbuilding in Japan. They find that, on average, initial returns for 481 auction-priced IPO’s listed on the OTC market was 11.4%. Initial returns for bookbuilt underwriter-priced IPO’s averaged 48% over 469 listed companies on the OTC stock exchange. Market return, expected issue size and the firm age prior to the IPO are most significantly related to the initial returns. Additionally, according to Kaneko & Pettway (2003), one of the main reasons for this voluminous underpricing is the tendency for underwriters to set the upper limit of the initial price range too low. The vast majority of the bookbuilt issues are priced at the upper limit, which indicates that investors are likely willing to accept higher prices. They conclude that underwriters generally favour their investors, not the issuing firms.
3. Hypothesis Development
From previous literature I can infer that market-based offering techniques lead to different degrees in underpricing compared to fixed-price offerings. However, no previous literature has regressed these market-based pricing mechanisms against the previously dominant fixed-price offering technique, while taking into account firm characteristics that are known to affect first-day returns. This is the first study to test whether incorporating market information in the offer price affects underpricing, while controlling for other factors.
In the case of fixed-price offering, market information is not at all incorporated in the offer price. In the case of bookbuilding, a limited amount of market information is incorporated in the offer price, as institutional investors, who tend to be more experienced and better
informed, have the ability to participate in the offering process. Based on the suggested
20 to higher underpricing, as investors intentionally provide negative information in order to lower the offer price. This leads to the following hypotheses:
𝐻1 = 𝑈𝑠𝑖𝑛𝑔 𝑎 𝑏𝑜𝑜𝑘𝑏𝑢𝑖𝑙𝑑𝑖𝑛𝑔 𝑝𝑟𝑖𝑐𝑖𝑛𝑔 𝑚𝑒𝑐ℎ𝑎𝑛𝑖𝑠𝑚 𝑠𝑖𝑔𝑛𝑖𝑓𝑖𝑐𝑎𝑛𝑡𝑙𝑦 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑒𝑠 𝑢𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔 𝑐𝑜𝑚𝑝𝑎𝑟𝑒𝑑 𝑡𝑜 𝑓𝑖𝑥𝑒𝑑 − 𝑝𝑟𝑖𝑐𝑒 𝑝𝑟𝑖𝑐𝑖𝑛𝑔 𝑚𝑒𝑐ℎ𝑎𝑛𝑖𝑠𝑚𝑠
𝑤𝑖𝑡ℎ𝑖𝑛 𝐸𝑢𝑟𝑜𝑝𝑒𝑎𝑛 𝑚𝑎𝑟𝑘𝑒𝑡𝑠
In the case of open-auctions, all investors, institutional or not, have the ability to participate based on their own research on the financial data. Adhering to previous literature, stating that auctioning decreases underpricing due to the free rider issue and aggressive bidding, I assume a single direction in composing the second hypothesis:
𝐻2 = 𝑈𝑠𝑖𝑛𝑔 𝑎𝑛 𝑎𝑢𝑐𝑡𝑖𝑜𝑛𝑖𝑛𝑔 𝑝𝑟𝑖𝑐𝑖𝑛𝑔 𝑚𝑒𝑐ℎ𝑎𝑛𝑖𝑠𝑚 𝑠𝑖𝑔𝑛𝑖𝑓𝑖𝑐𝑎𝑛𝑡𝑙𝑦 𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝑠
𝑢𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔 𝑐𝑜𝑚𝑝𝑎𝑟𝑒𝑑 𝑡𝑜 𝑓𝑖𝑥𝑒𝑑 − 𝑝𝑟𝑖𝑐𝑒 𝑝𝑟𝑖𝑐𝑖𝑛𝑔 𝑚𝑒𝑐ℎ𝑎𝑛𝑖𝑠𝑚𝑠 𝑤𝑖𝑡ℎ𝑖𝑛 𝐸𝑢𝑟𝑜𝑝𝑒𝑎𝑛 𝑚𝑎𝑟𝑘𝑒𝑡𝑠
Lastly, I provide a general comparison by combining both market-based techniques with fixed-price offerings. As no prior literature exists that compares market-based and fixed-fixed-price offerings, I assume no particular direction. The final hypothesis is therefore as follows:
𝐻3 = 𝑈𝑠𝑖𝑛𝑔 𝑎 𝑚𝑎𝑟𝑘𝑒𝑡 − 𝑏𝑎𝑠𝑒𝑑 𝑝𝑟𝑖𝑐𝑖𝑛𝑔 𝑚𝑒𝑐ℎ𝑎𝑛𝑖𝑠𝑚 𝑠𝑖𝑔𝑛𝑖𝑓𝑖𝑐𝑎𝑛𝑡𝑙𝑦 𝑎𝑓𝑓𝑒𝑐𝑡𝑠 𝑢𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔 𝑐𝑜𝑚𝑝𝑎𝑟𝑒𝑑 𝑡𝑜 𝑓𝑖𝑥𝑒𝑑 − 𝑝𝑟𝑖𝑐𝑒 𝑝𝑟𝑖𝑐𝑖𝑛𝑔 𝑚𝑒𝑐ℎ𝑎𝑛𝑖𝑠𝑚𝑠
𝑤𝑖𝑡ℎ𝑖𝑛 𝐸𝑢𝑟𝑜𝑝𝑒𝑎𝑛 𝑚𝑎𝑟𝑘𝑒𝑡𝑠
4. Sample and Empirical Design
This chapter focuses on the data and methods used to test whether market-based pricing methods for European IPO’s affect first-day returns. The first paragraph gives an overview of the selected data and how it is retrieved. Secondly, in the next paragraph, the methodology will be discussed. Additionally, the equation regressions are provided and what proxies are used to measure the influence market-based pricing mechanisms have on underpricing. Finally, the different variables will be explained and what their relevance is in regards to answering the research question.
21
4.1. Sample selection and Data Retrieval
The chosen timeframe of this research ranges from 1990 to 2010, given the rise in bookbuilding in European countries. Moreover, all IPO’s in Europe are taken into account.
In order to collect all the required data, several data sources have been consulted. Initially, an overview of all the IPO’s that occurred in this time period are obtained from the Thomson One Banker database. This resulted in 9667 IPO’s that were conducted on European stock exchanges. Of these, 1392 companies were omitted, because no pricing technique was provided. This includes companies where solely a Best Efforts or Firm Commitment approach was shown. It indicates the relation between underwriter and issuer, but offers no information on how the offer price is established. For a complete overview of the distribution of pricing mechanisms over the various European countries, see appendix A. Offer prices, issued shares and first-day closing prices were also provided by Thomson One. Remaining firm characteristics were retrieved from Thomson DataStream, which allows insight in insider information, stock prices and macro-economic data. Data regarding firm age, the company’s auditor and market return were obtained and linked to the IPO’s from Thomson One by matching ISIN-codes. The founding years were manually retrieved from prospectuses and webpages after establishing a sample.
Because of missing ISIN-codes, firm characteristics of 4039 companies were impossible to retrieve and are therefore excluded from the sample. Additionally, observations with
insufficient data were removed from the sample. Considering the relative incompleteness of European databases, a rather large amount had to be excluded. The remaining sample consists of 1024 observations.
Of these 1024 observations, 3 were considered to be outliers based on the 95th percentile, 1 with regards to offer price, 1 due to an extremely high initial return and 1 with excessive proceeds. The offer price outlier was caused by the use of Denmark Krones, which was recalculated to euro’s using todays’ currency exchange rate. Recalculating the offer price also entailed that the proceeds were recalculated, considering these are based on the offer price. The initial return outlier was winsorized on the 95 percent level to ensure it would not distort the sample. And lastly, I removed the observation with the excessive proceeds. This concerned the IPO of the Russian diamond mining ALROSA. Although there is no reason to assume the data is incorrect, in order to ensure the validity of the statistics, the observation was removed, leaving a final sample of 1023 observations, spread over 25 countries.
22
4.2. Empirical Design
For the sake of completeness, I have made several assumptions prior to establishing a regression equation. Firstly, Benveniste & Wilhelm (1990) analysed the effects of a multi-price rule for book building. Underwriters are allowed to charge different prices to different investors. This means that, with the issuance of share from a single company, different initial returns could arise as some investors obtained the shares at a price closer to the first-day closing price. However, due to the limited data available I will use a single-price approach. This will generally be the price with which a company enlists on the public exchanges. This will cover the majority of the shares issued.
Additionally, managerial and underwriter conflicts are assumed to be non-existent. As previously stated, managerial conflicts refer to the notion that a lower price invites more investors and therefore company control is spread over more stockholders. Furthermore,
essentially relying on the presence of information asymmetry, Ruud (1993) indicated that there is a possibility that underwriters intentionally offer shares below a fair offer price in order to boost sales and improve client relations, which inherently results in more underpricing. In composing the equation regression, these aspects are not taken in to account as it is impossible to derive the incentives of management and investment banks.
Lastly, I assume there are no variances between hot and cold IPO markets. Although an abundance of literature focuses on the differences between hot and cold IPO markets, evidence suggests they differ predominantly on firm characteristics and long-term performance (Loughran & Ritter, 1995). No direct influence is shown of the kind of market on initial returns. All the characteristics that are known to be affected by the state of the market are controlled for (see next paragraph). In running my ordinary least squares regression, I therefore build on the
premises that initial returns are not directly affected by whether the IPO is performed in a hot or cold market.
23
4.3. Measuring Offer Price Accuracy through Initial Returns
To conclude whether market-based pricing methods result in offer prices closer to the market value, I use first-day returns (FDR). This is the variance between the price with which a security is offered to the market and the first-day closing price. The market value a securities’ worth on the open market. In the case of a fixed-price offering however, until a security is actually offered to the public, the market has no input in its valuing. After the first day of trading, the market has contributed and reached an equilibrium on the right price. This is what the security is actually worth. Which entails that, the higher the amount of underpricing, the less accurate the
underwriter and issuer were in their determination of the offer price. In order to test whether a market-based approach, where the market is (partly) taken into consideration, affects the extent of underpricing, two regression analyses will be performed. Relying on the study performed by Kaneko & Pettway (2003), which performs a regression analysis on auctioned versus bookbuilt IPO’s, similar variables are used. In the first model, bookbuilding and auction will be regressed separately against the first-day results, taking other explanatory variables into account. This model is used to test hypothesis 1 and 2 and results in the following equation:
Model 1:
𝐹𝐷𝑅 = 𝛼 + 𝛽1∗ 𝐴𝑈𝐶𝑇𝐼𝑂𝑁 + 𝛽2∗ 𝐵𝑂𝑂𝐾𝐵𝑈𝐼𝐿𝐷 + 𝛽3∗ 𝐴𝐺𝐸 + 𝛽4∗ 𝑂𝐹𝐹𝐸𝑅 + 𝛽5 ∗
𝑃𝑅𝑂𝐶𝐸𝐸𝐷𝑆 + 𝛽6∗ 𝐴𝑈𝐷𝐼𝑇𝑂𝑅 + 𝛽7∗ 𝑀𝐴𝑅𝐾𝐸𝑇 + 𝛽8∗ 𝑌𝐸𝐴𝑅 + 𝛽9∗ 𝐶𝑂𝑈𝑁𝑇𝑅𝑌 +
ε
To test hypothesis 3, a second model is used, pooling together auction-priced and bookbuilt offerings. This provides a general comparison of market-based offerings versus fixed-priced offerings. This model as follows:Model 2:
𝐹𝐷𝑅 = 𝛼 + 𝛽1∗ 𝑀𝐵𝑃𝑀 + 𝛽2∗ 𝐴𝐺𝐸 + 𝛽3∗ 𝑂𝐹𝐹𝐸𝑅 + 𝛽4∗ 𝑃𝑅𝑂𝐶𝐸𝐸𝐷𝑆 + 𝛽5∗ 𝐴𝑈𝐷𝐼𝑇𝑂𝑅 + 𝛽6∗ 𝑀𝐴𝑅𝐾𝐸𝑇 + 𝛽7∗ 𝑌𝐸𝐴𝑅 + 𝛽8∗ 𝐶𝑂𝑈𝑁𝑇𝑅𝑌 +
ε
24
4.4. Explanation of Variables
In this paragraph, the variables that are relevant for the equation regressions are explained.
(Dependent Variable) IPO First-day Returns (FDR)
The first-day return, or the initial return, is the percentage change of the stock price on the first day of trading. It is the difference between the offer price and the first day closing price. The use of percentages ensures potential currency differences are accounted for. Although the majority of the data is to the standard European currency, unavailable exchange currencies at IPO date has led to the offer price being stated in a nations respective currency unit.
𝐹𝐷𝑅 = 𝐹𝑖𝑟𝑠𝑡 𝑑𝑎𝑦 𝐶𝑙𝑜𝑠𝑖𝑛𝑔 𝑃𝑟𝑖𝑐𝑒 𝑂𝑓𝑓𝑒𝑟 𝑃𝑟𝑖𝑐𝑒 − 1
(Variable of Interest)
Pricing Technique (AUCTION; BOOKBUILD; FIXED; MBPM)
This is the main independent variable used to determine whether market-based pricing
mechanisms result in various degrees of underpricing. Three possibilities can occur: bookbuilt offerings (market-based), auctioned offerings (market-based) and fixed-price offerings (not market-based). This essay focusses on the effects of both market-based mechanisms compared to the fixed-price approach. In order to make statistical analysis possible for the used pricing
mechanism, dummy variables will be created. For model 1, each observation is given a 1 for bookbuilding, if they use bookbuilding approach, a 1 for an auctioning approach, if it concerns an auctioned offering and a 1 for fixed, if it regards a fixed-price approach. Allocating dummy variables to all pricing mechanisms is necessary to provide descriptive results. With the ultimate regression analysis, fixed offerings are excluded and embedded in the intercept. Hypothesis 1 and 2 are supported, if a significant regression coefficient in the correct direction is present for either using an auction-priced or bookbuilt method, as this method affects underpricing compared to fixed-price offerings. For testing hypothesis 3, using model 2, both auctioned and bookbuilt IPO’s are pooled together and regressed versus fixed-price offerings. If the analysis shows a significant regression coefficient in either direction, hypothesis 3 is supported.
25 (Control Variables)
Firm Age (AGE)
Beatty & Ritter (1986) have provided empirical evidence that there is a positive relation between ex ante uncertainty about an initial public offerings’ value and its initial return. A firms’ age is highly associated with the perceived risks. Younger firms tend to be associated with more risk. Additionally, they are generally less prepared for the consequences of going public and the necessary requirements. Firm age is determined by calculating the difference between the
founding year and the year of the IPO. I expect firm age and initial returns to be inversely related.
Offer Price (OFFER)
The offer price can be directly inferred from the retrieved data and indirectly provides some insight in the future prospects of the firm. A lower offer price tends to point towards greater ex ante uncertainty, as these companies are generally lower rated. A lower offer price will therefore increase underpricing.
IPO Proceeds (PROCEEDS)
Calculated by multiplying the number of shares offered times the offer price. This is relevant because higher gross proceeds indicate a larger firm size. This is generally associated with less risk, which consequently leads to lower underpricing (Beatty & Ritter, 1986).
Auditor Reputation (AUDITOR)
Previous literature has already proved auditor reputation affects the amount of underpricing. If the public market is used to set an offer price, poor auditor prestige can damage the validity of the financial statements and prospectuses. Consequently, if investors assume the financial statements possess several omissions and misstatements, they will try to lower offer price to compensate for possible future announcements. Michaely and Shaw (1995) and others report that auditor size and/or reputation is a valuable signal of investment quality and risk in IPO’s. The higher the auditor quality, the lower the initial return levels.
With regards to auditor reputation, I will focus on the traditional Big 5/non-Big 5 dichotomy (also taking in account Arthur Andersen, who audited multiple IPO’s prior to its bankruptcy in 2001). Auditor reputation is indicated by using a dummy variable (0 for non-Big 5 Firms, 1 for Big 5 Firms), creating a distinction between Big 5 Accounting firms and other external auditing partners. Considering the increased reputation associated with Big 5 accounting firms, I expect an inverse relation between Big 5 firms and initial returns.
26 Market Return (MARKET)
The last control variable which will be addressed is the market return. Due to overall market changes, the share price is sure to rise or decline, despite being influenced by the aforementioned control variables. In order to control for this average market return, the returns of the relevant stock exchange will be calculated. This will be done by reducing the end of day closing index with the opening day index. If securities trade on multiple stock exchanges, the first one provided by DataStream is used in order to find the accompanying market return. I expect market return and first-day returns to be positively related.
27
5. Results
5.1. Descriptive Results
Table 2 reports descriptive statistics for the various variables used in the regression.
Table 2: Descriptive statistics: Variable characteristics
Variables N Minimum Maximum Mean Std. Deviation Skewness Kurtosis
FDR 1023 -0.99 3.56 0.5517 0.51512 0.035 3.447 FIXED 1023 0.00 1.00 0.0723 0.25917 3.307 8.952 AUCTION 1023 0.00 1.00 0.3118 0.46347 0.814 -1.341 BOOKBUILD 1023 0.00 1.00 0.6158 0.48663 -0.477 -1.776 MBPM 1023 0.00 1.00 0.9277 0.25917 -3.307 8.952 OFFER 1023 0.50 480.64 12.7108 26.66643 7.934 105.9 PROCEEDS 1023 1710.00 9956084670.00 228551353.1 636465391.5 7.606 80.115 AUDITOR 1023 0.00 1.00 0.7331 0.44254 -1.056 -0.887 MARKET 1023 -810.60 252.2 -0.8126 52.10370 -4.130 62.403 AGE 1023 0.00 228.00 22.9795 35.18615 2.907 9.306
Table 2 provides some variable characteristics, such as the mean, the minimum and maximum, the standard deviation and the skewness and kurtosis in order to determine the distribution of the variable. This is more for the sake of inferring information, as normal distribution of the variables is not required for a valid regression model.
Regarding the offer price, winsorizing major outliers has resulted in a more representative number of the average offering price, with a mean of 12.71. The positive skewness implies an asymmetrical distribution, with a long tail to the right. This is in accordance with expectations, as it is reasonable to assume that offer price is not normally distributed. As offering price cannot be below zero, but can be any integer above zero, it is logical that the distribution is skewed to the left. The kurtosis however, is somewhat surprising. The extremely high kurtosis implies all offer prices lay in a relatively narrow range. A positive kurtosis indicates that a large amount of the observations lie close to the mean, rather than being spread out. In this regard, I expected a normal, or even negatively kurtosed, distribution. IPO’s of higher valued companies could result in higher offer prices, especially in the case of auction approaches. These tend to be more stable firms and provide less risk to investors, which would make it an appealing investment possibility. Auctions among investors drive up the amount they are willing to pay, resulting in an increased offer price. Yet, the descriptive results contradict this assumption. This means issuers and underwriters prefer increasing the amount of shares in order to decrease offer price, which
28 provides evidential basis for the existence of managerial and underwriter incentives to lower offering price in order to ease sales or spread company control.
Ideally, a mean of zero would be shown for initial returns (FDR). If so, the price established by the underwriter equals the fair value provided by the market equilibrium of offer and demand. But this is rarely the case. In nearly all IPO’s, some extent of underpricing is present, which is also visible here. The mean of 0.5517 suggest that, on average, stock prices increase by 55.17% on the first day after being offered to the market. This is in accordance with expectations, as well as results of prior studies (Sherman, 2005; Swaminathan & Purnandam, 2004).
Age and market returns vary quite extensively. But considering the relatively high amount of used markets, this was to be expected. Furthermore, the distribution of offering techniques is shown. Table 3 provides an overview for the distribution of pricing mechanisms among the various European countries, using this sample. Evidence shows that the bookbuilding approach is the preferred method to establish the offer price (61.58%), followed by auctions (31.18%) and fixed-price offerings (7.23%). The larger variation in the variable values of the sample should contribute to creating a more significant regression model.
29 As indicated, one of the main reasons for choosing this timeframe was the assumed increase in market-based IPO’s, which entails that distribution of pricing mechanisms has changed over the past decades. The overview provided by table 4 shows the amount of IPO’s and corresponding pricing mechanism per year. Although bookbuilding was introduced in Europe prior to 1990, the overview shows that companies using bookbuilding to establish an offer price has increased. The low amounts of IPO’s in early years is partly due to limitations on behalf of the database, especially with regards to firm characteristics. Yet, there is no reason to believe that this has skewed the distribution. Except for drawbacks in 2005 and 2006, it is apparent that the majority of the offer prices have been established by market-based bookbuilding methods. Country Total 1990 240,00% 0 0,00% 3 60,00% 5 1991 360,00% 0 0,00% 2 40,00% 5 1992 150,00% 0 0,00% 1 50,00% 2 1993 0 0,00% 0 0,00% 5 100,00% 5 1994 0 0,00% 1 20,00% 4 80,00% 5 1995 218,18% 1 9,09% 8 72,73% 11 1996 0 0,00% 0 0,00% 8 100,00% 8 1997 228,57% 1 14,29% 4 57,14% 7 1998 228,57% 0 0,00% 5 71,43% 7 1999 233,33% 1 16,67% 3 50,00% 6 2000 0 0,00% 11 78,57% 3 21,43% 14 2001 0 0,00% 19 95,00% 1 5,00% 20 2002 315,00% 17 85,00% 0 0,00% 20 2003 513,16% 32 84,21% 1 2,63% 38 2004 29 22,31% 100 76,92% 1 0,77% 130 2005 115 63,89% 62 34,44% 3 1,67% 180 2006 113 46,89% 115 47,72% 13 5,39% 241 2007 37 17,70% 170 81,34% 2 0,96% 209 2008 2 5,88% 30 88,24% 2 5,88% 34 2009 1 3,57% 27 96,43% 0 0,00% 28 2010 0 0,00% 43 89,58% 5 10,42% 48 Total 319 31,18% 630 61,58% 74 7,23% 1023 Auction Bookbuild Fixed
Pricing Mechanism Country Total Austria 7 50,00% 7 50,00% 0,00% 14 Belgium 8 40,00% 11 55,00% 1 5,00% 20 Bulgaria 2 100,00% 0,00% 0,00% 2 Cyprus 0,00% 3 75,00% 1 25,00% 4 Czech Republic 1 100,00% 0,00% 0,00% 1 Denmark 2 14,29% 11 78,57% 1 7,14% 14 Estonia 0,00% 1 100,00% 0,00% 1 Finland 4 57,14% 3 42,86% 0,00% 7 France 67 36,41% 106 57,61% 11 5,98% 184 Germany 24 32,43% 42 56,76% 8 10,81% 74 Gibraltar 1 50,00% 1 50,00% 0,00% 2 Greece 1 5,56% 7 38,89% 10 55,56% 18 Hungary 0,00% 0,00% 1100,00% 1 Ireland 6 27,27% 13 59,09% 3 13,64% 22 Italy 21 33,87% 41 66,13% 0,00% 62 Luxembourg 1 14,29% 3 42,86% 3 42,86% 7 Netherlands 6 20,00% 14 46,67% 10 33,33% 30 Norway 0,00% 1 50,00% 1 50,00% 2 Portugal 2 50,00% 2 50,00% 0,00% 4 Russia 2 20,00% 6 60,00% 2 20,00% 10 Spain 7 50,00% 7 50,00% 0,00% 14 Sweden 10 38,46% 14 53,85% 2 7,69% 26 Switzerland 13 34,21% 21 55,26% 4 10,53% 38 Ukraine 0,00% 2 100,00% 0,00% 2 United Kingdom 134 28,88% 314 67,67% 16 3,45% 464 Total 319 31,18% 630 61,58% 74 7,23% 1023
Auction Bookbuild Fixed Pricing Mechanism
30
Table 5: Average initial returns per pricing mechanism variable
Table 5 gives an overview of the main characteristics regarding initial returns for the various pricing mechanisms. The average initial returns for fixed-price offerings is 0.1836, which implies that, when isolating the fixed-price offerings, the share price rises by 18.36% in the first day following the public offering. For the market-based pricing mechanisms auctioning and bookbuilding, the average initial returns are 57.94% and 58.09%, respectively. These are
differences of 39.58% and 39.73%, compared to using a fixed offer price. When focusing on the pooled dummy variable for both market-based pricing mechanisms (MBPM), used for model 2, average initial returns are 58.04%, which differs 35.72% from fixed-price offerings.
In general, since the ordinary least squares regression is not sensitive to variables which are not normally distributed, this data set qualifies for an OLS regression.
Method N Mean Std. Deviation Std. Error Mean
FDR
FIXED 74 0.1836 0.52189 0.06067
AUCTION 319 0.5794 0.52565 0.02943
BOOKBUILD 630 0.5809 0.49261 0.01963
31 Table 6 provides the Pearson correlation matrix.
Table 6: Pearson Correlation Matrix
OF FE R PR OC E E DS FDR A UDI T OR D M A R KE T A GE A UC T ION B OOKB UIL D FI XE D OFFER 1 PROCEEDS 0.32 1 (0) FDR -0.502 -0.235 1 (0) (0) AUDITORD 0.128 0.16 -0.206 1 (0) (0) (0) MARKET 0.027 -0.141 0.042 -0.035 1 (0.383) (0) (0.174) (0.262) AGE 0.088 0.122 -0.117 0.078 -0.027 1 (0.005) (0) (0) (0.012) (0.394) AUCTION 0.035 0.013 0.036 -0.028 -0.019 0.08 1 (0.267) (0.666) (0.247) (0.371) (0.543) (0.01) BOOKBUILD -0.057 -0.058 0.072 -0.036 0.063 -0.076 -0.852 1 (0.069) (0.064) (0.022) (0.253) (0.045) (0.015) (0) FIXED 0.045 0.085 -0.2 0.117 -0.084 0 -0.188 -0.354 1 (0.152) (0.007) (0) (0) (0.007) (0.993) (0) (0)
The Pearson correlation test calculates the separate correlation coefficients between the multiple variables used in the regression. This is therefore a univariate analysis. The first number is the correlation coefficient and the number between parentheses refers to its value. Significant p-values (below 0.01) are shown in bold. As can be seen, multiple key variables are significantly correlated on the 0.01 correlation level. However, the only actual issues regarding
multi-collinearity arises between the dummy variables for auction and bookbuilding pricing methods. This is rather self-evident, as the majority of this sample either uses an auction or bookbuilding pricing technique. Therefore, if an observation did not use a bookbuilding approach, it is likely an auction approach was used instead (dummy variable trap). Multi-collinearity is a phenomenon where two or more variables in a multiple regression model are highly correlated, to the extent
32 that one of the variables can be linearly predicted by another variable. Multi-collinearity does not reduce the predictive power of the model as a whole, but solely affects inferences about the predictive powers of single variables. As I focus on the predictive powers of certain variables, this is something to take into account. Issues of multi-collinearity of variables could arise when the correlation coefficient between two independent variables exceeds 0.8 (positively correlated), or is beneath -0.8 (inversely correlated). In this table, there exists inverse correlations between the various pricing mechanisms. This entails that it is possible to predict if a certain pricing mechanism is used, based on whether another pricing mechanism is used. This makes sense, considering the limited amount of options.
I also expected a high degree of correlation between the offer price and a firms proceeds, considering the latter is calculated by multiplying the offer price with the amount of shares issued. And although it shows a higher correlation coefficient than average, it does not imply multi-collinearity, which entails that variance in the number of issued shares is large enough to ensure that no relation can be drawn.
5.2. Results of Hypothesis Tests
My hypothesis focuses on the extent to which market-based pricing methods, i.e., auctioned and bookbuilt offerings, affect first-day returns after an IPO, compared to a fixed-price method. Using an ordinary least squares regression, the results are shown in table 6. The predicted sign, if shown, indicates the expected direction of the explanatory variable. According to hypothesis 1, a positive relation is expected between bookbuilding and underpricing. According to hypothesis 2, an inverse relation is expected between auctioning and underpricing. Hypothesis 3, where both market-based pricing mechanisms are pooled, assumes no direction. Alternatively, the variance inflation factor (VIF) is included to further ensure no multi-collinearity issues are present.
33
Table 7: Results Ordinary Least Squares Regression.
P-values reported in parentheses.
*, ** Significant at 5%, 1% levels, respectively, based on one-tailed tests where the sign of the coefficient is predicted, and two-tailed otherwise.
The Sample consists of 1023 European IPO’s from January 1, 1990, through December 31, 2010.
In table 6, models 1 and 2, both show an adjusted R-squared of 0.566. This implies that 56.6% of the difference between offer and first-day closing price is explained by the variable of interest and the control variables. This is higher than the adjusted R-squared provided by the ordinary least squares regression by Kaneko & Pettway (2003), who compared auctioned vs bookbuilt IPO’s. They showed an adjusted R-squared of 0.2075.
When taking in account all variables, several conclusions can be drawn regarding the control variables. Offer price (OFFER) is significant in predicting initial returns (FDR).
However, the regression coefficients of 0 indicate that, ceterus paribus, neither an increase nor a
Variable Predicted Sign Model 1 VIF Model 2 VIF
Intercept 0.489** 0.508** (0.008) (0.006) AUCTION -/- 0.225 7.221 (0.000) BOOKBUILD + 0.197** 7.736 (0.001) MBPM 0.209** 2.115 (0.000) AGE 0.000 1.106 0 1.095 (0.273) (0.320) OFFER -0.004** 1.566 -0.004** 1.566 (0) (0) PROCEEDS 0 1.305 0 1.303 (0.856) (0.826) AUDITOR -0.061 1.121 -0.062 1.117 (0.17) (0.14) MARKET 0.001* 1.408 0* 1.401 (0.033) (0.039)
YEAR Yes Yes
COUNTRY Yes Yes
N 1023 1023
