IPOs in pharmaceutical industry : determining factors which influence pharmaceutical companies' underpricing and post-IPO returns

(1)

IPOs in Pharmaceutical Industry

Determining factors which influence pharmaceutical companies'

underpricing and post-IPO returns

UNIVERSITEIT VAN AMSTERDAM

AMSTERDAM BUSINESS SCHOOL

MSc Finance

Master Specialisation Corporate Finance

Author:

A. Talacheva

Student number:

11625090

Thesis supervisor:

Dr.

J.J.G. Lemmen

(2)

2

PREFACE AND ACKNOWLEDGEMENTS

I would like to take this opportunity to express my gratitude to the people involved in the process of writing my thesis. First of all, I am very thankful to my supervisor – Dr. J.J.G. Lemmen, who was not only a great advisor and a thoughtful mentor, but who gave me an opportunity to find my own path in the IPO research field. I am very grateful for his flexibility and encouragement of students’ creative ideas, work ethic and decision-making.

Secondly, I would like to thank Sandoz NL – the pharmaceutical company where I currently have an internship in the Finance Department. They gave me a wonderful opportunity not only to see how the finances of a pharmaceutical company work in real-life, but were also very enthusiastic in answering the questions that naturally arose. I would especially like to thank Adri Simamora, the CFO of Sandoz NL, who provided me with an overview of the large gap between the theoretical and practical approaches to finance in the pharmaceutical industry. I would also like to express my gratitude to Else Vissers, Contract Management and Data Claim Analyst, whose theoretical background in the Global Health and practical involvement in the company’s day-to-day operations gave me many insights about pharmaceutical industry operations.

STATEMENT OF ORIGINALITY

This document is written by Student Anna Talacheva who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

ABSTRACT

This thesis concentrates its attention on Initial Public Offerings, specifically on the phenomenon of underpricing in the pharmaceutical industry. As the performance of pharmaceutical companies is largely dependent on their intangible assets, we argue that measurements of innovation activities (namely, the number of patents the company has prior to an IPO and R&D expenses) are important determinants of underpricing, first month return and first year return after an IPO. Using a sample of 224 U.S. based pharmaceutical companies, we prove that both of these measurements are significant, with the number of patents being the more accurate estimation tool.

Keywords: IPO, underpricing, pharmaceutical, patents, R&D expenses JEL Classification: I11, G32

(4)

4

LIST OF TABLES

Table 1: Regulation levels in Pharmaceutical industry ... 18

Table 2: Description of the variables used ... 26

Table 3: Shapiro-Wilk W test for normal data for the ratios ‘Number of patents/Assets’ and ‘R&D expenses/Assets’ ... 31

Table 4: Shapiro-Wilk W test for normal data for the variables log(Number of patents/Assets) and log(R&D expenses/Assets)... 32

Table 5: Summary of data management steps ... 32

Table 6: Distribution of Initial Sample and Final Sample by Decade ... 34

Table 7: Number of IPOs and average underpricing in the pharmaceutical industry by year ... 34

Table 8: Descriptive statistics of the dependent and independent variables ... 35

Table 9: Summary statistics of dependent variables: underpricing, first month return and first year return after an IPO ... 37

Table 10: OLS regression models of the First-Day IPO Returns ... 38

Table 11: Multicollinearity check (VIF test) for regression with patents variable... 39

Table 12: Multicollinearity check (VIF test) for regression with R&D expenses variable ... 40

Table 13: OLS regression models of the First-Day IPO Returns, comparison between unrestricted (“long”) and restricted (“short”) versions... 42

Table 14: Multicollinearity check (VIF test) for restricted “short” regression with patents variable ... 43

Table 15: OLS regression models of the First-Month IPO Returns ... 44

Table 16: OLS regression models of the First Year IPO Returns ... 45

Table 17: Comparison of OLS regression models for Underpricing, First Month Return and First Year Return ... 47

(6)

6

LIST OF FIGURES

Figure 1: First-day return fluctuations, average per month, % ... 10 Figure 2: First-day return fluctuations, average per decade, % ... 11 Figure 3: Plot for the comparison of the distribution of the 'Number of Patents/Assets' ratio with the reference line for the normal distribution ... 30 Figure 4: Plot for the comparison of the distribution of the 'R&D expenses/Assets' ratio with the reference line for the normal distribution ... 30 Figure 5: Plot for the comparison of the distribution of the variable log (Number of patents/Assets) with the reference line for normal distribution ... 31 Figure 6: Plot for the comparison of the distribution of the variable log (R&D expenses/Assets) with the reference line for normal distribution ... 32

(7)

7

CHAPTER 1 Introduction

1.1. Research question and motivation

For many years, the potential benefits and costs of a company conducting an Initial Public Offering (IPO hereafter) challenged the minds of financiers. The well-known phenomenon of underpricing seemed to trap almost every company that pursued an IPO. On average, first-day returns reached 16.9% in U.S. (Ibbotson, Sindelar and Ritter, 1988), leaving a significant amount of possible raised capital “on the table”. The decision to go public seems to be even more puzzling, as Loughran and Ritter (1995) find that issuing firms consistently underperform in comparison to control group firms with the same level of market capitalization that remained private. This pattern is consistent for at least five years following an IPO, resulting in investors losing a cumulative 2% in stock market value to their best alternative option. A similar result is obtained during a Secondary Equity Offer (SEO hereafter) (Loughran and Ritter, 1995). Despite these facts, the number of companiesseeking to conduct an IPO has not decreased over time or fallen to zero. On the contrary, recent research (Ernst and Young, 2017, p.11) states the following insight considering America's IPO markets: “Strong end to 2017 points to more IPOs in 2018”.

This research study focuses its attention on IPOs within the pharmaceutical industry. The main reason for this decision is the difference between the pharmaceutical industry and traditional industries. High information asymmetry surrounding IPOs in the pharmaceutical industry results in the importance of non-classical finance determinants for investment decisions and a large role for unconventional behavioural bias in a pharmaceutical company’s valuation. Information asymmetry between firms and investors is always important, but for certain industries it is crucial. As with any other investment possibility, the decision of an investor to buy shares of a company conducting an IPO is, above all, an estimation of costs and benefits. With Research and Development and project investments in the pharmaceutical industry having an average life cycle of 13.9 years (Efrata, 2008), it is close to impossible for investors to evaluate companies' future cash flows, create a discounted cash flow model and calculate the true stock price today. Another obstacle is the industry’s dependence on intangible assets. One example is knowledgeable workers, which tend to leave a company in times of private-to-public transition due to future uncertainty (Bernstein, 2014) leading to a decrease in the quality of a company’s innovation which is unknown to investors. Another example is patent legislation, which not only varies from country to country, but also cannot be simply evaluated by the number of patents a firm is holding (Trajtenberg, Henderson and Jaffe, 1997; Bernstein, 2014). This kind of information is complicated for investors to process, resulting in them using shortcuts to evaluate the value of a firm instead (Cohen and Frazzini, 2008).

(8)

8 Considering all of the above, the pharmaceutical industry is very different from traditional ones and therefore is interesting to research. It is also not clear from previous research whether the pharmaceutical industry should be considered speculative and whether difficulty in determining future cash flows results in a higher level of underpricing. It might have different determinants responsible for post-IPO pricing and stock price volatility than traditional industries. Moreover, the same determinants might have different impacts than for comparable companies from traditional industries.

Therefore, the research question of this work is:

What are the size and determinants of short- and long-term performance of US Pharmaceutical IPOs?

To fully answer this question, several hypotheses were introduced:

Hypothesis 1: Companies with more patents at the time of Initial Public Offering have less underpricing;

In the pharmaceutical industry the number of patents highly correlates (one-to-one relationship) with the number of drugs a company is selling. An intuitive explanation for this hypothesis is that the number of patents is a proxy for the future potential growth of a company, so that with a larger number of patents, a pharmaceutical company’s valuation is more accurate for investors, resulting in less underpricing.

Based on traditional IPO theories:

Hypothesis 2: Larger companies have smaller initial returns at the first trading day; companies in the

pharmaceutical industry act in a similar way to companies in more traditional industries;

An intuitive explanation is that larger companies (firms with more sales) are considered to be more stable, therefore investing in its stock issuing is less risky and the company’s true stock price is considered to be evaluated more accurately, resulting in relatively smaller underpricing. If firms in the pharmaceutical industry differ significantly from traditional industries, investors’ sentiment being more important (more optimistic or hopeful) hypotheses 2 might be rejected. The rejection of this hypothesis will indicate that for valuation of a pharmaceutical company the traditional approach alone is not sufficient.

Hypothesis 3: For the estimation of the underpricing of a pharmaceutical company, the number of patents

provides more accurate results than research and development expenses.

We expect this hypothesis to be confirmed as in theory the pharmaceutical industry has a strong one patent - one medicine correlation. Otherwise, if research and development expenses are a more accurate estimator of underpricing, we can conclude that pharmaceutical industry is quite similar to more traditional industries.

(9)

9

Hypothesis 4:The implemented model more accurately describes short-term average returns than

long-term average returns.

An intuitive explanation is that any long-term investment is made with higher uncertainty. This results in determinants of yearly returns being less significant and having less explanatory power than those of monthly and first day returns.

1.2. Main results

This study obtains determinants specifically for pharmaceutical industry underpricing, first month and first year post-IPO performance. All hypotheses received confirmation after empirical analysis. We found a negative correlation between underpricing and innovation activity measurements: the number of patents, research and development expenses. The more innovation activity investors observe, the more quickly the true price is discovered. Correlation between first day return and sales was also negative, so that, on average, larger firms experience smaller underpricing. The number of patents was determined to be a more efficient proxy of innovation activity than research and development expenses, as the one patent – one drug correlation helps investors to estimate true price of a company’s stocks. Finally, we found that longer term average returns after an IPO are more difficult to predict as they are subject to higher level of uncertainty. The main results of this research are largely in line with the theoretical predictions based on the literature review.

1.2. Thesis Organisation

This thesis is organized as follows. In the next chapter we will discuss theoretical and practical aspects of going public by a pharmaceutical company. As there is a significant gap in the literature covering the topic of a pharmaceutical company conducting an IPO, chapter 2 can be conceptually divided into two major parts: the phenomenon of underpricing and specific characteristics of the pharmaceutical industry. Chapter 3 presents the methodology of our research including the models we are using and is followed by a thorough description of the data collection process in Chapter 4. Chapter 5 presents the main results and critical analysis. This chapter also discusses the confirmation or rejection of hypotheses and the intuitive logic behind the results. In the final chapter, the results are reviewed with a particular interest towards answering the research question, correlating results with the previous literature, and examining the place of the results in the field of study and possibilities for future research.

(10)

10

CHAPTER 2 Literature review

2.1. Definition of underpricing

Underpricing is a financial phenomenon when during an Initial Public Offering – the first time a company enters public capital markets – the price of the stock rises substantially during the first day of trading (Ritter and Welch, 2002). This way the selling price (also called offer price) turns out to be lower than the true value of the stock (closing price).

Underpricing =(First day closing price−Offer price)

Offer price (1)

Underpricing has been thoroughly studied over the years. Starting with early works by Logue (1973) and Ibbotson (1975) it has been researched extensively, but it is still not explained fully. The underpricing discount remains on average constant around 17% in the U.S. throughout the years, albeit with significant fluctuations over the business cycle (Ljungqvist, 2007).

Figure 1: First-day return fluctuations, average per month, %

This figure illustrates overall volatility of the underpricing over time. The absolute peak was reached in September, 1998 with average underpricing equal to 163.2%. This somewhat outlying result is due to “IPO bubble period” from September 1998 to August 2000 (Lowry, Officer and Schwert, 2010). The figure is built by the author based on the data from J. Ritter’s website (https://site.warrington.ufl.edu/ritter/ipo-data/) -50,00 0,00 50,00 100,00 150,00 200,00 ja n-1 96 0 de c-1 96 1 no v-1 96 3 okt -1 96 5 se p-19 67 au g-1 96 9 ju l-1 97 1 ju n-19 73 m ei-19 75 ap r-1 97 7 m rt -1 97 9 fe b-1 98 1 ja n-1 98 3 de c-1 98 4 no v-1 98 6 okt -1 98 8 se p-19 90 au g-1 99 2 ju l-1 99 4 ju n-19 96 m ei-19 98 ap r-2 00 0 m rt -2 00 2 fe b-2 00 4 ja n-2 00 6 de c-2 00 7 no v-2 00 9 okt -2 01 1 se p-20 13 au g-20 15 ju l-2 01 7

(11)

11 Figure 2: First-day return fluctuations, average per decade, %

This figure illustrates average underpricing per decade. The lowest point was achieved in 1970s with average first-day returns being around only 11% and the peak was achieved in 1990 with 23%. This figure is built by author based on the data from J. Ritter’s website (https://site.warrington.ufl.edu/ritter/ipo-data/)

A significant and constant level of underpricing results in a large amount of money being “left on the table”, a potential sum of money a company could have reached but did not due to an incorrect evaluation of offer price (Loughan and Ritter, 2002).

Money left on the table = (First day closing price − Offer price) ∗ Number of shares (2)

For example, the largest IPO ever conducted, by Alibaba Group Holding in September 2014, had an offer price of $68 per share (top end of the expected range). However, the closing price of Alibaba reached $93.89 per share the first day of trading, with 38% growth from the offer price. This means that Alibaba Group Holding left $8.3 billion on the table, the money this company could have received with an initially correct stock offer price. In comparison, in the pharmaceutical industry money left on the table during IPOs for the whole year of 2012 is equal to around $1.2 million. An average pharmaceutical firm going private in 2012 left around $130,000 on the table (Williams, 2015). Loughran and Ritter (2002) discovered that the combined underpricing of all firms from 1990 till 1998 is equal to 3 years of combined profits. However, when calculating the amount of the ‘money left on the table’, the formula binary assumes all shares are sold either at offer price or closing price.This is probably an unrealistic scenario in real-life. 19% 11% 16% 23% 17% 15% 0,00 0,05 0,10 0,15 0,20 0,25 1960 1970 1980 1990 2000 2010

(12)

12 2.2. Determinants of underpricing

Many economists believe that underpricing happens due to several reasons simultaneously. Ljungqvist (2004) places underpricing theories into four main groups: asymmetric information, institutional factors, ownership/control reasons and behavioural approaches. All of them are discussed more thoroughly below.

1. Explanations due to asymmetric information: beneath the decision to go public lies information asymmetry. There are three major parties responsible for conducting an IPO: the issuing firm, the underwriter and new investors. If information asymmetry is present, one of the parties knows more than the others. This results in an increase in trading volume, which drives the first day closing price further away from offer price, resulting in a higher level of underpricing.

One of the most well-known asymmetric information models considering underpricing is Rock’s (1986) winner’s curse model. When the whole stock market is growing, investors are in general optimistic. Therefore, conducting an IPO in a growing market will initially give an advantage, sufficiently large that going public during a crisis is not justified. From this point of view, underpricing can be explained as a “lemon market” (Akerlof, 1970). If IPOs are truly happening based on the market conditions, some of the offerings are not a profitable deal for investors, however, due to information asymmetry, investors are not aware of that. Assuming this theory is true, the cumulative value of public offerings is being consistently overvalued. Buying shares from a company that is going public seems to be a curse now – any new investor overvalued the company and therefore paid more than it is worth. In the research study conducted by Brau and Fawcett (2006) it was found that 70% of Chief Finance Officers (CFOs hereafter) strongly agree that industry condition highly influenced the decision to go public.

Underpricing can be used by the management team as a positive signal to investors. When insider information about present value of risk or future cash flows is present, large first day return can become a hint to the future investors of the higher returns a firm can bring (Grinblatt and Hwang, 1989). As costly of a signal as underpricing is, it does create a positive experience (‘good after-taste’) for investors involved (Ibbotson, 1975), which increases the possibility of future investments in this firm. Research by Boehmer and Fishe (2001) supports the hypothesis that underpricing increases after-IPO trading volume of the company’s stock.

2. Institutional explanations: There are three underpricing theories about the impact of institutions. The first to be introduced was the theory of avoiding lawsuits or, as it is more commonly referred, the theory of litigation risk (Alexander, 1993). Almost 6% of companies that went public between 1988 and 1995 in the U.S. were sued for violations of the IPO process (Lowry and Shu, 2002). Research by Alexander determined that the probability for a firm to be sued is

(13)

13 directly dependent on the possible size of the settlement reward. Smaller settlement rewards reduce the probability of a company receiving a lawsuit as the process is very time- and resource-consuming, so that it is only worth starting legal procedures when there is a chance to gain large sums of money. Therefore, it is plausible that firms intentionally allow underpricing of their stocks so as to avoid the possibility of future lawsuits. The compensation (direct costs) might reach on average 20% of capital raised (Lowry and Shu, 2001). More underpricing reduces IPO income but also reduces settlement size in case of lawsuits. Indirect costs are reputation costs, lawyer (legal) fees, and opportunity costs in terms of the time spent by management team and other employees on the lawsuit regulation.

Secondly, the underwriting bank may be responsible for underpricing. Part of an underwriter’s responsibilities is so-called price support or price stabilization. The role of the bank is to decrease the probability of a price drop in the days or weeks following an IPO. To achieve this, the underwriter might be willing to initially underprice the stocks so that true price discovery will result in positive returns for a company. In the study of Benveniste, Busaba, and Wilhelm (1996) price stabilization is considered a bonding mechanism for the underwriter bank and investors, which ensures both parties are satisfied with the IPO and might be willing to work together again in the future. Underwriters prefer to build good relations with their investors, as they may work with them again. The relationship with the issuing company is less important, as each company can conduct an IPO only once. Therefore, underwriters try to please investors at the expense of the issuing firm. The consequences of investors’ disappointment were thoroughly studied by Krigman and Jeffus (2015) using the example of Facebook, whose first-day closing price remained almost the same as its offer price. The Facebook IPO, which occurred during a period of no underpricing was followed by a sharp increase in average IPO first-day return. Investors were compensated for the absence of large first-day returns on the Facebook IPO with large underpricing after Facebook IPO; average first-day returns next month spiked from 11% prior to Facebook IPO to 20% afterwards. Krigman and Jeffus argue that this increase is a result of underwriters “compensation” to investors for the Facebook IPO, which had no underpricing.

Finally, there might be certain tax advantages which result in consistent underpricing (Rydqvist, 1997). The study conducted by Ljungqvist and Wilhelm (1993) discovered that companies are selling a smaller part of the firms during an IPO than they used to, which proves that firms are avoiding dilution of ownership if they can. To avoid attenuation of ownership and control, a firm can compensate their workers with equity and stocks. In this scenario, their compensation will be based on the offer price. When insiders (employees who own their company’s equity) execute their options during an IPO they might not qualify for paying the tax at the spread

(14)

14 between offer and closing prices but at an alternative minimum tax rate (Taranto, 2003) and receive large tax savings. Taranto discovers in his research that the tax benefit (particularly a company’s tax regulations) was highly correlated with the level of underpricing in companies, where the management team is compensated with stock options. The larger the tax savings, the larger the underpricing.

3. Ownership and control reasons: going public is one step further in the separation of ownership and control roles inside the firm. Underpricing can help the managerial team to allocate shares in such a way that there will not be one large investor – block holder – so that control rights stay in managers’ hands (Brennan and Franks, 1997). This pattern was discovered in UK-based IPOs. Sun and Tong (2011) prove that only 0.72% of eligible companies actually proceed with IPO process. One of the main reasons is the reluctance of the management team to transfer their ownership and control rights. Moreover, the same reasoning is provided by CFOs themselves in the study by Brau and Fawcett (2006). The authors conducted a private survey with 336 CFOs and found out that the main objective for the company was to stay private, and indeed, the intention of the CFOs was to keep control of decision-making and the ownership structure.

4. Behavioural reasoning unites two approaches, one assumes investors are subject to behavioural bias and the other assumes that the issuing company can be irrational.

Irrationality or sentiment (Ljungqvist, Nanda and Singh, 2002) of investors can come in two channels: overoptimism (investors believe the deal to be more profitable than its expected value is) and overconfidence (investors believe the risk to be smaller than it actually is). Investors tend to base their bids on other investors’ (earlier investors’) bids, resulting in ‘informational cascades’ which disregard individual perception of information in favour of earlier decisions of other investors, herding behaviour (Welch, 1992). This way underpricing can be a positive cascade with earlier investors getting compensated for their involvement and setting up the cascade of the investment in the company. In the study by Baker and Wurgler (2000) positive correlation was found between the amount of IPOs and investors’ optimism. The optimism was measured as higher abnormal returns. Moreover, certain information in Pharmaceutical industry is very difficult for non-specialists to comprehend, which may also result in investors being irrational. As was found in several studies (Cohen and Frazzini, 2008; Huberman and Regev, 2001), investors tend to disregard subtle or difficult to process information. And since all information considering product launches or clinical testing of new medicine can be very difficult to comprehend for non-specialists, investors might simply ignore it. As Aboody and Lev (2000) state, the newest updates about the process of clinical testing of new drugs do not

(15)

15 update investors' valuation as the information is too hard to process. This inattention to crucial information results in less accurate pricing.

Issuing companies can also be irrational, sometimes overestimating their valuation. For example, AkzoNobel refused several profitable acquisition offers from PPG despite the desires of its shareholder and suffered negative share price pressure afterwards1_{. The company might}

also be irrational regarding the timing of an IPO. In particular, going public in unfavourable macroeconomic conditions will result in more uncertainty among investors and, therefore, larger underpricing (Lucas and McDonald, 1990).

Combining these four areas of underpricing determinants, Ljungqvist (2004) states the most popular ‘traditional industry’ proxies from previous studies to be:

 Age of the firm: on one hand, older firms are considered to be more stable and have less information asymmetry, therefore less underpricing is present (Ritter, 1984). On the other hand, an IPO can be considered to be part of the life cycle of a company, so that investors perceive the company as more mature with less information asymmetry on the market (Megginson and Weiss, 1991, and others);

 Size of the firm, which is usually measured as the function of sales (Ritter, 1984). The more sales the more stable the firm seems to be to investors and the less it is dependent on intangible assets. This should result in smaller underpricing;

With IPOs happening in clusters of firms (Ibbotson and Jaffe, 1975) it seems that firms going public take advantage of a “window of opportunity” on the market. In the model, implemented by Rajan and Servaes (2002) all the investors are subdivided into three groups based on their trading patterns:

1. rational speculators, investors who avoid taking any risks (risk averse) and simply maximize their final wealth;

2. passive investors, whose investment decisions are determined by the expected true value of the shares and the current offer price. These investors purchase shares only when they believe a stock's intrinsic value to be higher than the current price on the market;

3. trend chasers, those investors who believe the price will move the same way (with a proportion factor) as it did in the past. This group of investors bases its decision to buy stocks not on the current stock price but on past market performance.

The first two groups represent rational investors as these investors base their decision on expected profits (true value of a stock) and costs (current offer price), while the third is irrational as its trading decision

1_{Reuters (May, 2017) “Akzo Nobel rejects third takeover proposal from PPG”}

(16)

16 is based not on the current situation but on past performance. Each company conducting an IPO is meant to behave differently according to its own true value, pricing and IPO strategy. We would like to concentrate attention on trend chasers, who assume the direction of pricing movements remains the same throughout time. As investors assume the market behaves as in the past, when they observe that the average stock return is bigger this month than the previous month, they expect the market to continue to grow and therefore invest more. The increase in demand due to IPO market conditions might increase underpricing, as excessive trading can spike the closing price.

The volume of IPOs tends to be very volatile, with fewer companies going public in times of financial distress (Lucas and McDonald, 1990). A financial crisis leads to an increase in the level of overall market uncertainty, which increases underpricing. All sources of financing, including issuing stocks, becomes more expensive during a crisis. So, with all else held constant, going public in times of financial turbulence will raise less capital. King and Banderet (2014) find that the underpricing of U.S. firms conducting an IPO in times of financial crisis is 26%, which is 4% higher than average (22%). With the pharmaceutical industry being highly dependent on subjective and expert valuations, an increase in the level of overall uncertainty can increase underpricing or decrease the demand for the stocks resulting in fewer successful IPOs. The overall volatility of the IPO market and the pattern of firms conducting IPOs in clusters reminds us how vital timing can be to an IPO. To track the periods of high IPO activity many researchers (Helwege and Liang, 1995; Lowry and Schwert, 2001; Brau, Francis and Kohers, 2000) use a special measure to determine the “hotness” of the IPO market. The market is considered to be “hot” when either the number of IPOs during a certain period of time is higher than the overall average (“hot” issue by number of IPOs) or when the underpricing is higher than average (“hot” issue by underpricing). “Hot” IPO periods are naturally characterized by higher first-day returns.

Based on the investors’ attitude and sentiment and IPO timing three more determinants are added:  Market condition, which indicates overall optimism when the average first day returns of an

IPO are consistently increasing (Baker and Wurgler, 2000);

 Crisis, which indicates those IPOs that were conducted in the times of financial distress;  “Hot” issue period, which indicates whether IPO is happening in times of high IPO market

activity.

There are two essential characteristics of economic agents involved in the IPO process, which influence the level of underpricing. The first is, of course, underwriter reputation. Having a prestigious underwriter (an underwriter with a high ranking) is associated with lower levels of information asymmetry. These underwriters are usually big investment banks, therefore they tend to have more IPO market expertise and fewer time- and resource constraints. This leads to more accurate estimation of stocks' prices and, consequently, faster determination of true price and smaller underpricing (Beatty and Ritter, 1986;

(17)

17 Carter and Manaster, 1990). Moreover, Carter, Dark and Singh (1998) find that the level of underperformance (in comparison to private firms) in the long-term period (3 years) is smaller for IPOs with a top-ranking underwriter. On the other hand, underwriters might work on behalf of investors’ best interests, which will result in higher first-day returns. Higher underpricing creates a positive signal and keeps investors working with the same investment bank.

The second characteristic of IPO economic agents is whether the IPO was venture capital-backed (VC-backed hereafter) or not. VC-(VC-backed companies prior to an IPO are funded mostly via private funds. As backed companies have more access to market expertise and more funds to complete an IPO, VC-backed firms tend to have faster price discovery and smaller underpricing. However, the literature on the empirical evidence is not as straight-forward. Some studies find no significant difference between VC-backed and non-VC-backed firms (Belghitar and Dixon, 2012) and some argue that venture capital involvement directly impacts only the choice of a high-ranking underwriter and not underpricing itself (Liu and Ritter, 2011). Taking into consideration the two characteristics of economic agents involved in the IPO process, two more determinants of underpricing are introduced:

 Underwriter’s reputation – underwriter ranking, which indicates how prestigious and reliable underwriter seems to be to the market and investors; the higher the underwriter’s ranking, the more expertise he brings to the IPO process, which reduces underpricing;

 VC-backed – variable that indicates involvement of venture capital funding prior to an IPO. Venture Capital provides extra resources (especially, market expertise) which results in faster price discovery and a smaller level of underpricing.

2.3. Characteristics of the Pharmaceutical industry

Research of the pharmaceutical industry usually unites all medicine related industries. Following Higgins and Rodriguez (2005), this research study uses SIC codes 2833-2836 to cover all industry, that includes, for example, pharmaceutical preparations (SIC code: 2834), medicine-patents (SIC code: 283406), manufacturing (SIC code: 283401), production and sale of different medications (SIC code: 446110) as well as medical devices (SIC code: 423450).

The pharmaceutical industry is truly one of the most interesting gambles from an investor’s perspective. Information asymmetry between scientists (medicine developers), the management team and potential investors is large. Needless to say, the pharmaceutical industry is also one of the most researched industries (Comanor and Scherer, 2000). Pharma’s objective is not only to generate profits but to improve people's lives, which cannot be accomplished easily or quickly. For example, one of the growing approaches is biosimilars2_{- technological medication, which is created on the basis of}

(18)

18 outstanding patents. Once the patent stops being valid only for one company on the market, many other companies get involved into creating similar drugs based on the now open patent,allowing drugs to be less expensive and more available to a wider audience.

One of the most interesting characteristics of the pharmaceutical industry is that society can suffer both from it and in the absence of it. People are affected both by lack of medicine needed and badly prescribed, outdated or poor quality medicines (Management sciences for health, 2012, ch.6). To avoid negative effects, the pharmaceutical industry is, on one side, highly regulated by the government and, on the other side, given subsidies, sponsorship and protection by it. The pharmaceutical industry is often argued to have monopoly power and is characterized by strict legislation, high dependency on patents – it is one of the leading industries by the number of patents issued per company, and large investments in innovation, research and development (Mazzucato and Tancioni, 2012). Every feature of this industry will be discussed thoroughly below.

Legislation in the pharmaceutical industry in the U.S.

Legislation in the pharmaceutical industry consists of three major categories: pharmaceutical laws, regulations and guidelines. Laws are the ground level of regulations, they are fundamental and very difficult to change. Its priority status helps to regulate relationships in Pharma industry. Passing a new law is a very time-consuming process because the final approval of a certain law needs to be granted not by one advisor but by the country’s legislative branch. Therefore, existing laws might become outdated and insufficient for solving the newest problems. Regulations, on the other hand, can be passed much faster and more easily than laws, as they usually only need to be approved by one ministry or even by a group of experts. If a regulation is approved, it gains the power of law and the responsibility of involved parties to respect and follow it, otherwise they can get punished. Thus, regulations seem to be a better version of laws, easier to implement and sharing same power. However, regulations are much easier to alter, making them more vulnerable to court hearings. Guidelines, unlike laws and regulations, are not obligatory to follow. Guidelines offer informal information on what should be done and how in a certain industry. Guidelines are updated or modified quite often, depending on how to present information in the easiest and most understandable way. Guidelines are created not to dictate the rules, but to avoid misunderstandings or favourable misinterpretations. Its informal character helps regulatory information to be more accessible to a wider public (Managerial sciences for health, 2012).

Table 1: Regulation levels in Pharmaceutical industry

Regulation level Mandatory Implementation process

Laws Yes Very difficult

(19)

19

Guidelines No Easy

This descriptive table is built by author based on the data from Managerial sciences for health (2012)

This research focuses its attention on patent laws, as it is the most important regulation in the Pharmaceutical industry, which has an impact on each firm in particular and not on the industry as a whole. A patent is a special property right, which innovators receive for new discoveries. This right protects their innovation from anyone who is willing to replicate and use it for personal advantage for 20 years, which can be prolonged if significant time has passed between the discovery and market-placing of a product. Patents can be used not only as an indicator of current research and development activities of a firm but, more importantly, of future progress and growth rate, future cash flows. The number of patents is used as a proxy for the level of innovation of a company (Bernstein, 2014) or for more technical tasks – such as determining value of potential target in mergers and acquisitions (Breitzman and Thomas, 2002). While in electronic or high-tech industries patents can be shared between competitors (so-called cross-licensing or pooling) since each product requires many patents at the same time as it consists of many parts, in the pharmaceutical industry one patent represents one particular medicine the company is selling (Lehman, 2003). Therefore, every patent is protecting one medicine and with that - the competitive advantage of a firm. This legal protection is vital since the production process is relatively easy to replicate as it is not as expensive or time-consuming as research, development and clinical testing prior to the discovery (Grabowski, 2002).

Research and Development

The role of innovation in the pharmaceutical industry is crucial. The pharmaceutical industry is one of the most forward-thinking industries: each drug takes almost 14 years of development until it begins to generate profits (Efrata, 2008). According to the U.S. Department of Health and Human Services, Food and Drugs Administration3_{(FDA hereafter), a medicine discovery goes through 4 major steps:}

1. Discovery and development: a research department tests many different compounds (5,000-10,000 compounds) of existing medicine and decides which ones seem promising. Then the process of gathering together specific information like side effects of a drug, the best way to take it in, how the new medicine interacts with already existing drugs, testing if the new drug has different impact on different socio-demographic groups (different by gender, age or ethnicity) starts. After all necessary information is obtained, the effectiveness of a drug is checked. When the new medicine works flawlessly but is less effective or more expensive to produce than already existing medicine, there is no point in implementing it.

3_{U.S. Department of Health and Human Services, Food and Drugs Administration}

(20)

20 2. Preclinical Research: all new drugs go through several tests including laboratory and animal testing to ensure medicine safety. At this step, only 250 out of 5,000-10,000 compounds from the development phase proceed further. The first two steps combined take approximately 6 years (Efrata, 2008).

3. Clinical Research: this step implies tests of new medicine on the group of people. The trial is designed in such a way that it answers specific questions about the product’s safety and effectiveness. For every trial a special plan called protocol is developed. All the procedures and rules concerning the conduct of clinical research are put in the protocol and are mandatory to follow. At the third stage patients start to give feedback on the drug, potential side effects are discovered, reported and summarized. If the drug is considered to be working effectively and not doing harm, it proceeds to the final stage. After the third step on average only five compounds are left. Testing takes around 7.5 years.

4. FDA review: if all the previous tests proved new medicine is both safe and effective, the company can finally apply for market access for the drug. After all the documents are submitted the Food and Drugs Administration thoroughly examines proofs of efficiency and decides to approve or reject it. If a company has an objective to sell medicine in other countries apart from the U.S. it has to apply to Health Department of each particular country separately. This process takes around 1.5 years and usually results in only one compound left, which becomes the new medicine and can be launched on the market.

This way, creating a new medicine requires around 5,000-10,000 initial ideas and around 15 years of hard work and massive investments. Moreover, after the new drug is fully reviewed and approved by FDA, the Food and Drugs Administration continues monitoring all drugs currently on the market. The major concern is that with larger sample of people using the medicine, the drug might turn out to be unsafe and harmful. The FDA also examines conditions in which medicine is being made and checks for signs of any illegal activity.

With medicine development having such a long cycle, the pharmaceutical industry uses profits from already successful products on the market to invest into research and development of new drugs. According to a study conducted by the National Bureau of Economic Research (NBER hereafter), on average 17% of total current sales is invested in Research and Development (Danzon, 2006). During the master thesis internship we had a chance to ask directly the CFO of Sandoz NL, Adri Simamora, on his view about large R&D expenses and its added value. According to him, there are two main sources of competitive advantage for a pharmaceutical company: patents and an exclusive commercial right from a manufacturer. With commercial agreements being easy to change depending on market conditions, patents represent a more expensive but more sustainable competitive advantage.

(21)

21 The valuation of a Pharmaceutical company largely depends not only on assets, both tangible and intangible, but also on potential discoveries and the probability of innovation happening in the future. As already existing drugs are used mostly as a source of money for future discoveries, traditional valuation methods based on cash flows cannot be a representative valuation tool. Moreover, a breakthrough in the future by itself is not an endgame goal, a discovery needs to happen before it will be found and legally protected by competitors. A new drug has to be market-adjusted and easy to scale for a large market - large numbers of patients. If the new medicine cannot be scaled, it is basically closed for wide audiences, impossible to distribute and ultimately unprofitable.

Since it is impossible to measure the future outcome of today’s innovation or the possibility of future discoveries, for measuring innovation researchers (Trajtenberg, Henderson and Jaffe, 1997; Bernstein, 2014) usually take into account a combination of these five components:

1. R&D expenses;

2. Scaled patents counts – the sum of all patents the company is holding;

3. Scaled patent citations - measurement of breakthrough innovation, if the patent is granted in the sector where there are a lot of patents already, it receives less weight;

4. Scaled originality - how many citations has the patent received, was it cited by a broader array of technology classes, originality is scaled by the average calculated in the same year and the same technology class;

5. Scaled generality - whether the patent is cited by a more “technologically varied array of patents”, meaning that the patent can be used as a solid base for other patents, generality as well as originality is scaled by the average of the same year and the same technology class.

The choice of which proxy or which combination of the above to use as a measure of innovation usually depends on the industry being researched, as different industries have different patent legislation. In the pharmaceutical industry each patent usually stands for one product, therefore checking the number of patents will provide an estimation of the number of medicines a firm will have on the market (Lehman, 2003; Grabowski, 2002).

2.4. The determinants of underpricing of pharmaceutical IPOs

There is a significant gap in the literature regarding specifically Pharmaceutical companies going public. Moreover, it is unclear if we should expect a higher or lower average level of underpricing in the pharmaceutical industry in comparison to the market as a whole. Several general studies are rather relevant for the topic of the pharmaceutical IPOs even though they do not touch upon the industry in particular. One is the study by Bernstein (2014), who evaluates how going public has an impact on the innovation activity of a company. Bernstein discovers that an IPO does not have an impact on the volume

(22)

22 of company’s innovation, or what he refers at as “innovation rate”. However, the quality of the innovation declines following going public. This could be happening due to the departure of knowledgeable workers (e.g. scientists or doctors), who prefer not to be involved with a company in a period of significant uncertainty. This results in additional changes to the underlying innovation strategy. Information asymmetry plays large role in this situation: if outsiders estimate company’s innovation activity by volume of activity but not its quality, their estimation of company’s innovation activity will not change. Considering the timeframe, a large time lag occurs before outsiders find out the quality of innovation dropped even though number of patents might have increased.

Despite the need to value innovation activity’s quality, investors (and researchers) historically concentrate their attention on investments in innovation as they are easier to measure. A common method is checking its research and development expenses (Kerr and Nanda, 2014). However, an innovation measure based on R&D expenses only is not considered sufficiently accurate in industries where future cash flows are highly impacted by new technologies. In other words, large investments in innovation do not necessarily transform into break-through discoveries, and discoveries can be challenging to monetize. For innovation-intensive industries researchers prefer to use patents measures (Pandit, Wasley and Zach, 2011; Bernstein, 2015), which might include simple counts of patents or patents citations depending on industry features. The pharmaceutical industry is characterized by a one patent - one drug relation, therefore the number of patents should sufficiently represent the success rate of innovation activity (Lehman, 2003). Using patents results in estimating the quality of innovation, its potential and therefore a more accurate valuation of a pharmaceutical firm. Both R&D expenses and the number of patents need to be normalized by a company’s size variable, so it can be compared throughout a sample of different-sized firms. In the work of Efrata (2008) company’s assets are used for this purpose.

After filing for an IPO, around 20% of American firms decide not to proceed and withdraw the application (Dunbar and Foerster, 2008). This happens more often when the valuation of a company is subjective (e.g. the company of interest has a lot of intangible assets or the company is in an actively growing phase when it is hard to predict future cash flows), so that the valuation of the company’s market value is an approximation. In situations of uncertainty, investors’ perception of negative news becomes much stronger than that of positive (Busaba et al., 2001). Even neutral news can build a negative bias around investment in a firm, whose valuation is subjective. If a company is threatened by negative bias and therefore downward pressure on the offer price, it can decide not to conduct a public offering at all. In the pharmaceutical industry withdrawal mainly happens due to rejection of a new drug by the Food and Drugs Administration, environmental scandals considering medicine manufacturing or a slowdown in the stock market. From investors’ point of view, a history of withdrawn applications increases the level of uncertainty whether offering will be conducted at all and if it is going to be

(23)

23 successful as company had troubles with proceeding with an IPO beforehand.This uncertainty is often reflected in a decrease in IPO pricing for the second IPO attempt (Dunbar, 1998), hence an increase in underpricing. However, there is a literature gap considering empirical evidence of the correlation between whether the company withdrew its IPO application and the level of underpricing it had once it did proceed with an IPO.

When a pharmaceutical company is preparing all the documentation for issuing shares, including prospectus, information about successful ongoing projects becomes public and widely known not only for investors, but also for competitors (Guo, Lev and Zhou, 2004). Unlike other industries, pharmaceutical companies publish a broad summary considering its investments as they represent the firm’s involvement in innovation activity. One of the main characteristics of these companies are their research and development expenses. However, including this kind of information in a prospectus can be a poor move by the management team, since competitors might take advantage of it; especially sensitive is information on the clinical tests of various drugs. Therefore, many companies include only rough aggregate numbers considering R&D expenses and the information provided is not particularly relevant for valuation. Similarly, firms avoid including information on the drugs that were rejected at the clinical testing stage so that competitors would not obtain and implement this data (Aboody and Lev, 2000). As this strategy increases information asymmetry, less information can be accessed by investors, resulting in less accurate price discovery and, therefore, larger underpricing (Sherman and Titman, 2002). The main objective of going public for a company might not be to raise capital, but to prepare for future acquisitions (Brau and Fawcett, 2006). In the pharmaceutical industry with its dependency on patents, companies can use intellectual property as leverage for bargaining power during an acquisition. It was also found that the more patents a company has, the less likely it is to go bankrupt (Saotome, Nakaya and Abe, 2016). When a company is conducting an IPO with a strategy of getting acquired soon after issuing stocks, it is particularly interested in sending a positive signal to investors, e.g. having large first day returns (underpricing). This creates an aftertaste for investors of the true pricing of a company being much higher than the offer price is, resulting in an increase in company’s expected growth rate and higher pricing for an acquisition (Schultz, 2000).

Combining all the above, important determinants specifically for the pharmaceutical industry are:  Research and Development expenses: as a signal for investors about company’s volume of

innovation and the probability of future break-through discoveries hence future cash flows;  Number of patents: as it shows how many drugs the company has discovered already, the

quality of investments in innovation, and the company's potential for future development;  A variable that represents if an IPO application was withdrawn previously: as this is a signal

for investors of an increased level of uncertainty around a company’s going public process that might result in larger underpricing.

(24)

24

CHAPTER 3 Method

3.1. Main Variables

This research study examines three different independent variables. First and, arguably, the most important among them is underpricing. Underpricing is calculated as the difference between first day closing price and offer price, divided by offer price. It is also considered to be the first-day initial return.

𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔 =

𝐹𝑖𝑟𝑠𝑡 𝑑𝑎𝑦 𝑐𝑙𝑜𝑠𝑖𝑛𝑔 𝑝𝑟𝑖𝑐𝑒−𝑂𝑓𝑓𝑒𝑟 𝑝𝑟𝑖𝑐𝑒

𝑂𝑓𝑓𝑒𝑟 𝑃𝑟𝑖𝑐𝑒

(1)

The second independent variable is the first month return of a stock after an IPO. It is calculated with this formula:

𝐹𝑖𝑟𝑠𝑡 𝑀𝑜𝑛𝑡ℎ 𝑅𝑒𝑡𝑢𝑟𝑛 =

𝐹𝑖𝑟𝑠𝑡 𝑚𝑜𝑛𝑡ℎ 𝑐𝑙𝑜𝑠𝑖𝑛𝑔 𝑝𝑟𝑖𝑐𝑒−𝑂𝑓𝑓𝑒𝑟 𝑝𝑟𝑖𝑐𝑒

𝑂𝑓𝑓𝑒𝑟 𝑃𝑟𝑖𝑐𝑒

(2)

The last independent variable is the first-year return of a stock after going public. It is calculated with this formula of first year return:

𝐹𝑖𝑟𝑠𝑡 𝑌𝑒𝑎𝑟 𝑅𝑒𝑡𝑢𝑟𝑛 =

𝐹𝑖𝑟𝑠𝑡 𝑦𝑒𝑎𝑟 𝑐𝑙𝑜𝑠𝑖𝑛𝑔 𝑝𝑟𝑖𝑐𝑒−𝑂𝑓𝑓𝑒𝑟 𝑝𝑟𝑖𝑐𝑒

𝑂𝑓𝑓𝑒𝑟 𝑃𝑟𝑖𝑐𝑒 (3)

We use five explanatory variables as determinant of returns after an IPO: research and development expenses, number of patents prior to an IPO, sales of a company, ranking of the underwriter and the age of a firm. R&D expenses and number of patents need to be normalized by assets of a company so that we can compare firms of different sizes. As relationship between these two determinants and independent variables can be non-linear, we also need to construct logarithms of variables (Efrata, 2008; Guo, Lev and Shi, 2005; Heeley, Matusik and Jain, 2007). The non-normal distribution of the innovation activity measures and the non-linear relation with the dependent variables are more thoroughly discussed in chapter 4.

𝐿𝑜𝑔(

𝑃𝑎𝑡𝑒𝑛𝑡𝑠 𝐴𝑠𝑠𝑒𝑡𝑠

) = 𝑙𝑛(

𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑝𝑎𝑡𝑒𝑛𝑡𝑠 𝑝𝑟𝑖𝑜𝑟 𝑡𝑜 𝐼𝑃𝑂 𝐴𝑠𝑠𝑒𝑡𝑠

)

(4)

𝐿𝑜𝑔 (

𝑅&𝐷 𝐴𝑠𝑠𝑒𝑡𝑠

) = 𝑙𝑛 (

𝑅&𝐷 𝑒𝑥𝑝𝑒𝑛𝑠𝑒𝑠 𝐴𝑠𝑠𝑒𝑡𝑠

)

(5)

Apart from explanatory variables, this research uses several control variables, namely:  Crisis: variable, that indicated that IPO was conducted during financial crisis;

(25)

25  Withdrawn before: variable, that indicates those firms that already listed its application to go

public but decided to withdraw it;

 Market condition: variable, that indicates IPOs conducted in time of higher first-day returns than in the past month. This variable is meant to control for investors’ sentiment and optimism;  VC-backed variable: variable, that indicates whether the IPO was conducted under venture

capital analysts' support and expertise;

 “Hot” issue: variable, which indicates periods of time when either underpricing was higher than the all-time average (“hot” issue by underpricing) or the number of completed IPOs was higher than the all-time average (“hot” issue by number of IPOs).

3.2. Models

To study the impact different determinants have on Underpricing in the Pharmaceutical industry, we constructed six ordinary least square (OLS) models. For each of the three independent variables we ran two regressions with the same set of determinants apart from number of patents or research and development expenses, which we used once per regression. This was done to evaluate and compare predictive power of volume measurement of innovation activity (R&D expenses) and quality measurement (number of patents).

This way, two regressions were:

𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔 = ∝ + 𝛽1∗ 𝑙𝑛(𝑃𝑎𝑡𝑒𝑛𝑡𝑠 /𝐴𝑠𝑠𝑒𝑡𝑠) + 𝛽2∗ 𝑆𝑎𝑙𝑒𝑠 + 𝛽3∗ 𝑙𝑛 𝐴𝑔𝑒 + 𝛽4∗ 𝐷1𝐶𝑟𝑖𝑠𝑖𝑠+ 𝛽5∗ 𝑈𝑛𝑑𝑒𝑟𝑤𝑟𝑖𝑡𝑒𝑟 𝑅𝑎𝑛𝑘𝑖𝑛𝑔 + 𝛽6∗ 𝐷3𝑊𝑖𝑡ℎ𝑑𝑟𝑎𝑤𝑛 𝑏𝑒𝑓𝑜𝑟𝑒+ 𝛽7∗ 𝐷4𝑀𝑎𝑟𝑘𝑒𝑡 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛+ 𝛽8∗ 𝐷5"Hot" 𝑖𝑠𝑠𝑢𝑒 𝑏𝑦 𝑢𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔 + 𝛽9∗ 𝐷6"Hot" 𝑖𝑠𝑠𝑢𝑒 𝑏𝑦 𝑛𝑢𝑚𝑏𝑒𝑟+ 𝛽10∗ 𝐷7𝑉𝐶−𝑏𝑎𝑐𝑘𝑒𝑑+ 𝜀𝑡 (6) 𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔 = ∝ + 𝛽1∗ 𝑙𝑛(𝑅&𝐷 𝑒𝑥𝑝𝑒𝑛𝑠𝑒𝑠 /𝐴𝑠𝑠𝑒𝑡𝑠) + 𝛽2∗ 𝑆𝑎𝑙𝑒𝑠 + 𝛽3∗ 𝑙𝑛 𝐴𝑔𝑒 + 𝛽4∗ 𝐷1𝐶𝑟𝑖𝑠𝑖𝑠+ 𝛽5∗ 𝑈𝑛𝑑𝑒𝑟𝑤𝑟𝑖𝑡𝑒𝑟 𝑅𝑎𝑛𝑘𝑖𝑛𝑔 + 𝛽6∗ 𝐷3𝑊𝑖𝑡ℎ𝑑𝑟𝑎𝑤𝑛 𝑏𝑒𝑓𝑜𝑟𝑒+ 𝛽7∗ 𝐷4𝑀𝑎𝑟𝑘𝑒𝑡 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛+ 𝛽8∗ 𝐷5"Hot" 𝑖𝑠𝑠𝑢𝑒 𝑏𝑦 𝑢𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔 + 𝛽9∗ 𝐷6"Hot" 𝑖𝑠𝑠𝑢𝑒 𝑏𝑦 𝑛𝑢𝑚𝑏𝑒𝑟+ 𝛽10∗ 𝐷7𝑉𝐶−𝑏𝑎𝑐𝑘𝑒𝑑+ 𝜀𝑡 (7)

As this research concentrates on the importance of determinants throughout time, two similar regressions were run with dependant variables being, respectively, return for the first month and the return for the first year:

𝐹𝑖𝑟𝑠𝑡 𝑚𝑜𝑛𝑡ℎ 𝑟𝑒𝑡𝑢𝑟𝑛 (𝐹𝑖𝑟𝑠𝑡 𝑦𝑒𝑎𝑟 𝑟𝑒𝑡𝑢𝑟𝑛) = ∝ + 𝛽1∗ 𝑙𝑛(𝑃𝑎𝑡𝑒𝑛𝑡𝑠 /𝐴𝑠𝑠𝑒𝑡𝑠) + 𝛽2∗ 𝑆𝑎𝑙𝑒𝑠 + 𝛽3∗ 𝑙𝑛 𝐴𝑔𝑒 + 𝛽4∗ 𝐷1𝐶𝑟𝑖𝑠𝑖𝑠+ 𝛽5∗ 𝑈𝑛𝑑𝑒𝑟𝑤𝑟𝑖𝑡𝑒𝑟 𝑅𝑎𝑛𝑘𝑖𝑛𝑔 + 𝛽6∗ 𝐷3𝑊𝑖𝑡ℎ𝑑𝑟𝑎𝑤𝑛 𝑏𝑒𝑓𝑜𝑟𝑒+ 𝛽7∗ 𝐷4𝑀𝑎𝑟𝑘𝑒𝑡 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛+ 𝛽8∗ 𝐷5"Hot" 𝑖𝑠𝑠𝑢𝑒 𝑏𝑦 𝑢𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔 + 𝛽9∗ 𝐷6"Hot" 𝑖𝑠𝑠𝑢𝑒 𝑏𝑦 𝑛𝑢𝑚𝑏𝑒𝑟+ 𝛽10∗ 𝐷7𝑉𝐶−𝑏𝑎𝑐𝑘𝑒𝑑+ 𝜀𝑡 (8) 𝐹𝑖𝑟𝑠𝑡 𝑚𝑜𝑛𝑡ℎ 𝑟𝑒𝑡𝑢𝑟𝑛 (𝐹𝑖𝑟𝑠𝑡 𝑦𝑒𝑎𝑟 𝑟𝑒𝑡𝑢𝑟𝑛) = ∝ + 𝛽1∗ 𝑙𝑛(𝑅&𝐷 𝑒𝑥𝑝𝑒𝑛𝑠𝑒𝑠 /𝐴𝑠𝑠𝑒𝑡𝑠𝛽2∗ 𝑆𝑎𝑙𝑒𝑠 + 𝛽3∗ 𝑙𝑛 𝐴𝑔𝑒 + 𝛽4∗ 𝐷1𝐶𝑟𝑖𝑠𝑖𝑠+ 𝛽5∗ 𝑈𝑛𝑑𝑒𝑟𝑤𝑟𝑖𝑡𝑒𝑟 𝑅𝑎𝑛𝑘𝑖𝑛𝑔 + 𝛽6∗

(26)

26 𝐷3𝑊𝑖𝑡ℎ𝑑𝑟𝑎𝑤𝑛 𝑏𝑒𝑓𝑜𝑟𝑒+ 𝛽7∗ 𝐷4𝑀𝑎𝑟𝑘𝑒𝑡 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛+ 𝛽8∗ 𝐷5"Hot" 𝑖𝑠𝑠𝑢𝑒 𝑏𝑦 𝑢𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔 + 𝛽9∗

𝐷6"Hot" 𝑖𝑠𝑠𝑢𝑒 𝑏𝑦 𝑛𝑢𝑚𝑏𝑒𝑟+ 𝛽10∗ 𝐷7𝑉𝐶−𝑏𝑎𝑐𝑘𝑒𝑑+ 𝜀𝑡 (9)

Below is a table with descriptions of all variables, their expected signs and the intuition behind the expected outcome.

Table 2: Description of the variables used

Variable Expected sign Explanation

Number of patents

-

The number of drugs a company is selling largely correlates with the number of patents company has. The number of patents is a proxy for potential future growth of the company. This results in more accurate valuation and, therefore, smaller underpricing.

R&D expenses

-

R&D expenses are a common proxy for the innovation activity of a company. The correct estimation of innovation activity results in a more accurate estimation of company’s future cash flows and growth and, therefore, smaller underpricing.

Sales of a company

-

Sales is used as a proxy for the size of a company. Larger companies provide more information on their performance and are viewed by investors as less risky since they have larger cash flows and a longer performance history. Their valuation tends to be more accurate, which results in smaller underpricing.

Age of a firm

-

Age of the firm signifies the level of maturity of the firm, how stable it is. From the Life Cycle theory only mature firms should conduct an IPO since going public is a natural step in company’s life. Hence, valuation of older firms should be more accurate and lead to smaller underpricing. Crisis

-

Companies tend to avoid going public in times of financial

crisis as the overall market uncertainty increases. Financial distress leads to smaller closing price and smaller first-day returns.

Underwriter reputation

-/+

Underwriters with high rankings are seen by the market as more prestigious agents with higher level of expertise and more resources, which results in more accurate estimation of companies' value and a smaller level of underpricing.

(27)

27 Nonetheless, if underwriters act in the best interest of investors and not the company conducting an IPO, they might intentionally increase first-day returns via decreasing offer price.

Withdrawn before

-

Dummy-variable equal to one if the company had applied to conduct an IPO but withdrew its application in the past. A history of withdrawn applications leads to larger uncertainty for investors which results in larger underpricing.

Market condition

+

Market condition is a dummy variable which indicates times when IPOs having larger first-day returns for the second month in a row. It is used as a proxy for investors’ optimism, which results in more speculative trading, higher demand for stocks and an increase in the closing price. “Hot” issue market

+

The IPO market is considered to be “hot” when either

underpricing is higher than the all-time average or number of IPOs is higher than the all-time average. The “hotter” the market, the more deals are happening and the more speculative trading seems to be, which results in bigger underpricing.

VC-backed

-

Companies, which conduct IPO with backing from venture capital have more access to market expertise and larger resources for correct true price evaluation, which leads to smaller underpricing.

To check how important the control variables are, we also ran a restricted (short) regression to compare the results with the very first model:

(28)

28

CHAPTER 4 Data

Collecting data for this research was a multi-step process, which required using several databases. The main data was obtained from CapitalIQ database. The dependent variables were first day initial returns (underpricing), first month returns and first year returns. Using longer periods of time would result in a smaller sample of firms, e.g. using five-year returns would result in a 42% decrease in sample size, therefore this research concentrated its attention only on underpricing, first month and first year returns. Number of observations in this dataset was equal to 351. This dataset had all the observations on the IPOs, which happened in the Pharmaceutical industry with country of interest being United States. To reach representative amount of observations, the timeframe begins in 1980 and finishes in May 2017 to have an opportunity to check first year returns. This way, the oldest observation was from the 1st of July 1983 (Immunex Corporation) and the newest was of 12th of April 2017 (Tocagen, Inc.).

With the next step two control variables (dummies) were constructed in the main dataset:

1. Withdrawn variable - “1” was given to companies that had previously withdrawn a filed IPO, before or after they went public, “0” otherwise;

2. Crisis variable - “1” was given if the IPO was conducted in times of financial crisis, namely the dot-com bubble crisis (March 2001 - November 2001) or the subprime mortgage crisis (December 2007 - June 2009) and “0” otherwise. Following market-timing IPO theory, very few companies went public during these periods.

Data on the age of the firm was obtained from database collected by J. Ritter and available on his website. As a lot of observations had missing tickers, therefore we decided to merge on the variable ‘Cusip’. After manually collecting foundation date for those companies, which still had missing values, our main dataset had 224 observations. For all the companies left, we obtained patent data from the Orbis database. The variable of interest was the ‘number of patents prior to an IPO’. The initial idea was to simply obtain the number of patents from the Orbis database with the filter prior to the first trading day. However, the Orbis database does not allow this kind of filtering, so that first all the available patent data was obtained and date of publishing the patent was obtained and then patent data prior to an IPO was collected manually. The number of observations remained 224.

To check if investors’ perception of market conditions has an influence on the underpricing, the ‘market_condition’ variable was introduced. It was constructed with the average first-day returns monthly data available on J. Ritter’s website. If the average first day return last month was higher than two months ago, we assume the market to be growing from investors’ perspectives and the market condition variable is equal to one. Otherwise, it remains equal to zero. This approach was taken from theoretical work by Rajan and Servaes (2002) to determine behavioural bias of ‘trend chasing’ investors.

IPOs in pharmaceutical industry : determining factors which influence pharmaceutical companies' underpricing and post-IPO returns