UNIVERSITY of AMSTERDAM
AMSTERDAM BUSINESS SCHOOL
MASTER in INTERNATIONAL FINANCE
MASTER THESIS
DOES THE INVESTOR TYPE AFFECT THE FIRST DAY RETURN OF IPOs IN THE TURKISH STOCK MARKET?
PREPARED BY
Murat Bas
THESIS ADVISOR
Jens Martin, PhD, Assistant Professor of Finance, University of Amsterdam
ABSTRACT
This thesis addresses the question of whether the type of investors has any impact on the first day return of initial public offerings (IPOs) in Turkish Stock Market for the period of 2006-2013. In other words, the relationship between the first day return and the percentile distribution of the shares sold in the IPOs among local individual investors, local institutional investors and foreign institutional investors are tested. The aim is to clarify whether any type of investors benefits most, if the stocks are underpriced and vice versa in the case of overpricing. The data show that the shares are underpriced on average of 4.61 % in line with the underpricing phenomenon in many countries. Since the shares are underpriced, it means that some money are left on the table by the issuing companies which is calculated by multiplying the first day return with the total shares issued. Consequently, all the investors who buy the shares in IPO benefit from this underpricing. The total money left on the table or in other words the money which the issuers could not benefit because of the underpricing corresponds totally to USD 590 million. Regarding the impact of the type of investors on the first day return, the regression results show that the local individual investors and foreign institutional investors have an explanatory power on the first day return in such a way that the more the foreign institutions join in the IPO, the higher is the first day return, and the opposite situation for the local individual investors.
TABLE OF CONTENTS
1-INTRODUCTION ... 1
2-LITERATURE REVIEW ... 3
3-BRIEF INFORMATION ABOUT IPO PROCESS IN TURKISH STOCK MARKET ... 8
4-DATA AND DESCRIPTIVE STATISTICS ... 12
5-METHODOLOGY, HYPOTHESIS AND RESULTS ... 16
6-CONCLUSION ... 25
7-REFERENCES ... 27
8-WEBSITES ... 28
1 1-INTRODUCTION
Companies need funds in order to finance their business activities. There are many ways of financing that companies can use. Issuing debt through bank loans, capital increase by shareholders, attracting new investors can be examples for those ways.
One other way of financing is going public which is based on the selling of some shares of the company to investors in a stock market. If a privately held company goes public for the first time, then this process is called as Initial Public Offering (IPO) (Jonathan Berk and Peter DeMarzo, 2011).
IPOs are widely used as source of financing and are very popular in Today’s World, as they have many advantages at the expense of their some disadvantages.
Advantages of IPO for a Company:
It ensures capital to continue to grow. The generated cash can be used also for merger and acquisitions.
The companies can achieve financing with a lower cost and with a long term compared to alternative financings (bank loan etc.).
The shares of the company may win liquidity which provides advantages to current stockholders and employers having stock options.
It may increase the recognition of the company, since publicly traded companies are analyzed regularly by analysts and they take place more commonly on the media. This may contribute to sales of the company.
Since a publicly traded company is subjected to extensive regulations, it may increase the level of corporate governance.
If companies need new financing, then they may again issue new shares through secondary public offerings.
Disadvantages of IPO for a Company:
It necessitates high cost (hiring an investment bank etc.). The current shareholders can lose ownership control.
Since the company would face much more intense scrutiny and regulations and consequently many details about the company's business and its owners will become
2
public, this can give valuable information to competitors. Additionally complying with regulations cause an increase in costs (hiring independent auditors for the approval of balance sheet etc.).
The current shareholders should share the future profits with the new shareholders.
One of the important parts of an IPO process is the valuation through which the stock price is determined. There are many researches in the literature showing that following an IPO, subject stock has a positive return at the end of the first trading day, known as underpricing in the finance literature. In this case, it is clear that some money are left on the table by the issuers, as the stock is sold at a lower price than it is worth.
Because of the underpricing case in IPOs and hence some money are left on the table by the issuers, it is really interesting to search for who benefits from this underpricing. Of course, if the stock price is overpriced, then it is clear that the issuers are benefiting from this situation, where the investors who buy the stock during IPO process suffer losses.
In this thesis, the IPOs in the Turkish Stock Market, known as Istanbul Stock Exchange, for the period of 2006-2013 are analyzed. The thesis is concentrated on whether there is a relationship between first day return and type of investors, which are classified as local individual investors, local institutional investors and foreign institutional investors. In other words, the question is that to which type of investors and at which percentage are the shares sold in IPOs and does it have any effect in the first day return of the shares. The aim is to clarify whether any type of investors benefits most, if the stocks are underpriced and vice versa in the case of overpricing.
In order to test the existence of this relationship, a multiple regression model is built using the estimation method of ordinary least squares.
3 2-LITERATURE REVIEW
There are many researches in the literature studied IPOs. Many of them conclude that underpricing of stocks is a common phenomenon. Some of them are concentrated in the reasons of underpricing. However there are not so many research regarding who benefits from the underpricing and at what level.
Table 1 gives a summary of the equally-weighted average initial returns on IPOs in a number of countries around the world:
Table 1-Equally Weighted Average Initial Returns for 52 Countries
Country Author(s) of Article(s)/Source Sample Size Time Period Average Initial Return
Argentina Eijgenhuijsen & van der Valk; Dealogic 26 1991-2013 4.2% Australia Lee, Taylor & Walter; Woo; Pham; Ritter 1,562 1976-2011 21.8%
Austria Aussenegg 103 1971-2013 6.4% Belgium Rogiers, Manigart & Ooghe; Manigart
DuMortier; Ritter
114 1984-2006 13.5% Brazil Aggarwal, Leal & Hernandez; Saito;
Ushisima
275 1979-2011 33.1% Bulgaria Nikolov 9 2004-2007 36.5% Canada Jog & Riding; Jog & Srivastava; Kryzanowski,
Lazrak & Rakita; Ritter
720 1971-2013 6.5% Chile Aggarwal, Leal & Hernandez; Celis &
Maturana; Dealogic
81 1982-2013 7.4% China Chen, Choi, & Jiang; Jia, Xie & Zhang 2,512 1990-2013 118.4% Cyprus Gounopoulos, Nounis, and Stylianides;
Chandriotis
73 1997-2012 20.3% Denmark Jakobsen & Sorensen; Ritter 164 1984-2011 7.4%
Egypt Omran; Hearn 62 1990-2010 10.4% Finland Keloharju 168 1971-2013 16.9% France Husson & Jacquillat; Leleux & Muzyka;
Paliard & Belletante; Derrien & Womack; Chahine; Ritter; Vismara
697 1983-2010 10.5% Germany Ljungqvist; Rocholl: Ritter; Vismara 736 1978-2011 24.2%
Greece Nounis, Kazantzis & Thomas; Thomadakis, Gounopoulos & Nounis
373 1976-2013 50.8% Hong Kong McGuinness; Zhao & Wu; Ljungqvist & Yu;
Fung, Gul, and Radhakrishnan; Dealogic
1,486 1980-2013 15.8% India Marisetty and Subrahmanyam; Ritter 2,964 1990-2011 88.5% Indonesia Suherman 441 1990-2013 25.0% Iran Bagherzadeh 279 1991-2004 22.4% Ireland Dealogic 38 1991-2013 21.6%
Israel Kandel, Sarig & Wohl; Amihud & Hauser; Ritter
4
Italy Arosio, Giudici & Paleari; Cassia, Paleari & Redondi; Vismara
291 1985-2011 15.7% Japan Fukuda; Dawson & Hiraki; Hebner & Hiraki;
Pettway & Kaneko; Hamao, Packer, & Ritter; Kaneko & Pettway
3,236 1970-2013 41.7% Jordan Al-Ali and Braik 53 1999-2008 149.0%
Korea Dhatt, Kim & Lim; Ihm; Choi & Heo; Mosharian & Ng; Cho; Joh; Dealogic
1,720 1980-2013 59.3% Malaysia Isa; Isa & Yong; Yong; Ma; Dealogic 474 1980-2013 56.2% Mauritius Bundoo 40 1989-2005 15.2%
Mexico Aggarwal, Leal & Hernandez; Eijgenhuijsen & van der Valk; Villarreal
123 1987-2012 11.6% Morocco Alami Talbi; Hearn 33 2000-2011 33.3% Netherlands Wessels; Eijgenhuijsen & Buijs; Jenkinson,
Ljungqvist, & Wilhelm; Ritter
181 1982-2006 10.2% New Zealand Vos & Cheung; Camp & Munro; Alqahtani;
Dealogic
242 1979-2013 18.6% Nigeria Ikoku; Achua; Dealogic 122 1989-2013 13.1% Norway Emilsen, Pedersen & Saettem; Liden;
Dealogic
209 1984-2013 8.1% Pakistan Mumtaz 57 2000-2010 32.0% Philippines Sullivan & Unite; Dealogic 155 1987-2013 18.1% Poland Jelic & Briston; Woloszyn 309 1991-2012 13.3% Portugal Almeida & Duque; Dealogic 32 1992-2013 11.9% Russia Dealogic 64 1999-2013 3.3% Saudi Arabia Al-Anazi, Forster, & Liu; Alqahtani 80 2003-2011 239.8%
Singapore Lee, Taylor & Walter; Dawson; Dealogic 609 1973-2013 25.8% South Africa Page & Reyneke; Ali, Subrahmanyam &
Gleason; Dealogic
316 1980-2013 17.4% Spain Ansotegui & Fabregat; Alvarez Otera;
Dealogic
143 1986-2013 10.3% Sri Lanka Samarakoon 105 1987-2008 33.5% Sweden Rydqvist; Schuster; de Ridder 374 1980-2011 27.2% Switzerland Kunz, Drobetz, Kammermann & Walchli;
Dealogic
164 1983-2013 27.3% Taiwan Chen; Chiang 1,620 1980-2013 38.1% Thailand Wethyavivorn & Koo-smith; Lonkani &
Tirapat; Ekkayokkaya and Pengniti; Vithessonthi
500 1987-2012 35.1% Tunisia Hearn 32 2001-2013 24.3% Turkey Kiymaz; Durukan; Ince; Kucukkocaoglu 355 1990-2011 10.3% United Kingdom Dimson; Vismara; Levis 4,932 1959-2012 16.0% United States Ibbotson, Sindelar & Ritter; Ritter 12,496 1960-2013 16.9%
Source: http://bear.warrington.ufl.edu/ritter/Int.pdf
This Table is an updated version of the table placed in the article of ‘’Initial Public Offerings: International Insights’’ written by Tim Loughran, Jay R, Ritter and Kristian Rydqvist and published in Pacific-Basin Finance Journal in 1994. The table is updated in 2014.
As seen in Table 1, the stocks in many countries finalize the first day with a positive return which confirms the underpricing phenomenon.
5
Rock (1986) states that investors have different information about the fair value of the shares. While uninformed investors join to every IPO, informed investors only buy new shares, if the issue price is less than the fair value. This cause a "winner's curse" for the uninformed investors. Therefore, share must be offered at a discount to hold uninformed investors in the market, because none of the investors group have enough money to absorb the initial public offering.
According to the research of Loughran and Ritter (2004) on US market, the average first-day return on initial public offerings (IPOs) is 7 % in the 1980s, then it doubles to almost 15 % during 1990-1998, before jumping to 65 % during the internet bubble years of 1999-2000 and then reverting to 12 % during 2001-2003. At the end, they conclude that this variation in underpricing could be attributed to the winner’s curse problem and dynamic information acquisition in the 1980s and analyst coverage, side payments to CEOs and venture capitalists are the main reasons for underpricing during the internet bubble.
Sullivan and Unite (1999), study all initial public offerings that occurred in the Philippines from 1987 through 1997. They conclude that investors joining in the initial public offer of a Philippine company earn large first-day returns and these returns, as in other countries, can be attributed to the underpricing of IPOs. Additionally they state that since initial returns of 22.69 % are greater than those documented for U.S. IPOs, this confirms the view that investors in smaller countries with a less developed capital market are subject to greater risks.
Loughran and Ritter (2002) state that companies going public in US during 1990-1998 left USD 9.1 million on the table on average, a number that worked out to an aggregate of more than USD 27 billion, where the money left on the table is defined as the first-day price gain multiplied by the number of shares sold.
Ljungqvist (1997) expands the research of Wasserfallen and Wittleder titled as ‘’Pricing Initial Public Offerings: Evidence from Germany’’ which is published in European Economic Review in 1994 and which provides evidence of the well-known underpricing phenomenon for 92 German IPOs coming to market in 1961–1987. In the expanded study, Ljungqvist uses a larger sample of 189 firms from 1970–1993. In that paper, he finds out that stock market returns, the macroeconomic climate, insider retention rates and inverse of real gross proceeds or inverse offer size all affect underpricing positively which is calculated as 9.2 % for the relevant period.
Pons-Sanz (2005), in his study, uses a unique sample of 175 Spanish equity offerings from 1985 to 2002 to test who benefits from IPO underpricing and why. He concludes that,
6
institutions receive nearly 75 % of the profits in underpriced issues, while they have to bear only 56 % of the losses in overpriced offerings. He states that; superior information regarding first day underpricing cannot completely explain the institutional abnormal profits, underwriters are better informed about the companies they take public, and use that information to favor their long term clients, the preferential treatment of institutional investors, however, does not come at the expense of retail investors, retail investors earn positive profits from participating in the new issues market, the driving factor behind the relative retail large allocation in overpriced issues when compared to underpriced offerings is not the underwriter allocation bias in favor of institutional investors and retail investors subscribe more heavily to underpriced issues, consistent with individuals being partially informed.
As far as the researches done for Istanbul Stock Exchange are concerned, some studies within the context of underpricing are listed below:
Erdoğan, Güntürkün and Gürarda (2012), study the Istanbul Stock Exchange for the period of 2006-2011. They conclude that there exists underpricing in Turkey for that period. Regarding which macroeconomic factors are affecting the underpricing, they find that Istanbul Stock Exchange 100 index, consumer price index, interest rate, GDP per capita do not affect underpricing. However, consumer confidence index (CCI) is weakly correlated to initial return in that period which is the only determinant of underpricing in that period. Lastly, world oil price is not a determinant of underpricing if the period is considered. They analyze also the pre-crisis and after-crisis periods separately (2006-2007 and 2010-2011). ISE 100 index, consumer price index, interest rate, GDP per capita don’t have any effect on initial return in these periods, too, where the effect of CCI increases. When world oil prices are concerned, it is not a determinant for the pre-crisis period. However after the crisis period it affects the underpricing.
Baspinar (2008), studies 240 public offerings taken place in Istanbul Stock Exchange between January 1993 and May 2007 with the purpose of identifying how far the stock prices set in the process of public offering deviates from their fair values; of marking those who derive benefit from this deviation, of testing the methods used to reduce price deviations and of proposing a model to ensure fair pricing. In the performed analyses, the findings are such that underpricing appears at an average of 9.16 % on the first day of public offerings and therefore the stocks offered to public are not priced on their fair values, mainly foreign institutional investors and domestic individual investors benefit from this underpricing and transactions applied to ensure price stability prove to be efficient in general. In this framework, Internet Based Single-Price Closed-Bid Auction method is a suggested method for reasons that it results in a lower
7
underpricing, that it provides lower interagency commission and that it creates a more efficient equal opportunity, with respect to other public offering methods.
Kucukkocaoglu (2008), addresses the question of what kind of selling and underwriting procedure might be preferred for controlling the amount and volatility of underpricing in the Istanbul Stock Exchange (ISE) using data in the period of 1993-2005. He compares the three substantially different IPO methods available in the ISE. One is very similar to the book building mechanism used in the U.S., another is the fixed price offer, and the third one is the sale through the stock exchange method. He concludes that the empirical analysis reveals significant first day underpricing of 7.01 % in fixed price offer, 11.47 % in book building mechanism, and 15.68 % in sale through the stock exchange method. Finally, he states that fixed price offers can better control the impact of market information on underpricing than sale through the stock exchange method.
Otlu and Olmez (2011) search the level of underpricing in the Istanbul Stock Exchange and the factors effecting the short term performance (not the first day performance) of the stocks for the period of January 2006-June 2011. They find out that the underpricing level is 6.99 % and standard deviation and the first day return of stocks are the independent variables which are explaining the short term performance most. The short term performance is determined as the 1st, 2nd, 3rd and 4th week of trading.
8
3-BRIEF INFORMATION ABOUT IPO PROCESS IN TURKISH STOCK MARKET
The Istanbul Stock Exchange (ISE) was founded as an autonomous, professional organization at the end of 1985. In 2013, Istanbul Stock Exchange, Istanbul Gold Exchange and Turkish Derivatives Exchange were merged under the institutional body of Borsa Istanbul A.S. (BIST). Now, BIST provides trading in equities, bonds and bills, revenue-sharing certificates, private sector bonds, foreign securities, precious metals etc. As of 30.06.2014, the shares of 403 companies are traded in the equity market of ISE (including 17 ETFs) and total market capitalization corresponds to USD 269 billion.
The Capital Markets Board of Turkey (CMB) is the main regulatory and supervisory authority in charge of the securities markets. Empowered by the Capital Markets Law, the CMB is regulating and supervising the markets, investment instruments and institutions. BIST has some self-regulatory authority on its members, but major decisions are subject to CMB approval.
The IPO process in Turkey is regulated through the communiques mentioned below:
Communiqué Serial: II No:5.1 on Principles Regarding Prospectus and Issue Documents
Communiqué Serial: II No:5.2 on Principles Regarding Issue of Capital Market Instruments
Below brief information on the steps of public offerings in Turkey are given which are compiled from the above mentioned communiques and documents (reports, snapshots, handbooks etc.) released in the website of the Capital Markets Board of Turkey (www.spk.gov.tr), Borsa Istanbul A.S. (www.borsaistanbul.com) and Turkish Capital Markets Association
(www.tspakb.org.tr).
Step 1: Preparation for the IPO
This step includes the resolution of the company board for IPO, adaptation of the articles of association to capital market regulations, audit of the financial reports by an independent auditor, preparing legal report by lawyers, selection of a brokerage firm or an investment bank with the intermediary purposes and determination of the company value and sale price. Additionally, detailed information about the company is presented in a prospectus, including
9
the scope of the business, sector analysis, current shareholder structure, financial position, etc. Circulars, on the other hand, should summarize general features of the public offering.
Step 2: Application to BIST and the CMB and Inspections
After step 1, the candidate company should submit all the required documents to the CMB in order to receive its approval for public offering, and at the same time, it should apply to BIST to be listed in the relevant market of the Exchange. Following the successfully filing of all required documentations, experts from both the CMB and BIST pay a visit to the company’s headquarters and production facilities to perform an on-site investigation. Additionally, the prospectus and circulars are reviewed by the CMB.
Step 3: Application to the Settlement and Custody Bank (Takasbank), the Central Registry Agency (CRA) and Public Disclosure Platform (PDP)
After step 2, the company should get an ISIN code (International Securities Identification Number) defined by the International Organization for Standardization (Standard no: 6166) in order to create a uniform exchange and custody process to ensure smoother transactions. This code is provided by Custody Bank (Takasbank) in Turkey whose major purpose and activity are to provide clearing, settlement and custody services within the framework of capital market and related exchange regulations as well as rendering investment banking services within the scope of the Banking Law and other banking regulations.
All issuers, whose shares are expected to be traded in the Exchange, should apply to Central Registry Agency (CRA) for membership prior to the trading date in order to satisfy the terms of the relevant legislation. CRA is established to keep the records of the capital market instruments as well as rights affixed thereon with respect to issuers, intermediary institutions, and right owners, oversee the consistency of actual records maintained by member groups, and manage and represent the Investors Protection Fund.
Furthermore, membership to the Public Disclosure Platform (PDP) is obligatory for companies whose capital markets instruments are to be traded in the Exchange. PDP is an electronic platform, utilized by the traded companies to announce their material disclosures using digital signatures.
10 Step 4: Designation of the Trading Market
The market (national market etc.) within the Exchange in which the issuer’s shares will be traded is determined by the decision of the Board of BIST, after the Exchange experts complete their investigation.
Step 5: Approval of the Prospectus
The CMB examines the IPO prospectus from a public disclosure standpoint and once the prospectus is seen fit for its purpose, it is approved to grant permission to the public offering.
Step 6: Public Offering and Reporting the Sales Results
After step 5, the prospectus is disclosed to the public and the shares are offered to the public via an investment/brokerage house or sometimes via a consortium of investment/brokerage houses. Once the sales results are finalized, the investment/brokerage house publishes the figures at the Public Disclosure Platform (www.kap.gov.tr) and also sends the results to the CMB and BIST.
Regarding selling methods of shares, the sale of shares to investors is conducted in two ways, book building and sales on the stock exchange. In the book building method, investors’ demand for shares is collected through “fixed price” or “price bid” or within “price range”. In sales on the stock exchange, shares are offered to the public on the Primary Market of the Exchange.
Step 7: Listing and the Inception of Trading in the Exchange
A final evaluation regarding whether the sales results satisfy the relevant requirements is carried out by the Board of BIST. Subsequently, shares begin trading in the related market following the announcement of BIST in Public Disclosure Platform.
11
Table 2 presents an average timeframe for the IPO process in Turkey:
Table 2-Timeframe for the IPO Process in Turkey
Timeframe Week 1 Week 2 Week 3 Week 4 Week 5 Week 6
Application to the CMB and BIST 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 Investigations of the CMB and BIST Decision of the Board of BIST Prospectus Approval by CMB Disclosure of the Prospectus and the Sales Announcement to the Account Owners Public Offering Sales Results and Reporting to BIST Beginning of Trading Source: www.borsaistanbul.com
12 4-DATA AND DESCRIPTIVE STATISTICS
Since in this thesis, whether there is a relationship between first day return and the type of investors in the initial public offerings in Turkish Stock Market which is known as Istanbul Stock Exchange, is searched, the data relate to the first day return and the percentile distribution of the shares sold in the IPOs among local individual investors, local institutional investors and foreign institutional investors.
The data are limited to IPOs realized in the period of 2006-2013, as the thesis is focused on
that period. According to data provided from the website of Istanbul Stock Exchange (www.borsaistanbul.com), 99 companies went public during that period.
Based on the relevant law in Turkey, in order to save the rights of investors and to set transparency, the companies whose shares are traded or to be traded in Istanbul Stock Exchange should share important information about themselves with Public Disclosure Platform which is a public institution and which stores those information for the use of investors in its website (www.kap.gov.tr). The information of at which rate the sold IPO shares are distributed among investor types should also be disclosed to public through the website of Public Disclosure Platform. However, when the website is checked, it is seen that the reporting of this information is not at a standardized form and the relevant information is missing or partly available for 41 companies. In order to complete the missing information, I had contacted to the relevant companies and/or to the Lead Underwriters. However it was not possible to get the relevant information. Hence, information about 58 companies out of 99 companies are available for my research which I think would be enough to go further in my thesis. The first day return of the relevant shares is collected from the websites releasing data about the stock markets including the website of Istanbul Stock Exchange.
Table 3 below presents some brief statistics about the IPOs for the period of 2006-2013 in Istanbul Stock Exchange. It is seen that, the number of companies went to public decreases in 2008 and 2009, mainly because of the effects of the global financial crisis in 2008. However, starting from 2010, the number increases as the effects of the crisis diminished. Regarding the first day return, the data prove that the shares are underpriced on average of 4.61 %. Since the shares are underpriced, it means that some money are left on the table by the issuing companies and the investors who bought the shares in IPO benefit from this underpricing. The total money left on the table or in other words the money which the issuers could not benefit because of the underpricing corresponds totally to USD 590 million.
13
Table 3-Brief Statistics for IPOs for the Period of 2006-2013 in Istanbul Stock Exchange
Year Number of Companies Went
Public
Average First Day Return (%)
Money Left on the Table (USD)
(First day Return*Total Shares Issued)
2006 15 6.39% 124,785,518 2007 9 7.96% 306,711,783 2008 2 10.72% 9,095,118 2009 1 -7.47% -561,290 2010 22 4.51% 86,072,426 2011 25 6.52% 44,786,739 2012 16 6.23% 4,988,184 2013 9 2.02% 14,384,167 Total 99 4.61% 590,262,647
In this thesis, the portion of IPO shares sold to local individual investors, local institutional investors and foreign institutional investors are the main explanatory variables for the regression, since I search whether there is a relationship between first day return and the distribution of IPO shares to the type of investors and whether any of those investors benefits most from underpricing, where the first day return is the dependent variable.
In addition to the main explanatory variables mentioned above, some control variables are used in order to test whether the individual regression coefficients are significant while controlling for some other variables. The control variables are determined as; net profit margin, leverage ratio, share overhang, age and industry.
The accounting measures which are net profit margin and leverage ratio are obtained from the last annual audited financials of the companies before they went public. Net profit margin is the ratio of profitability and calculated as net income divided by total sales, where leverage ratio shows indebtedness of the companies and relates to financial risk and calculated as total liabilities divided by total assets. Share overhang refers to the percentage of the shares of the companies which goes to public which can be used as an indicator of how much the old shareholders would lose control in their companies. Age of the companies refers to the years between the date of establishment and going public. Industry is the sector in which the company operates.
Table 4 below presents some brief statistics about the above mentioned main explanatory variables and control variables. Since data are available only for 58 companies because of the
14
reasons mentioned above, the statistics could be given for 58 companies, not for the whole companies went to public in the period of 2006-2013.
Table 4-Summary Statistics for Dependent Variable, Main Explanatory Variables and Control Variables
Statistics Dependent Variable Main Explanatory Variables Control Variables First Day Return Portion of Loc. Ind. Inv. in IPO Shares Sold Portion of Loc. Inst. Inv. in IPO Shares Sold Portion of Foreign Inst. Inv. in IPO Shares Sold Net Profit Margin
Lev. Age Share Over. Mean 4.26% 62.28% 9.53% 28.20% 36.37% 58.04% 13.94 32.64% Median 0.22% 61.96% 7.16% 20.00% 9.58% 49.29% 12.53 29.57% Maximum 21.82% 100.00% 95.80% 75.00% 651.48% 434.33% 69.68 100.00% Minimum -7.69% 4.20% 0.00% 0.00% -2.19% 1.13% 1.00 5.34% Std. Dev. 8.91% 27.17% 14.75% 27.01% 102.02% 56.94% 11.95 20.32% Skewness 0.87 -0.10 4.03 0.38 5.01 5.01 2.18 2.07 Kurtosis 2.42 1.76 22.33 1.58 28.23 34.16 10.12 7.62 Total Obs. 58 58 58 58 58 58 58 58
In Table 4, it is seen that local individual investors bought 62.28 % of the total shares in money terms (IPO revenue) on average. This statistics indicates that the local individual investors showed great interest in IPOs. The age of the companies ranges from 1 year to 70 years, where the average age is 13.94 years.
As per the distribution of first day returns, we may say that the returns are not normally distributed, as the skewness differs from zero and the kurtosis differ from 3, which are the requirements of being normal distribution.
15
Figure 1 below gives information on the sectoral diversification:
As seen in Figure 1, there is no concentration in any sector, where real estate investment trusts (REITs) have the biggest share among the companies went public, with a percentage of 22 %.
Holding 10% Finance 16% Production 17% Wholesale and Retail Trade 19% REIT 22% Others 16%
16 5-METHODOLOGY, HYPOTHESIS AND RESULTS
The thesis is concentrated on whether there is a relationship between the first day return and the type of investors who buy the shares of the companies which went public in Turkish Stock Market during the period of 2006-2013. The type of investors are classified as local individual investors, local institutional investors and foreign institutional investor.
In order to test the existence of this relationship, the multiple regression model below is described:
Y = a + β1X1+ β2X2+ β3X3+ ℇ
Where;
Y =First Day Return
X1 =Percentage of the Shares Bought by Local Individual Investors X2 =Percentage of the Shares Bought by Local Institutional Investors X3 =Percentage of the Shares Bought by Foreign Institutional Investors ℇ =Error Term
In this regression, the first day return (Y) is assigned as the dependent variable and shows the return between the offer price of the shares and the closing price of the shares at the end of the first trading day, percentage of the shares bought by local individual investors (X1), percentage of the shares bought by local institutional investors (X2), and percentage of the shares bought by foreign institutional investors (X3) are assigned as main explanatory variables or main independent variables.
Regarding the model, the hypothesis are as follows:
H0: There is no relationship between dependent variable and the independent variables.
Hence; β1=0, β2=0, β3=0
H1: There is a relationship between dependent variable and the independent variables.
17
The above mentioned hypothesis have the structure of joint hypothesis. Hence, they are tested through F-statistics. Single hypothesis (e.g. β1=0 or β1≠0) are tested through t-statistics for
significance.
First of all, the main independent variables defined above are subjected to the regression. Later, the control variables are added to that regression model in the following parts of this section in order to test whether the regression coefficients of the main independent variables are significant while controlling for some other variables.
As the estimation method for the regression, Ordinary Least Squares method is used. As the econometric software, EViews 8 Student Version is used. The data is cross sectional.
Table 5 below presents the abbreviations of variables which are used in following tables:
Table 5-The Abbreviations in the Regression Tables Regarding Variables
Abbreviation Status Definition
Return Dependent variable The return between the offer price of the shares and the closing price of the shares at the end of the first trading day Locind Main independent variable Percentage of the shares bought by local individual
investors
Locinst Main independent variable Percentage of the shares bought by local institutional investors
Forinst Main independent variable Percentage of the shares bought by foreign institutional investors
Margin Control variable Net profit margin (net income/total sales) Leverage Control variable Leverage ratio (total liabilities/total assets)
Shareover Control variable Share overhang (the percentage of the shares going to public)
Agedum Control variable/Dummy Age dummy. The companies which are older than 10 years take the value of 1, where the companies which are 10 years old or younger take the value of 0.
Findum Control variable/Dummy Finance sector dummy. The companies in financial sector take the value of 1, the others take the value of 0.
Holdum Control variable/Dummy Holding sector dummy. The companies in holding sector take the value of 1, the others take the value of 0.
Reitdum Control variable/Dummy REIT sector dummy. The companies in REIT sector take the value of 1, the others take the value of 0.
Produm Control variable/Dummy Production sector dummy. The companies in production sector take the value of 1, the others take the value of 0. Wholedum Control variable/Dummy Wholesale/retail sector dummy. The companies in
wholesale/retail sector take the value of 1, the others take the value of 0.
As mentioned before, first of all, the main independent variables are subjected to the regression. In order to ensure soundness for the model, firstly the data are checked for multicollinearity and heteroscedasticity. Multicollinearity relates to the correlation between
18
independent variables and existence of high correlation would mean the estimated coefficients are not biased, but the t-statistics will fall. Heteroscedasticity relates to existence of a non-constant variance for the error term and it would mean the estimated coefficients are not biased, but the t-statistics will be misleading.
Table 6 below presents the correlation between the main independent or main explanatory variables:
Table 6-Correlation Matrix for the Main Independent Variables
Locind Locinst Forinst
Locind 1.00 -0.28 -0.85
Locinst -0.28 1.00 -0.26
Forinst -0.85 -0.26 1.00
As seen in Table 6, since percentage of the shares bought by local individual investors and percentage of the shares bought by foreign institutional investors are very highly correlated with each other, in order to eliminate this problem, instead of running a regression including all independent variables, it is decided to run regressions in which the independent variables take place in the regression alone or pairwise. Hence, the below regressions are run:
Regression 1: Only Locind is the independent variable. Regression 2: Only Locinst is the independent variable. Regression 3: Only Forinst is the independent variable.
Regression 4: Locind and Locinstare the independent variables. Regression 5: Locind and Forinstare the independent variables. Regression 6: Locinst and Locinstare the independent variables.
Table 7 below presents the regression results in which only the main independent variables are included. The control variables are added to these regression models in the latter parts of this section.
19
Table 7- Results of the Regressions Including Only Main Independent Variables
The dependent variable is first day return. In the first 3 regressions, the main independent variables take place alone, where in the last 3 regressions they take place pairwise. The control variables are not included here. The first line of the results shows the coefficients of the independent variables, where the standard errors are given in parenthesis.
Statistics Reg. 1 Reg. 2 Reg. 3 Reg. 4 Reg. 5 Reg. 6
Constant 0.118312 (0.027628) 0.043210 (0.014093) 0.007373 (0.015874) 0.132752 (0.031417) 0.057078 (0.069076) -0.000424 (0.019110) Locind * -0.121564 (0.040717) - - * -0.133176 (0.042472) -0.057502 (0.077754) - Locinst - -0.006338 (0.080742) - -0.075674 (0.078231) - 0.057502 (0.077754) Forinst - - *0.124952 (0.040824) - 0.075674 (0.078231) *0.133176 (0.042472) F Statistic 8.913757 0.006163 9.368084 4.919604 4.919604 4.919604 P-Value of F Statistic **0.004191 0.937709 **0.003387 ***0.010825 ***0.010825 ***0.010825 R-Squared 0.137317 0.000110 0.143313 0.151748 0.151748 0.151748 Adjusted R-Squared 0.121912 -0.017745 0.128015 0.120902 0.120902 0.120902 P-Value of Heteroscedasticity ****0.8178 ****0.8770 ****0.9273 ****0.9596 ****0.9119 ****0.9991 ****0.8218 ****0.9000 ****0.9302 ****0.9627 ****0.9184 ****0.9992 Observations 58 58 58 58 58 58 Notes
*The independent variable is significant with the confidence level of 99 %. Probability of t statistic (P value-t
statistic) is less than 0.01.
**The joint null hypothesis of no relation should be rejected with the confidence level of 99 %. Probability of F
statistic (P value-F statistic) is less than 0.01.
***The joint null hypothesis of no relation should be rejected with the confidence level of 95 %. Probability of F
statistic (P value-F statistic) is less than 0.05.
****White test is used for heteroscedasticity. The null hypothesis of no heteroscedasticity should be accepted,
since the probability of Chi-Square of Obs*R-squared (presented in the first line) and probability of Chi-Square of Scaled explained SS (presented in the second line) are considerably in excess of 0.05.
20
Table 7 shows also the statistics regarding the heteroscedasticity, where the White test is used. Since the probability of Chi-Square of Obs*R-squared and probability of Chi-Square of Scaled explained SS are considerably in excess of 0.05, we may conclude that there is no heteroscedasticity regarding the regressions.
When the results of regressions are analyzed:
First 3 Regressions: The independent variables take place alone in those regressions.
According to results, based on the p-value of t-statistics and F-statistics, we may conclude that; percentage of the shares bought by local individual investors is a significant variable and has a negative relation with the first day return, since the coefficient of this variable is negative (Regression 1). On the contrary to that, percentage of the shares bought by local institutional investors is not a significant variable at least at the confidence level of 99 % or 95 % and has no effect in the first day return (Regression 2). As far as percentage of the shares bought by foreign institutional investors is concerned, it is a significant variable and has a positive relation with the first day return, since the coefficient of this variable is positive (Regression 3).
Last 3 Regressions: The independent variables take place pairwise in those regressions. The
results of these regressions are confirming the significance of the independent variables in the first 3 regressions. According to Regression 4, where percentage of the shares bought by local individual investors and percentage of the shares bought by local institutional investors are independent variables together, again percentage of the shares bought by local individual investors is significant, where percentage of the shares bought by local institutional investors is not. According to Regression 5, as mentioned before, since there is a high correlation between percentage of the shares bought by local individual investors and percentage of the shares bought by foreign institutional investors, the results are not meaningful. According to Regression 6, where percentage of the shares bought by local institutional investors and percentage of the shares bought by foreign institutional investors are independent variables together, again percentage of the shares bought by foreign institutional investors is significant, where percentage of the shares bought by local institutional investors is not.
As seen in Table 7, the statistics of R-Squared, which shows how well the regression model actually fits the data or how well does the model containing the explanatory variables actually explain variations in the dependent variable, is at a low level. However, our main aim is just to test whether there is a relationship between the dependent variable and independent variables, but not to clarify all the variables which fully explain the independent variable. Hence, this situation does not destroy the results interested in.
21
After the relationship between dependent and main independent variables are analyzed, some control variables, which are net profit margin, leverage ratio, share overhang, age and industry are added to the above defined econometric model in order to test whether the regression coefficients of the main independent variables are significant while controlling for some other variables.
The data for age and industry are used in a dichotomous form where dummy variables are assigned for them. Regarding age, the companies are split as young and old companies, where the companies older than 10 years are defined as old and the companies which are 10 years old or younger are defined as young companies. The old companies take the value of 1, where the young companies take the value of 0. Regarding the industry, the companies are distributed among the sectors of finance, holding, REIT, production, wholesale/retail and others. For each sector except the sector of ‘’others’’, a dummy variable is assigned. Because as a rule of thumb in statistics, at least one of the categories should be omitted from the model, in order not to create multicollinearity.
Again a correlation check is done by including main independent variables and non- dichotomous control variables together.
Table 8 below presents the correlations:
Table 8-Correlations Between All Independent Variables
Locind Locinst Forinst Margin Leverage Shareover
Locind 1.00 -0.28 -0.85 -0.05 -0.07 0.21 Locinst -0.28 1.00 -0.26 0.11 -0.19 0.10 Forinst -0.85 -0.26 1.00 -0.01 0.18 -0.27 Margin -0.05 0.11 -0.01 1.00 -0.14 -0.07 Leverage -0.07 -0.19 0.18 -0.14 1.00 -0.10 Shareover 0.21 0.10 -0.27 -0.07 -0.10 1.00
As seen in Table 8, the control variables have no high correlation with themselves and with the main independent variables. These control variables are added to the regressions run for the main independent variables which is presented in Table 7.
Table 9 below presents the regression results in which the independent variables and control variables are included together. The heteroscedasticity test results are also given in the table.
22
Table 9- Results of the Regressions Including the Main Independent Variables and Control Variables
The dependent variable is first day return. In the first 3 regressions, the main independent variables take place alone, where in the last 3 regressions they take place pairwise. The control variables are also included in each regression. The first line of the results shows the coefficients of the variables, where the standard errors are given in parenthesis.
Statistics Reg. 1 Reg. 2 Reg. 3 Reg. 4 Reg. 5 Reg. 6
Constant 0.147215 0.078283 0.036212 0.153951 0.111393 0.027240 Locind *-0.118517 - - *-0.126711 -0.084154 - (0.042135) (0.044999) (0.075628) Locinst - 0.029838 - -0.042557 - 0.084154 (0.078332) (0.077515) (0.075628) Forinst - - **0.113940 - 0.042557 *0.126711 (0.043621) (0.077515) (0.044999) Margin -0.005285 -0.006422 -0.003756 -0.004653 -0.004653 -0.004653 (0.011947) (0.012935) (0.012095) (0.012092) (0.012092) (0.012092) Leverage **0.045778 **0.049851 **0.040587 **0.043615 **0.043615 **0.043615 (0.018566) (0.020339) (0.018969) (0.019116) (0.019116) (0.019116) Shareover -0.091665 -0.126583 -0.077928 -0.083943 -0.083943 -0.083943 (0.068710) (0.073910) (0.070612) (0.070641) (0.070641) (0.070641) Agedum -0.049461 -0.029430 -0.045677 -0.049787 -0.049787 -0.049787 (0.027125) (0.028239) (0.027153) (0.027336) (0.027336) (0.027336) Findum 0.066004 0.070490 0.065706 0.065528 0.065528 0.065528 (0.041929) (0.045221) (0.042360) (0.042253) (0.042253) (0.042253) Holdum 0.006524 -0.003137 0.012287 0.009351 0.009351 0.009351 (0.041713) (0.045105) (0.042366) (0.042341) (0.042341) (0.042341) Reitdum -0.024814 -0.031040 -0.024634 -0.024235 -0.024235 -0.024235 (0.038897) (0.041915) (0.039301) (0.039204) (0.039204) (0.039204) Produm -0.028172 -0.053998 -0.032419 -0.027530 -0.027530 -0.027530 (0.036727) (0.038357) (0.036852) (0.037021) (0.037021) (0.037021)
23 Wholedum 0.020121 -0.002076 0.024862 0.023626 0.023626 0.023626 (0.035652) (0.037837) (0.036558) (0.036483) (0.036483) (0.036483) F Statistic 2.928005 1.849128 2.776741 2.649662 2.649662 2.649662 P-Value of F Statistics ***0.006198 0.077576 ***0.008834 ****0.010263 ****0.010263 ****0.010263 R-Squared 0.383849 0.282347 0.371384 0.387861 0.387861 0.387861 Adjusted R-Squared 0.252754 0.129655 0.237636 0.241479 0.241479 0.241479 P-Value of Heterosce. *****0.1895 *****0.2219 *****0.2587 *****0.2687 *****0.2840 *****0.2732 *****0.3676 *****0.6224 *****0.5752 *****0.5319 *****0.5484 *****0.5369 Observations 58 58 58 58 58 58 Notes
*The independent variable is significant with the confidence level of 99 %. Probability of t statistic (P value-t statistic) is less than 0.01.
**The independent variable is significant with the confidence level of 95 %. Probability of t statistic (P value-t
statistic) is less than 0.05.
***The joint null hypothesis of no relation should be rejected with the confidence level of 99 %. Probability of F
statistic (P value-F statistic) is less than 0.01.
****The joint null hypothesis of no relation should be rejected with the confidence level of 95 %. Probability of
F statistic (P value-F statistic) is less than 0.05.
*****White test is used for heteroscedasticity. The null hypothesis of no heteroscedasticity should be accepted,
since the probability of Chi-Square of Obs*R-squared presented in the first line and probability of Chi-Square of Scaled explained SS presented in the second line are considerably in excess of 0.05.
When the results of regressions presented in Table 9 are analyzed:
Considering the t-statistics and F-statistics; the results for the main explanatory variables are confirming the results in the previous regressions presented in Table 7. In other words, percentage of the shares bought by local individual investors and percentage of the shares bought by foreign institutional investors are significant variables at the confidence level of either 99 % or 95 %, where percentage of the shares bought by local institutional investors is not. Additionally the coefficients decreased just slightly.
The results above show that the control variables have no major impact on the significance and the level of the coefficients of the main independent variables.
24
When the controlling variables are concerned from the perfective of explaining the dependent variable, it is seen that, all the control variables, except the leverage ratio are not significant meaning that they have no explanatory power on the dependent variable which is the first day return as the p-value of the t-statistics are high. Only leverage is found significant and has a positive relationship with the dependent variable.
25 6-CONCLUSION
In this thesis, whether the type of investors has any impact on the first day return of initial public offerings (IPOs) in Turkish Stock Market for the period of 2006-2013 is searched. In other words, the relationship between the first day return and the percentile distribution of the shares sold in the IPOs among local individual investors, local institutional investors and foreign institutional investors are tested. The aim is to clarify whether any type of investors benefits most, if the stocks are underpriced and vice versa in the case of overpricing.
During 2006-2013, 99 companies went public in Turkey. The data show that the shares are underpriced on average of 4.61 %. This results confirm that the underpricing phenomenon, which means that subject stocks have a positive return on average at the end of the first trading day and the stocks are sold at a lower price than they are worth, is also valid for that period in Turkey in line with the underpricing phenomenon in many countries. Since the shares are underpriced, it means that some money are left on the table by the issuing companies and all the investors who bought the shares in IPO benefit from this underpricing. The total money left on the table or in other words the money which the issuers could not benefit because of the underpricing corresponds totally to USD 590 million. This situation confirms the study of Loughran and Ritter (2002) which clarifies that on an aggregate level more than USD 27 billion are left on the table during 1990-1998 in US market, where the money left on the table is defined as the first-day price gain multiplied by the number of shares sold.
Regarding the impact of the type of investors on the first day return, relevant regressions are run. The regressions could not run over all the sample size which is 99 companies, as the relevant data for the independent variables which are the percentage of the IPO shares sold to local individual investors, local institutional investors and foreign institutional investors are not available. Hence the sample could cover 58 companies.
First of all, the main independent variables defined above are subjected to the regression. Later, the control variables, which are net profit margin, leverage ratio, share overhang, age and industry are added to the regression model in order to test whether the regression coefficients of the main independent variables are significant while controlling for some other variables.
According to the results of regressions, the percentage of the shares bought by local individual investors is a significant variable and has a negative relation with the first day return, the percentage of the shares bought by local institutional investors is not a significant variable and
26
has no effect in the first day return and the percentage of the shares bought by foreign institutional investors is a significant variable and has a positive relation with the first day return. Hence the results show that the local individual investors and foreign institutional investors have an explanatory power on the first day return in such a way that the more the foreign institutions join in the IPO, the higher is the first day return, and the opposite situation for the local individual investors.
When the controlling variables are concerned from the perfective of explaining the dependent variable, it is seen that, all the control variables, except the leverage ratio are not significant meaning that they have no explanatory power on the first day return. Only leverage is found significant and has a positive relationship with the first day return.
I need to say that percentile distribution of shares among investor types would not be enough to fully explain the first day return as there could be many factors effecting the first day return. Hence the statistics of R-Squared, which shows how well the regression model actually fits the data or how well does the model containing the explanatory variables actually explain variations in the dependent variable, is at a low level. However, since the aim is just to test whether there is a relationship between the dependent variable and independent variables, but not to clarify all the variables which fully explain the independent variable, this situation does not destroy the results interested in.
27 7-REFERENCES
1- Baspinar, Ahmet, 2008, Interest Relations Based on the Price Anomalies in Public Offerings, PhD Thesis, Baskent University, Turkey
2- Berk Jonathan and DeMarzo Peter, 2011, Corporate Finance, Second Edition, p. 776 3- Erdoğan Hilal Hümeyra, Güntürkün Mustafa Hakan and Gürarda Şevin, ‘’ Impact of
Macroeconomic Factors on Underpricing of Initial Public Offerings before and after the Recent Global Financial Crisis: Evidence from Istanbul Stock Exchange’’, Journal of Applied Finance & Banking, Volume 2, No. 5, p. 261-273
4- Kucukkocaoglu, Guray, 2008, A Comparison of the IPO Methods, International Research Journal of Finance and Economics, Issue 13, p. 73-87
5- Loughran Tim and Ritter Jay, 2002, “Why Don’t Issuers Get Upset About Leaving Money on the Table in IPOs?’’, The Review of Financial Studies, Special Edition, Volume 15, No. 2, p. 413-443
6- Ljungqvist, Alexander P., 1997, Pricing Initial Public Offerings: Further evidence from Germany, European Economic Review 41 p. 1309-1320
7- Loughran Tim and Ritter Jay, 2004, “Why Has IPO Underpricing Changed Over Time?’’, Journal of the Financial Management Association International, Volume 33, Issue 3, p. 5-37
8- Otlu, F., Olmez. S., 2011, Researching Short-Term Price Performance and Factors Influencing Price Performance of Stock Certificates Offered to Public for the First Time, An Implementation In Istanbul Stock Exchange, Journal of Academic Approaches, Volume 2, Issue 2
9- Pons-Sanz, V., 2005, Who Benefits from IPO Underpricing? Evidence Form Hybrid Bookbuilding Offerings, European Central Bank, Working Paper Series, No. 428 10- Rock, K., 1986, ‘’Why New Issues Are Underpriced?’’, Journal of Financial Economics,
Volume 15, Issues 1-1, p. 187–212
11- Sullivan, Michael J., Unite, Angelo A., 1999, ‘’ The Underpricing of Initial Public Offerings in the Philippines from 1987 to 1997’’, Review of Pacific Basin Financial Markets and Policies; Volume 2, Issue 3, p. 285-300
28 8-WEBSITES 1- http://bear.warrington.ufl.edu/ritter/Int.pdf 2- www.spk.gov.tr 3- www.borsaistanbul.com 4- www.tspakb.org.tr 5- www.kap.gov.tr
