This paper 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 1993-2005 firm and issue data, we compare 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. 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, we also show that fixed price offers can better control the impact of market information on underpricing than sale through the stock exchange method.

1. Introduction

Extensive amount of research from a variety of different markets have documented the presence of first-day underpricing upon the listing of initial public offerings. The evidence is well documented by Loughran, Ritter, and Rydqvist (1994) and Ritter1 (1998), (2003) in many developed and emerging markets. In developed markets, in the absence of restrictions on intra-day price movements, first-day underpricing is observed in broad price bands. However, in emerging markets, in the presence of daily volatility limits, first-day underpricing is observed in narrow price bands. In contrast to the daily price limits, significant positive short run returns are observed in a number of emerging markets and substantial amount of money is “left on the table" by issuers. Besides empirical evidence, most of the theoretical models explaining IPO underpricing are grouped under four broad headings by Ljungqvist (2005), these are (i) information asymmetry between the investors, the issuing firm and the underwriter, these models assume that one of these parties knows more than the others, (ii) institutional reasons, institutional theories focus on three features of the marketplace: litigation, banks' price stabilizing activities once trading starts, and taxes, (iii) control considerations, control theories argue that underpricing helps shape the shareholder base so as to reduce intervention by outside investors once the company is public, (iv) behavioral approaches, behavioral theories assume either the presence of 'irrational' investors who bid up the price of IPO shares beyond true value, or that issuers suffer from behavioral biases causing them to put insufficient pressure on the underwriting banks to have underpricing reduced. These theoretical models almost always end with the conclusion that the average IPO is undervalued at the offer price, where the initial investors, in most cases, benefit from possessing information by receiving allocations of shares in IPOs and earn the largest first-day returns. The expectations of issuing firms, investors and underwriters in IPO pricing are considerably different. In an offering, the issuer generally wants to receive the highest possible price to maximize cash flows to the firm. Investors like to purchase shares at a deep discount so that they can realize positive returns *Baskent University, Faculty of Economics and Administrative Sciences, Baglica Kampusu, Ankara, 06530, Turkey, + 90 (312) 234 10 10 /1728, gurayk@baskent.edu.tr 1 Ritter (1998), (2003) provides an update on the compilation of Loughran, Ritter, and Rydqvist (1994)

in a short investment period. Underwriters, acting as an intermediary between investors and issuing firms, suffer from a dilemma, if an underwriter determines IPO prices too low, where the foreseen amount of money left on the table will be huge, the issuing firm may withdraw or switch to another underwriter. On the other hand, if an underwriter determines IPO prices relatively high, investors will hesitate to buy new issues, which would result in low commissions and an unwanted effort in aftermarket stabilization activities. Underwriters, however, have an incentive to underprice the shares to ensure that they can sell the offering, and, unsurprisingly, there is extensive evidence that IPOs are, on average, underpriced. Hence, pricing of stocks in IPOs may be the most critical stage of the IPO process. More recently, the literature on IPOs, both theoretical and empirical, focuses on the efficiency mechanisms of the following methods for pricing initial public offerings.

At the center of this literature, book building, auctions and fixed price offers differ mainly in price-discovery and share allocation process.

  1. Book Building - in which the underwriters do road shows and take non-binding orders from
  2. Auctions - in which the company sets a price range to be used as a non-restrictive guideline for investors, than accepts bids, each specifying a number of shares and a price the investor is willing to pay for them, finally, the market-clearing price set by the investors approximates the real price the shares will command in the market.
  3. Fixed Price Offer - in which the issue price is set first and than orders are taken from investors who typically pay in advance for part or all of the shares that are ordered.
  4. Sale through the Stock Exchange - in which the sale is initially conducted in the primary market of the stock exchange by a designated underwriter. Those investors who buy the shares in the primary market must wait until the shares trade in the secondary market in order to sell their shares. The price designated at the time of registration with the securities exchange commissions is set as the opening price.
  5. Hybrid Offerings - in which the underwriters combine the preceding IPO methods, and design auction/fixed price, auction/book building and book building/fixed price hybrids. For most hybrids, the most common combination is the book building/fixed price offer, where the underwriter uses the book building method to set the price and allocate shares to institutional and foreign investors, and retain the fixed price offer to the domestic retail investors who do not participate in the price-setting process.

This paper 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). In this regard, we first compare the three IPO methods available in Turkey. 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. Then, we estimate a binary probit on the issuer's choice between fixed price offer and sale through the stock exchange method, however, because of the declining importance of the book building mechanism in Turkey, we excluded the book build IPO sample from our binary probit estimations. Finally, we determine the factors that are expected to have an effect on the IPO returns. Our results indicate that, the comparison of the two mechanisms yield that for certain values, namely first day underpricing, IPO amount and fractions of equity sold, fixed price offer outperforms the sale through the stock exchange method. To the best of our knowledge, this is the first empirical study on the comparison between fixed price offer and sale through the stock exchange method in the IPO literature. The uniqueness of the data and the availability of the sale through the stock exchange method in the ISE make it possible to conduct a study on the comparison between these two methods.

The remaining part of this paper is organized in six sections. In the next section, we provide a comparison of the theoretical and empirical research conducted on IPO methods across many countries around the world. In section 3, we describe the three important Turkish IPO market selling procedures. In section 4, we describe the data and the methodology we used in our empirical tests. Section 5 documents the relationship between market conditions and underpricing of IPOs in different time series and the last section concludes.

2. Comparison of the IPO methods in the literature: Theory and Evidence

The efficiency of the IPO methods has been the subject of an academic research over a decade, both empirical studies and theoretical models have tried to explain the advantages of one method over another. The argument that is often made in favor of IPO methods is often empirical as well as theoretical. Researchers studying on the efficiency of the IPO methods try to answer the most challenging question, “Which one of the IPO mechanism is the most efficient?"2. However, according to our comprehensive literature research, both empirical studies and theoretical models listed in Table

  1. have some mixed answers.
  2. 1. Book Building vs. Fixed Price Offer and/or Auctions Comparison of the IPO methods in the literature goes back to Benveniste and Spindt3 (1988), (1989)

and Spatt and Srivastava (1991), they suggest that the American bookbuilding procedure is efficient since it encourages investors to reveal their beliefs about the issue's value at a cost of initial underpricing. Book building allows investors to collect information about the value of the stock and price the issue more accurately. To compensate the investors who reveal information, underwriter will favor them when allocating shares. However, fixed price mechanism does not utilize any information about realized buyer valuationsin setting the issue price and is generally inefficient. Loughran, Ritter and Rydqvist (1994) present the first international evidence on the short-run and long-run performance of companies going public in many stock markets around the world. They document that the fixed price method is associated with greater underpricing because of the greater probability of the issue failing and the increased uncertainty associated with the longer time delay between offer and issuance time.

Chowdhry and Sherman (1996) point out that two features of fixed price offers tend to lead to greater underpricing, relative to the book building method. The first one is the length of the bidding process, as the time gap between the offer and first day market price widens “price information leakage" occurs, the second one is the common requirement that investors pay in advance for their entire order. Benveniste and Busaba (1997), extend Welch's4 (1992), model of information cascades in investment decisions and present a theoretical comparison of the fixed-price and book-building mechanisms. They exhibit that issuers with a greater concern for risk will prefer a fixed-price offer, because book- building might generate higher expected proceeds, and exclusively provides an opportunity to sell additional shares at full value but it also exposes them to higher risk. Ritter (1998) demonstrates that countries that use bookbuilding typically have less underpricing than countries using fixed-price offerings, more underpricing under fixed-price offering procedures can be attributed to informational cascades. However, Loughran, Ritter and Rydqvist (1994) and Ritter (1998) point out that IPOs with discretionary allocation (Fixed Price Offering and Book-building) are underpriced more than those with non-discretionary allocation (Offer for sale and Auctions), especially in Auctions. Under discretionary allocation, the first day price increase averaged 37% in fixed price offerings, 12% in book building. Under non-discretionary allocation, the first day price increase averaged of 27% in Offer for sale and 9% in Auctions. 2 In terms of controlling the amout and volatility of underpricing, share allocation and pricing. 3 The literature on underpricing in initial public offerings goes back to Logue (1973), Ibbotson (1975), Chalk and Peavy (1987), Miller and Reilly (1987), Ritter (1984), Rock (1986), Allen and Faulhaber (1989), Benveniste and Spindt (1988, 1989), Grinblatt and Hwang (1989), and Welch (1989). However, the mechanism by which initial issues are sold has largely been ignored until Benveniste and Spindt (1988), (1989). 4 Welch (1992) focuses on the fixed-price procedure used in some non-US countries, and shows that this procedure can cause informational cascades: investors who observe the investment choice made by previous investors can update their beliefs about the value of the issued shares.

Sherman (2000), (2002) shows that fixed price offer, can lead to higher underpricing than book building. Contrary to the fixed price offer and the auction method, in book building underwriters discriminate investors in the allocation of shares to establish long-run relationship with intermediates. Book building gives the underwriter greater flexibility in designing a solution that reflects the individual issuer's preferences. By controlling investor access to IPO shares, book building controls both the winner's curse problem that affects discriminatory auctions and the free rider problem that affects uniform price auctions. Book building also reduces uncertainty for both issuers and investors. In a study that covers 47 countries, Sherman (2002) has found that in all countries in which the bookbuilding mechanism has been introduced, pre-existing auction systems have decreased in popularity or disappeared altogether.

Ljungqvist, Jenkinson and Wilhelm (2000) use a unique dataset containing information on 2,051 initial public offerings in 61 stock markets around the world, during the period of 1992-1999. The authors examine the relative direct and indirect costs of offerings carried out by book building and fixed-price methods. They find that, the direct costs of book building are typically twice as large as direct costs for fixed-price offers. Compared to fixed price offerings, book building efforts - though more expensive - produce far less underpricing. Nevertheless, fixed price offering is still an extremely common method that is not likely to be abandoned by the underwriters completely. Compared to book building efforts, fixed price offering is an efficient, low cost way to distribute shares to retail investors, avoiding the high fixed costs of road shows. Aorsio, Giudici and Paleari (2000), Guidici and Paleari (2001) present an empirical study conducted on the Milan Stock Exchange companies between 1985 and 1999. Authors distinguish between fixed- price offers and open-price offers with bookbuilding and find different underpricing levels and statistically significant determinants. They state that if the offering is preceded by book building, the underpricing is significantly lower (8.32 % vs. 28.33% in fixed-price offerings), this method allows the issuing parties to collect information from the institutions and to signal good news or bad news to retailers through the revision of the prospectus price range. Therefore, the cost of raising private information is reduced and the requested underpricing is lower. The evolution of the placing procedure, from fixed price to book building, has considerably improved the efficiency of Italian IPO market.

Biais and Faugeron-Crouzet (2002) analyze and compare the performance of book building, fixed price offering, uniform price auction, internet-based Open IPO mechanism, and an auction like mechanism called the Mise en Vente in France. Conclusions emerging from their analysis are; Fixed price offerings lead to inefficient pricing and winner's curse. Dutch auctions can also lead to inefficiencies, to the extent that they are conductive to tacit collusions by investors. The book building and an auction like mechanism Mise en Vente can lead to optimal information elicitation and price discovery.

Chahine (2002) investigates the relationship between underpricing and the investors' interest prior to and after the IPO day on 305 French issues. Empirical results show that book-built issues have a lower underpricing, on median, but a higher variance level, than the auction-like and fixed-price offerings. Despite the high initial underpricing of some book-built issues, book-building procedure appears to better control the information gathering from investors participating in the offering, and to be a more efficient pricing system than the auction-like procedure. Paney (2004) examines the initial returns, characteristics of issuers and long run performance of Indian IPOs on a sample of 84 Indian IPOs between 1999 and 2002. In terms of initial returns or underpricing, Paney (2004) finds that fixed price offering yields higher initial returns on average, as compared to book building. In terms of issuer characteristics, Paney (2004) finds that fixed price offering are used by issuers offering large proportion of their capital by raising a small amount of money. In contrast, book building is opted for by issuers, offering small portion of their stocks and mobilizing larger sums of money.

Kutsuna and Smith (2004) present an empirical study conducted on the Japanese IPOs between 1995 and 1999. Using a sample of 163 book-built and 321 auctioned IPOs by JASDAQ companies, authors document that average total issue cost, measured as a percent of aftermarket price, was significantly higher in the book-building regime than in the earlier auction regime. However, when results are weighted by issue size, the estimated aggregate costs of auctioning and book building are similar. This outcome favors book building over auctions for two reasons. First, auction-method estimates do not reflect opportunity costs related to underinvestment. Second, issue cost estimates ignore other benefits of the more-accurate pricing that book building affords. Anand (2005) examines the differences between book building and Dutch-auction, and shows that the bookbuilding method of offering securities is superior to the Dutch-auction IPOs. Stated by Anand (2005), while the Dutch-auction may seem to lead to efficient price discovery based on investor demand, recent transactions suggest that price discovery is not always accurate and that, indeed, underpricing occurs even in the Dutch auction. Further, even if the Dutch auction is more fair than the bookbuilt process in terms of allocating securities, the Dutch auction can lead to less capital market efficiency overall and can therefore be questioned as a basis for promoting this type of offering. Jagannathan and Sherman's (2005) research on the efficiency of IPO mechanism show that hybrid bookbuildings5, unlike auctions, have proved effective in many different countries, cultures, time periods, and market conditions. Jagannathan and Sherman (2005) propose a new IPO mechanism that could overcome the problems with standard auctions. A method that retains the advantages of bookbuilding, while modifying it to increase transparency. Although not a direct comparison between book building, auctions and fixed-price offers, Cornelli and Goldreich6 (2001), (2003) examine a unique data set of international book building allocations and find that the underwriter favors both regular investors and investors that supply information on the value of the issue. Degeorge, Derrien and Womack (2005) have presented empirical evidence from France's IPO market that underwriters employing the bookbuilding process implicitly committed to providing more favorable coverage to the companies they took public in the aftermarket. Authors find convincing empirical evidence that in addition to placing the IPO shares with investors, underwriters employing book-building implicitly commit to providing more favorable coverage to the companies they take public in the aftermarket. Specifically, analysts, affiliated with the lead underwriter of the offering, issue more favorable recommendations for recent book-built IPOs than for auctioned offerings.

2.2. Fixed Price Offer vs. Book Building and/or Auctions

The pricing of Initial Public Offerings (IPOs) in the short-run has been analyzed by several theoretical and empirical studies referring to the major international stock markets. Extensive research has revealed that the fixed-price offering all over the world suffer from IPO underpricing especially in these major markets. However, studies conducted by Busaba and Cheng (2001), Bierbaum and Grimm (2003), Chemmanur and Liu (2003), Hsu and Hung (2005) present some evidence on the efficiency of fixed price offering over book building and auctions. Busaba and Cheng (2001) show that the bookbuilding process elicits much information from informed traders at the IPO stage by promising larger allocation of valuable stocks to investors who truthfully reveal their information, and therefore reduce the impact that such informed traders have in the after- 5 In the hybrid bookbuilding offers, all retail investors are allowed to place orders in a public offer tranche, and all have an equal chance of getting shares. The prices, however, are set by professional investors who are given incentives to attend the road show.

Jenkinson and Jones (2004), cast some doubts about the findings of Cornelli and Goldreich (2001) upon the extent of information production during the bookbuilding period.

market trading. In contrast, the fixed price method, that does not elicits such private information at the IPO stage, enables informed traders to use such information in the after-market at the expense of the uninformed traders. In this regard, if the underwriter building a book can not successfully target a subset of the informed investors, a simple fixed price strategy that involves allocating the issue to retail investors produces higher proceeds on average. The comparatively high adverse selection problems associated with the fixed-price method will spill over from the IPO stage to the after-market. This in turn means that liquidity will be relatively more important for IPOs carried out via a fixed- price method than via bookbuilding. Authors show that, compared with a fixed-price offering, the bookbuilding process elicits more information from informed traders at the IPO stage, and therefore reduce adverse selection problems in the after-market trading. However, by the same token bookbuilding may require larger informational rents to be paid at the IPO stage. This suggests that underpricing should be larger for IPOs carried out via bookbuilding than via a fixed-price method. Bierbaum and Grimm (2003) compare the fixed price and the uniform auction in a game theoretic framework. The comparison of the two mechanisms yields that for certain parameter values, namely a low variance of the asset and, at the same time, a sufficiently high probability of low demand, fixed price method outperforms the auction in terms of revenue. Moreover, the revenue in the fixed price mechanism is typically less volatile than the revenue in the auction. Chemmanur and Liu (2003) model the effect of costly information production on issuers' choice of a fixed-price offer or a uniform-price auction with exogenous entry of bidders. Their model predicts that IPO auctions will exhibit a significantly lower mean and variance of underpricing compared to fixed- price offerings. This is due to the fact that the offering price in an IPO auction aggregates the information produced by outsiders to a significant degree, so that this offering price is greater for higher intrinsic-value firms and lower for lower intrinsic-value firms in IPO auctions than in fixed- price offerings. At the same time, there is less information production in IPO auctions compared to fixed-price offerings where the offering price is set by insiders to induce the optimal degree of information production, so that a lower amount of information is reflected in the opening price of the shares listed in the stock market. Thus, Chemmanur and Liu (2003) demonstrated that, in many situations, firms will prefer to go public using fixed-price offerings rather than IPO auctions in equilibrium, since such offerings allow the firm to induce the optimal extent of information production.

Hsu and Hung (2005) present an empirical study conducted on the Taiwanese companies between 1996 and 2000. Using a sample of 280 pure fixed-price offers and 84 hybrid auctioned, authors find that, Taiwanese hybrid auctions are associated with less under-pricing and with a lower variance of under-pricing than versus the pure fixed-price offers, but these differences are not statistically different. On the other hand, we find that the market index returns prior to the IPO pricing date have a strong influence on the under-pricing of Taiwanese IPO auctions and of the pure fixed-price offers. Authors provide empirical evidence of how Taiwanese issuers make the choice of IPO method. Taiwanese issuers that float large IPOs, or which have a pricing conflict with underwriters, will likely use a hybrid auction to distribute shares. On the other hand, when the relative risk level of IPO auctions to fixed-price offers has increased, the issuers will likely avoid an IPO auction. Empirical evidence also explains why Taiwanese IPO auctions have lost market share to fixed-price offers. Further results reveal that Taiwanese IPO auctions are not associated with less under-pricing and with a lower variance of under-pricing, nor are they better at incorporating recent market information into the IPO price than the pure fixed-price offers. Authors' examination on issuers' choice of hybrid auctions or fixed-price offers indicates that Taiwanese issuers condition their choice of IPO method not only on firm characteristics, but also on IPO size and on market conditions. This is why Taiwanese issuers prefer a pure fixed-price offer to a hybrid auction are based on market volatility and the pricing conflict. In doing so, under a volatile market where Taiwanese hybrid auctions have become much riskier relative to the pure fixed-price offers, issuers will prefer a pure fixed-price offer to a hybrid auction, resulting in a lower popularity of Taiwanese hybrid auctions.

As listed in Table 1, Fixed Price Offering seems to be the less favorable method comparing to Book building and Auction Methods. It is a fact that, the worldwide introduction of book building method during the 90's has promoted efficiency in the major equity markets. However, Sherman (2002) states that stock markets listing few IPOs each year, fixed price offering is still be the optimal method. 2.3. Auctions vs. Book Building and/or Fixed Price Offer Using a sample of 108 French firms marketed on the Second Marché between 1984 and 1991, Leleux and Paliard (1995) show that initial returns are significantly higher for firms issuing through the fixed- price procedure than for firms using auction-like procedures. Leleux and Paliard (1995) state that the auction mechanism is associated with less underpricing and thus more efficient, since this procedure is able to incorporate more information from recent market momentum into the pricing of the IPO. Beierlein (2000) compares the book-building method to two commonly used auction mechanisms, the discriminatory price auction and the uniform price auction in terms of underpricing and the long run performance of IPOs relative to the market. Using data from Japan, Israel and the U.S., author finds evidence that the U.S. book building is less efficient than the auction mechanisms are. Specifically, underpricing is significantly higher in the U.S. than it is in Japan or Israel and bookbuilding appears to incorporate less demand information into the offer price than the auction mechanisms do. Bennouri and Falconieri (2001) suggest that auction mechanisms are the optimal way to sell new shares because auction procedures are more informationally efficient than bookbuilding. Assuming ex ante uncertainty about the firm true value, then auction mechanisms are able to elicit and incorporate more information from the market as well as from investors into the pricing of IPOs.

Draho (2001) suggests that underpricing in bookbuilt IPOs is due to the uncertainty about the price on the secondary market rather than about the firm value, as most of the literature assume. Nonetheless, his results indicate auction-like mechanisms as the most efficient ones, since they are open to all investors who are moreover required to submit price-quantity bids. McDonald (2001) examines the efficiency mechanisms of the sealed-bid uniform-price auctions over book building method in a theoretical framework and concludes that the uniform-price auction, due to its generalized Vickrey auction properties, is indeed an efficient auction mechanism especially for the sale of IPOs over the Internet. Biais, Bossaert, and Rochet, (2002) study the optimal IPO mechanism by which the seller can extract private information to maximize the expected net IPO proceeds. They find that the optimal mechanism they characterize is similar to auction-like IPO procedures used in the U.K. and in France. Kaneko and Pettway (2003) examine the Japanese initial returns before and after the introduction of book building, and find that underpricing in book building method is significantly higher than auctions, especially during hot markets. Results suggest that the move from auction-priced to underwriter-priced IPOs using book building in Japan has significantly reduced the wealth of issuing companies while increasing the wealth of underwriter-selected investors. Derrien and Womack (2003), use the French IPO data for the 1992-1998 period and compare the three underwriting/selling mechanisms available on the French market. One is very similar to the book building mechanism used in the United States. Another is a fixed price procedure. The third one is an auction-like procedure. Authors show that the auction procedure is better than the others at controlling underpricing in general as well as the variance of underpricing of the issued shares in “hot" versus “cold" markets. Fixed price offering method is indeed inefficient and leads to greater underpricing compared to IPOs sold through book-building and auctions. However, the main empirical comparison in this paper is between the two main procedures auction and book building. Authors find evidence that during hot markets auctioning is associated with less underpricing than book building. They attribute the result to the auction method's ability to incorporate more information about recent market performance into the offer price. This result provides empirical support for the theoretical work by Biais, Bossaerts, and Rochet (2002) who suggests the auction procedure is optimal.

In line with the evidence of Derrien and Womack (2003) that an auction procedure is more efficient in incorporating recent market momentum in the offer price compared to fixed price procedure, Vandemaele (2003) uses the French IPO data for the 1984-1995 period and points out the factors that may influence issue procedure choice. Results indicate that, firms facing relatively high valuation uncertainty are high likely to opt for an auction-like procedure and the likelihood of opting for an auction increases as the investment bank reputation associated with the issue decreases.

Although not a direct comparison between auctions, book building and fixed-price offers, studies in Pettway and Kaneko's (1996) examination on Japanese auctions, Kandel, Sarig and Wohl's, (1999) examination on Israeli auctions, and Liu, Wei and Liaw's (2003) examination on Taiwanese auctions seem to suggest that IPO auctions lead to less under-pricing. Biais and Faugeron-Crouzet (2001) show that a uniform price auction can prevent tacit collusion among bidders and can truthfully elicit information from investors in much the same way as book building. Bulow and Klemperer (1998) also show that it can be optimal in an auction to set a price at which there is excess demand. 2.4. Research on IPOs in the Istanbul Stock Exchange (ISE) Firms in Turkey may offer their shares to the public through, book building, fixed price offer and sale through the stock exchange method, however, they are mainly underwritten and sold using the fixed- price offering method; a method which is very common world wide is becoming much less common, particularly for more active markets. Recent empirical studies, focused mainly on the initial returns and under pricing, conducted by Ozer (1999), Kiymaz (2000) and Durukan (2002), show that underwriters of the Turkish companies listed in the Istanbul Stock Exchange do not fully incorporate all available information into the IPO offer price. The first day returns of IPOs average approximately 12.41%-13.10%-14.61% respectively, indicating that systematic underpricing largely observed in the ISE. Further, Ozer (1999) finds that IPOs provide significant excess returns in the first three days following the offer. Abnormal returns are the highest on the first day, decrease in the second and the third event days and approximate the market movement after the third day of trading. Kiymaz (2000) finds that these initial returns are related to the size of issuer, rising stock market between the date of public offering and first trading day, institutional ownership and self-issued offering. Durukan (2002) finds that these initial returns are related to the size of issuer, gross proceeds, age of firm, debt level in the firm capital structure in the year prior to IPO, institutional ownership and self-issued offering. 3. Methods of Sale in the Istanbul Stock Exchange (ISE) The Turkish IPO market gives issuers and underwriters a choice of three different IPO selling mechanisms. Accordingly, firms may offer their shares to the public through one of the following three methods of sale.

3. 1. Fixed-Price Offering

In a fixed price offering method, a fixed price at which the securities are offered is known in advance by the investors. Investors fill in and sign the bid forms and submit them to the underwriter during the period designated in the prospectus. When the bid forms are submitted, investors must deposit the amount corresponding to the demanded shares to the underwriters' account. At the end of the bid collection period, the allocation of shares among investors is conducted through the pro-rata basis. Where the total amount of shares offered to the public is divided by the number of investors until the entire amount of shares has been allocated. After the bid collection process, the underwriter submits to the issuer a list of the allocation of shares among investors within two business days after the end of the bid collection period. The isuer then approves the allocation list and returns it to the underwriter within two business days. The lists of bids that are not met are announced by the underwriter and the corresponding amounts are returned immediately to investors. Also, the shares pertaining to bids that have been met are released to investors.

3.2. Book Building Method

In a book building method, price at which securities will be offered is not known in advance to the investor. Only an indicative price range is known and bids above this minimum price are collected. As in the fixed price method, investors submit the bid forms and deposit the corresponding amount to the underwriters' account. At the end of the bid collection period, the allocation of shares is conducted as follows. Starting from the highest-price bid, the bids are transformed into a table showing cumulative bid amounts at each price level. The price level at which the cumulative amount exceeds the amount of shares offered is set as the selling price. All bids above that price are met. After the bid collection process, the underwriter submits to the issuer a list of the allocation of shares among investors within two business days after the end of the bid collection period. The issuer then approves the allocation list and returns it to the underwriter within two business days. The lists of bids that are not met are announced by the underwriter and the corresponding amounts are returned immediately to investors. Also, the shares pertaining to bids that have been met are released to investors.

3.3. Sale through the Stock Exchange

The initial public offering of an issue using Sale through the Stock Exchange method can be carried out to the Istanbul Stock Exchange within the regulatory framework of the stock exchange after the approval of the Capital Markets Board of Turkey (CMBT). The sale is first conducted in the primary market of the ISE by a designated intermediary institution. Those investors who buy the shares in the primary market must wait until the shares trade in the secondary market in order to sell their shares. The sale can be made through the secondary market of the ISE after fulfilling all the required documentation at least 20 days prior to the offering. The price selected at the time of registration with the CMBT is set as the opening price. From then on, the price of the shares moves within the band determined by the daily limits (+/- 21 %) set by the ISE.

4. Data and methodology

4.1. Data

The sample we analyze is the widest ever examined to carry out a research on IPOs in Turkey. We obtain firm data and issue data from the Istanbul Stock Exchange. Firm data includes book value of assets, book value of equity, book value of debt, sales revenue, net profit, firm age. Issue data include the offer data, number of shares issued, number of days between pricing and first trade, amount raised, offer price, first aftermarket price, and other offering details. Our sample consists of 217 IPOs from January 1993 through October 2005. Panel A of Table 2 reports the number of IPOs and gross proceeds in Turkish market during this period. Of these 217 IPOs, 149 firms use fixed price offers (69%), 39 firms use Sale through the Stock Exchange (18%), 29 firms use book building (13%) to distribute their shares. The number of IPOs during this period peaked in 2000 than hit the bottom in 2001. As a consequence of the 2001 crisis7, the number of IPOs between 2001- 2003 totaled only 7. Nevertheless, the book building mechanism which was often used in mid 90's has lost issuers interest in the most recent years, there were 25 total book building offerings in 1994-1995, 7 The Turkish economy has suffered from an acute liquidity crisis in 2001. The exchange rate as measured by Turkish Lira/USD depreciated by 50% in a short period of time.

but there have only been 4 since the beginning of 19968. Thus, our main comparisons in this paper are between the Fixed Price Offer and Sale through the Stock Exchange method. Panel B of Table 2 reports some descriptive statistics of Turkish IPOs from 1993 through 2005 on firms specific and market related characteristics. In line with the evidence on initial returns documented by Ozer (1999), Kiymaz (2000) and Durukan (2002), IPOs in the sample also suffer from a significant underpricing on an average. The average initial returns computed on the basis of first trading day closing are positive and significant for all the three types of IPO methods, high levels of underpricing observed in sale through the stock exchange with a mean of 15.68%, low levels of underpricing observed in fixed price offers with a mean of 7.01%. The average initial returns computed on the basis of first month closing price are still positive for fixed price offers with a mean of 15.61% and sale through the stock exchange offers with a mean of 58.83%, however, book building offers lead their investors into a systematic under-performance at the same time period. Panel B also shows that the fractions of equity offered to the public is higher in the sale through the stock exchange method with a mean of 35.32%, than the fixed price offer (25.66%) and book building (24.28%) methods. Firm specific variables such as Net Sales, Total Assets, Total Equity, and Total Debt show that larger firms are more likely use the fixed price offers and smaller firms use sale through the stock exchange and book building methods. 4.2. Choice of IPO method In this part of the study, we try to answer whether “issue related characteristics", “issuing firm characteristics" and “market related characteristics" have a stronger influence on issuers' choice of IPO method. As discussed and confirmed by Derrien and Womack (2003) in French IPOs, and Hsu and Hung (2005) in Taiwanese IPOs, previous market conditions prior to IPOs and some firm characteristics at the time of the IPO have significant impacts on the level and the variability of initial underpricing. To answer the same question in Turkish IPOs, we follow Hsu and Hung's (2005) methodology and first estimate a binary probit.