Pricing Initial Public Offerings on New European Stock Markets

Transcription

1 Pricing Initial Public Offerings on New European Stock Markets Giancarlo Giudici a,* and Peter Roosenboom b,* Abstract. In 2000 more companies listed in Europe than in the US. This is mainly due to the success of new stock markets designed for high-growth companies. In this paper we analyse a sample of 482 Initial Public Offerings (IPOs), listed on five new European stock markets up to March We investigate the determinants of their first-day returns. We find a mean first-day return (underpricing) equal to %. Market returns, the IPO firm s risk and price revisions in the premarket are positively related to first-day returns, whereas IPO deal flow shows is inversely related to underpricing. We also show that the Internet euphoria impacts first-day returns. JEL Classification: G30, G32 Keywords: Initial Public Offerings, Underpricing, New European Stock Markets a Corresponding author, Università degli Studi di Bergamo - Dipartimento di Ingegneria Via Marconi, Dalmine (BG) Italy Ph Fax giudici@unibg.it b Tilburg University and CentER - Department of Finance P.O. Box LE Tilburg - The Netherlands Ph Fax p.g.j.roosenboom@kub.nl * Giancarlo Giudici acknowledges funding from CNR and from Cofinanziamento MURST. We thank Andrea Randone, Mark Koevoets and Willem Schramade for their valuable research assistance. 1

2 1. Introduction In 2000 more companies listed on European stock exchanges than in the United States. From January to December the NYSE and the NASDAQ hosted 445 IPOs, while in Europe over 500 companies went public. The increased interest in going public is largely due to the success of new European stock markets designed for high-growth and high-tech fledgling companies. These new markets have recently been established around Europe: the French Nouveau Marché (1996), the German Neuer Markt (1997), Euro.NM Belgium (1997), the Dutch NMAX (1997) and the Italian Nuovo Mercato (1999). These new stock markets meet the expectations of the European Commission that has long been eager to copy the success of financing high-tech and new economy enterprises in the United States (European Commission, 1996). Such well-developed stock markets dedicated to high-growth firms are believed to be instrumental in creating economic growth, technological innovation and jobs. Subrahmanyam and Titman (1999) provide theoretical support for this belief and show how jump-starting an economy s stock market can indeed improve economic efficiency. Vibrant stock markets may also increase the amount of venture capital funding available to finance young and start-up companies in an economy (Jeng and Wells, 2000; Black and Gilson, 1998). New European stock markets are therefore of interest to both economists and policymakers. In this study we focus on the pricing of IPOs on five new European stock markets. It is widely documented that companies that go public sell their shares at an offer price that is lower than the first-day closing market price. This positive first-day offer-to-close return is commonly known as underpricing. Several recent studies have investigated the underpricing phenomenon for Initial Public Offerings (IPOs) on the main European stock exchanges. Ljungqvist (1997) reports an average underpricing equal to +9.2% for German IPOs from 1970 to Derrien 2

3 and Womack (2000) document +9.5% underpricing in France from 1992 to Giudici and Paleari (2001) find that 164 Italian IPOs from 1985 to 2000 are underpriced by +23.9%, on balance. These numbers pale in comparison to the average first-day return of +38.1% of IPOs on new European stock markets we report in this study. This prompts the question what explains the significant underpricing of IPOs on new European stock markets. The existing literature suggests several explanations. Companies that go public on the new stock markets tend to operate in risky entrepreneurial sectors, most notably the Internet. The information asymmetry between the issuing firm and investors is large. As a result, pricing the shares of these firms becomes more difficult and investors require underpricing to compensate them for the higher valuation uncertainty. Alternatively, underpricing may serve to compensate investors for their release of private information during the bookbuilding process. The book building process starts with the publication of a preliminary prospectus containing a price range for the shares. The investment bank solicits non-binding bids from institutional investors in the premarket. After gauging investors demand in the premarket phase, the final offer price is determined. In order to reward investors for their information, underwriters only partially incorporate collected private information in the final offer price. As investors pay a lower offer price than the full information price in the secondary market, they pocket high first-day returns (Benveniste and Spindt, 1989). In this study we analyse the pricing of 482 Initial Public Offerings (IPOs) on the French Nouveau Marché, the German Neuer Markt, the Euro.NM Belgium, the Dutch NMAX and the Italian Nuovo Mercato from March 1996 to March We study the determinants of first-day 1 These five new stock markets were members of the Euro-NM network. However, the Euro-NM initiative failed to establish operating links between the member markets. The network was disbanded on 31 December 2000, leaving full autonomy to the single national markets. 3

4 returns and focus on the IPO price revisions in the premarket. These price revisions are measured as the percentage revision of the offer price from the midpoint of the file price range. We find that favourable private information released by institutional investors during the premarket phase of the bookbuilding procedure (as reflected in a greater price revision) is positively related to underpricing. We also find that first-day returns are correlated with public information available at the listing (i.e. the market returns), the IPO firm s risk and the IPO deal flow (i.e. the number of IPOs brought to the market from 60 before to 10 trading days after the IPO date). Consistent with the Internet hype documented in the United States, we report that Internet IPOs are more underpriced than non-internet IPOs. Interestingly, we document marketspecific determinants of first-day returns. In section two, we continue with an overview of the extant literature. In section three we describe the sample and summary statistics. In section four we provide the empirical results and in section five we present our conclusions. 2. Review of the literature Pricing IPOs is a difficult task: the market is not certain about the quality of the IPO firm, while the issuing firm and its underwriter do not know the market demand for IPO shares. The disclosure of information is crucial in order to avoid mispricing. The problem facing an underwriter wanting to collect information useful to pricing an IPO is that investors have no reason to truthfully reveal their private information during the premarket phase. Benveniste and Spindt (1989) show that, in order to induce investors to truthfully reveal their demand for IPO shares, they must be rewarded with more underpricing on deals for which there is strong demand. At the outset of the bookbuilding process, the underwriter proposes a 4

5 price range for the shares. Investors express their interest for IPO shares by placing non-binding orders at different prices within the price range. When investors place their orders, they factor in their expectations about what the market price of the stock will be on the first day of trading. From these indications of interest, the underwriter can therefore learn positive and negative information that can be used when setting the final offer price of the IPO. In order to compensate investors for revealing their private information, the final offer price only partially incorporates collected private information. On average, investors earn a high first-day return by paying a lower price for the stock than the full information price in the secondary trading market. An important prediction of the Benveniste and Spindt model is that underpricing is related to the level of interest in the premarket phase. This suggests that IPOs priced in the upper part of the price range are more likely to be heavily underpriced. Consistent with the model of Benveniste and Spindt, several U.S. studies report a correlation between the percentage revision of the final offer price from the midpoint of the price range and first day returns. Hanley (1993), Lowry and Schwert (2001) and Loughran and Ritter (2001) report that U.S. IPOs where the final offer price is revised upwards display higher first-day returns than those IPOs where the final offer price is revised downwards. If the private information of investors was entirely incorporated into the final offer price, then no relation between the price revision and underpricing should have been found. Loughran and Ritter (2001) show that first-day returns on IPOs are also predictable based upon market returns in the three weeks prior to the IPO date. The quantitative effect is large: each 1 percent increase in the market during the three weeks before the IPO results in a first-day return which is 1.3 percent higher. This finding is at odds with the Benveniste and Spindt model. The model does not predict there should be partial adjustment to public information, such as market returns. Instead public information should be fully incorporated into the final offer price. 5

6 Lowry and Schwert (2001), on the other hand, show that public information (proxied by market returns) is fully incorporated into the offer price, whereas private information is only partially incorporated. Their results suggest that negative information learned during the premarket phase is more fully incorporated into the offer price than positive information. The reason is that both underwriters and investors do not want to incur losses on overpriced issues (i.e. IPOs with negative first-day returns). In addition, Lowry and Schwert (2001) find that high technology firms and riskier firms tend to have higher first-day returns. The higher first-day returns compensate investors for the greater valuation uncertainty associated with these deals. Ljungqvist and Wilhelm (2001) investigate the pricing of both U.S. and international IPOs. In line with other studies, they report that price revisions explain underpricing. However, in contrast to Lowry and Schwert (2001), they do not find that underwriters are including negative information more fully than positive information in an international context. Another finding is that IPO deal flow has a negative effect on underpricing. A higher (expected) deal flow affords underwriters the opportunity to exclude investors from other, profitable deals as retaliation for distorting their private information. This increases the market power of underwriters and reduces the need to compensate investors for revealing their private information (Benveniste, Busaba and Wilhelm, 2001). The effect is large in economic magnitude since a one standard deviation increase in IPO deal flow reduces underpricing from 22 to 17.2 percent. 3. Data and Sample Description We consider a sample of 482 IPOs on five European stock markets for high-growth firms (the French Nouveau Marché, the German Neuer Markt, the Euro.NM Belgium, the Dutch NMAX 6

7 and the Italian Nuovo Mercato) from March 1996 to March This number excludes transfers from other stock markets, financial companies, spin-offs, and firms incorporated in a country other than France, Germany, Belgium, the Netherlands or Italy. Table 1 compares the listing requirements in the stock markets that we consider in our analysis. [Please insert Table 1 about here] The minimum book value of the equity is generally equal to 1.5 million euro. The Euro.NM Belgium and the Neuer Markt require a minimum age equal to three years, while one year old companies may list on the other markets. Companies with losses, but with an ambitious business plan and relevant growth opportunities qualify for listing on all new stock markets. The capital sold to the public must represent 20%/25% of the total equity, albeit in some cases exceptions are tolerated. All markets require that at least 50% of the IPO shares must be newly issued. This should boost IPO firms to make new investments and grow. Lock-up provisions, that prevent insiders from selling their shares immediately after the IPO, have to be implemented in most markets. Table 1 highlights that the listing requirements imposed by the exchanges we analyze are similar, allowing us to draw a comparison between the single national exchanges. The sample distribution, by stock market and by industry, is reported in Table 2. More than 50% of the companies we analyze went public on the German Neuer Markt and about 30% listed on the French Nouveau Marché. [Please insert Table 2 about here] 7

8 About 8%, 3% and 2% listed on the Italian, Dutch and Belgian stock market respectively. The Internet (115 firms) and computer software and services (114 firms) account for 47.5% of the sample. Other industries such as electronics (10.8%), biotechnology (8.1%), business services (7.9%) and media (7.1%) are heavily represented as well. The industry distribution confirms that an overwhelming majority of companies going public on new European stock markets operates in high-tech industries. Table 3 reports summary statistics about the sample. We collect data from IPO prospectuses. Market prices are taken from Datastream. [Please insert Table 3 about here] It is interesting to note that the mean market capitalization and offer size are significantly larger than the book value of total assets. The average company has a market capitalization that is more than 120 times its operating cash flow. Over 25% of the companies have no (or negative) operating cash flow. The IPO thus represents an important source of capital, allowing the company to fund its future growth plans. The mean company age is equal to 12 years. This differs from previous studies of IPOs on European main exchanges (e.g., Pagano, Panetta and Zingales, 1996), that show European IPO firms to be mature and established companies. The mean initial underpricing is equal to %, which is remarkably high if compared to first-day returns on the main stock exchanges in Europe during the same period, as reported by Ljungqvist (1997), Derrien and Womack (2000) and Giudici and Paleari (2001). Yet, more than 25% of the sample IPOs does not display underpricing, or are initially overpriced (i.e. experience a negative first-day return). We plot the monthly number of IPOs and the mean initial underpricing per month in Figure 1. We observe hot issue periods, namely April 1998 to September 1998, as well as 8

9 February 1999 to May 1999, and December 1999 to April These hot issue periods are characterised by unusually high underpricing, followed by high IPO volumes periods, in which the mean initial return is lower (see for example April 2000 to August 2000). [Please insert Figure 1 about here] Almost all IPOs (466; 96.7%) have been priced using the book building procedure, while in 13 cases the final offer price was fixed in the prospectus and in 3 cases the IPO was auctioned. Considering 464 IPOs in which the final offer price is not fixed, we find that on the average the final offer price is revised upwards (+3.573%) with respect to the midpoint price of the file range 2. Yet, no revision or downward revision with respect to the midpoint price is found in about 25% of the IPOs. Table 4 investigates any cross-market differences. We record several marked differences among the three major markets (Nouveau Marché, Neuer Markt and Nuovo Mercato). [Please insert Table 4 about here] The German Neuer Markt leads the other markets with regard to offer size, IPO price volatility, underpricing, and percentage price revision. The French Nouveau Marché hosts smaller and less underpriced IPOs. The Italian Nuovo Mercato heads the other markets with regard to market capitalization and operating cash flow. No differences are found concerning company age. In the 2 We could not identify the bookbuilding range for two IPOs reducing the bookbuilding sample from 466 to 464 firms. 9

10 next section we will investigate the effect of both cross-country and firm-specific characteristics on the first-day returns. 4. Empirical Results In this section we study the determinants of the first-day returns of IPOs on new European stock markets. In comparison to companies going public on the main European stock exchanges, IPOs on new stock markets are younger and smaller and operate in high-tech and fast-growing innovative sectors. Therefore, information asymmetries should be particularly large and it becomes more difficult to price the shares of these firms. Moreover, it is interesting to investigate whether market-specific determinants are at work. We investigate whether the market index return in the 50 days prior to the IPO influences first-day returns. We measure the index return by the MSCI Index of the country of listing (France, Germany, the Netherlands, Belgium or Italy). As argued earlier, the Benveniste and Spindt (1989) model does not predict that there is partial adjustment to public information, such as market returns. Instead public information should be fully incorporated into the final offer price. According to theory, market returns should therefore not able to explain first-day returns. We expect that investors require higher first-day returns to compensate them for the higher valuation uncertainty associated with risky deals. Market-adjusted volatility is used as a proxy for company risk. It is computed as the standard deviation of the firm s daily stock returns 10

11 during the 60 trading day interval of 20 to 80 days after its IPO date minus the standard deviation of daily returns to the MSCI index of the country of listing during that same period 3. We examine whether IPO deal flow affects first-day returns. Several studies (Hoffmann- Burchardi, 2001; Loughran, Ritter and Rydqvist, 1994) show that IPO markets are subject to cycles, with hot issue periods, that are characterised by high volume of IPOs and/or high firstday returns, alternating with cold issue periods. The transition from hot to cold periods is marked by heavy volume periods in which the number of IPOs is increasing, but the typical firstday return is lower. Figure 1 reinforces the occurrence of IPO cycles during our sample period. Similar to Ljungqvist and Wilhelm (2001), we measure IPO deal flow by the number of IPOs from 60 trading days before to 10 days after the IPO date. We introduce a dummy variable for Internet IPOs. We expect Internet-related firms to be more underpriced, due to the Internet euphoria documented by Cooper, Dimitrov and Rau (2001) in the United States. In accordance with the model of Benveniste and Spindt (1989) we hypothesise that the revision of the offer price is positively related to underpricing. Benveniste and Spindt (1989) argue that, in order to reward investors for their private information, underwriters only partially incorporate collected private information in the final offer price. The more valuable the private information is to pricing the IPO, reflected in a greater price revision, the higher first-day returns. We measure the percentage revision of the final offer price from the midpoint of the price range. We also include four control variables in the regression models. We include the log of the book value of total assets to control for differences in size. We include company age to control 3 Although our risk measure is not available at the time of the IPO itself, investors clearly form opinions about the firm s risk when pricing IPOs. Following Lowry and Schwert (2001), we argue that our ex post risk meaure is an unbiased estimator of the ex ante risk of the IPO firm. 11

12 for age differences. To control for differences between the national exchanges, we incorporate two dummy variables for Neuer Markt and Nouveau Marché, respectively. Table 5 reports the results of the regression analyses. [Please insert Table 5 about here] We first consider the empirical results for the full sample of 482 IPOs as shown in the first column of Table 5. The market index returns during the 50 days prior to the IPO is significantly related to first-day returns. A one percentage point increase in the market results (e.g. from 4% to 5%) results in a first-day return which is 2.4 percent higher. This suggests that underwriters do not fully adjust final offer prices to recent market returns. Analysing a sample of U.S. IPOs, Loughran and Ritter (2001) find that the market return during the three weeks before the IPO date displays a positive and significant relation with underpricing. Their and our results are inconsistent with the prediction of Benveniste and Spindt (1989) that public information should be fully incorporated into the final offer price. Market-adjusted volatility is positively correlated with underpricing. This suggests that investors demand higher underpricing as compensation for the valuation uncertainty associated with risky IPOs. A one standard-deviation increase in volatility increases the first-day return by 5.9 percent. This corroborates the findings of Lowry and Schwert (2001) for the United States. They report that riskier firms have higher first-day returns. The IPO deal flow is inversely related to underpricing. This effect is large in economic magnitude. A one standard deviation increase in IPO deal flow reduces the first-day return by 39.3 percent. One explanation for this finding may be that during periods of high IPO volume, the market power of underwriters increases (Benveniste, Busaba and Wilhelm, 2001). Investors are eager to buy shares in IPOs, reducing the need to compensate them for revealing their private 12

13 information by high first-day returns. Alternatively, IPO deal flow may proxy for the transition from hot to cold issue markets. As argued earlier, this transition is characterised by an increasing number of IPOs coupled with lower first-day returns. Other things equal, Internet IPOs have first-day returns that are 27.4 percent higher than the first-day returns of non-internet IPOs. This corresponds to the results of Lowry and Schwert (2001) that show that high-technology firms in the United States have higher first-day returns than IPOs from other industries. Interestingly, cross-market differences in first-day returns do appear, as the coefficient of market dummies are significantly different from zero. Consistent with Table 4, IPOs on the German Neuer Markt are more severely underpriced, while IPOs on the French Nouveau Marché are less underpriced. In the second column of Table 5 we consider 464 IPOs for which price revisions are available. We find that the price revision is significantly related to underpricing. This is consistent with the theoretical model of Benveniste and Spindt (1989) and the empirical results of Loughran and Ritter (2001), Lowry and Schwert (2001) and Ljungqvist and Wilhelm (2001). A 10 percentage point increase in price revision results in a first-day return that is about 9 percentage points higher. This suggests that investors are rewarded by higher first-day returns in exchange for their revelation of positive private information in the bookbuilding process. We run separate regressions for the three major exchanges to investigate whether the determinants of underpricing are different across markets. The market momentum is informative only in Germany, while the IPO valuation uncertainty (proxied by the volatility) is an important determinant of underpricing in France. These differences may be related to different institutional settings. For example, in France in recent cases the final offer price is set at the listing and must be in the initial price range. If the final offer price is not in the range, orders are cancelled and a new subscription period opens with a new range. This is not the case of Germany, allowing the first-day returns to be higher on the Neuer Markt. Because of these institutional differences, 13

14 French underwriters are better able to incorporate public information (as proxied by the market returns) into the offer price than German underwriters are. This may explain why the market returns in France are not significantly related to first-day returns in France. A one percentage point increase in market return, increases the underpricing of German IPOs by almost 2.9%, whereas in France it increases first-day returns by an insignificant 0.3 percent. Surprisingly, we do not find significant determinants in Italy. We attribute this result to the immature nature of the Nuovo Mercato. Italian IPOs are more recent, and seem to be more influenced by the market turbulence after the NASDAQ fall in Several Italian IPOs that listed after April 2000 had a downward revision of their offer price, which was appreciated by the market as a signal of prudence, and not interpreted as a negative feedback from book building activity 4. This may explain why the correlation between the first-day returns and the revision of the offer price is not significant in Italy. In additional tests we investigate whether underwriters respond differently to negative and positive information learned during the premarket phase of the bookbuilding procedure. Lowry and Schwert (2001) argue that negative information learned during the premarket is more fully incorporated into the final offer price than positive information since investors and underwriters want to avoid losses on overpriced issues (i.e. they want to avoid buying issues with negative first-day returns). We re-estimate the regressions with an additional explanatory variable that equals the price revision if it is positive and zero otherwise. The coefficient of this variable should be significantly positive if underwriters price in positive information less fully than negative information. However, in unreported tests we do not find any evidence that 4 For example, in June 2000 the Italian company CHL offered shares at 30 euro, against a minimum file price of 45 euro. Yet, the first-day price has been euro (+181.7%). The same phenomenon has been observed in other four Italian IPOs. No similar cases are found for the other stock markets. 14

15 underwriters incorporate positive information less fully than negative information. The coefficient equals (t-statistic = 0.793). 5. Conclusions This study contributes to literature on the pricing of IPOs, which is thus far dominated by U.S. studies. We verify that, in contrast to the evidence of main European exchanges, new markets attract young and high technology firms. We show that an average IPO on the new markets displays a % first-day return, about three times larger than the first-day return of a typical IPO on the main European stock exchanges. We find that several determinants explain the high first-day returns of IPOs on new stock markets. First, in contrast to the Benveniste and Spindt (1989) model, public information is not fully incorporated into the final offer price. This yields a positive association between market returns (our proxy for public information) and first-day returns. Second, we find that the IPO firm s risk is positively related to underpricing. Investors seem to demand higher first-day returns to compensate them for the higher valuation uncertainty associated with risky deals. Third, the IPO deal flow is inversely related to first-day returns. Companies that go public in a period of high IPO volume have lower first-day returns. In high IPO volume periods, the market power of underwriters may be stronger, reducing the need to compensate investors for the release of private information by means of high first-day returns. Alternatively, this result may capture the transition of a hot to a cold issue market, that is normally associated with an increase in the number of firms going public and a decline in the average first-day returns. Fourth, we find that consistent with the Internet euphoria in the United States, Internet IPOs are more underpriced than non-internet IPOs. Fifth, the revision of the offer price with respect to the price range published in the prospectus is informative. The more optimistic the 15

16 revision of the offer price, the higher first-day returns. The higher first-day returns serve to reward investors for releasing their private information about IPO value in the premarket phase of the bookbuilding process. Finally, we document cross-market differences exist in the determinants of first-day returns. These differences may be attributed to alternative institutional settings and to different market maturity. 16

17 References Benveniste, L.M., W. Busaba and W.J. Wilhelm, 2001, Information externalities and the role of underwriters in primary equity markets, forthcoming in Journal of Financial Intermediation. Benveniste, L.M. and P.A. Spindt, 1989, How investment bankers determine the offer price and allocation of new issues, Journal of Financial Economics 24, Black, B. and R. Gilson, 1998, Venture capital and the structure of capital markets: banks versus stock markets, Journal of Financial Economics 47, 3, Cooper, M., O. Dimitrov and P.R. Rau, 2001, A rose.com by any other name, forthcoming in Journal of Finance. Derrien, F. and K.L. Womack, 2000, Auction vs. book-building and the control of underpricing in hot IPO markets, working paper, Tuck School of Business and Dartmouth College. European Commission, 1996, New technology-based firms in Europe, EIMS publication, 31. Giudici, G. and S. Paleari, 2001, What drives the initial market performance of Italian IPOs? An empirical investigation of underpricing and price support, working paper, Università di Bergamo. Hanley, K.W., 1993, The underpricing of initial public offerings and the partial adjustment phenomenon, Journal of Financial Economics 34, Hoffmann-Burchardi, U., 2001, Clustering of initial public offerings, information revelation and underpricing, European Economic Review 45, Jeng, L.A. and P.C. Wells, 2000, The determinants of venture capital funding: evidence across countries, Journal of Corporate Finance 6, 3, Ljungqvist, A.P., 1997, Pricing initial public offerings: further evidence from Germany, European Economic Review 41, Ljungqvist, A.P. and W.J. Wilhelm, 2001, IPO allocations: discriminatory or discretionary?, working paper, New York University and Boston College. Loughran, T., J.R. Ritter and K. Rydqvist, K., 1994, Initial public offerings: international insights, Pacific-Basin Finance Journal 2, Loughran, T. and J.R. Ritter, 2001, Why don t issuers get upset about leaving money on the table in IPOs?, forthcoming in Review of Financial Studies. Lowry, M. and G.W. Schwert, 2001, Biases in the IPO pricing process, working paper, University of Rochester. 17

18 Pagano, M., F. Panetta and L. Zingales, 1996, The stock market as a source of capital: some lessons from initial public offerings in Italy, European Economic Review 40, Subrahmanyam, A. and S. Titman, 1999, The going-public decision and the development of financial markets, Journal of Finance 54, White, H. S., 1980, A heteroscedastic-consistent covariance matrix estimator and a direct test of heteroscedasticity, Econometrica 48,

19 Table 1: Listing requirements Stock Market Country Company age and size Issue size and IPO rules Free float Lock-up provisions Neuer Markt Germany Three years; equity book value higher than 1.5 million euro Nuovo Mercato Italy One year; equity book value higher than 1.5 million euro Nouveau Marché France Equity book value higher than (Euronext) 1.5 million euro NMAX The Equity book value higher than (Euronext) Netherlands 1.5 million euro Euro.NM Belgium Belgium Three years; market (Euronext) capitalization higher than 2 million euro Half of the offered shares must be newly issued; IPO proceeds higher than 5 million euro Half of the offered shares must be newly issued; IPO proceeds higher than 5 million euro Half of the offered shares must be newly issued; IPO proceeds higher than 5 million euro Half of the offered shares must be newly issued; IPO proceeds higher than 5 million euro Half of the offered shares must be newly issued; IPO proceeds higher than 5 million euro 25% or 10% if offer size is Insiders must lock larger than 5 million euro their shares for at (at least 100,000 shares) least 6 months 20% (at least 100,000 One year (insiders voting shares) must lock at least 80% of their shares) 20% (at least 100,000 One year (insiders voting shares) must lock at least 80% of their shares) 20% (at least 100,000 Discretionary voting shares) 25% (in some cases 10%) Not available 19

20 Table 2: Sample distribution Neuer Markt Nouveau Marché NuovoMercato NMAX Euro-NM Belgium Total Internet a (23.9%) Computer software & services b (23.6%) Electronics & instruments c (10.8%) Biotech & medical instruments d (8.1%) Business services e (7.9%) Media f (7.1%) Manufacturing g (6.0%) Retail & Wholesale h (5.6%) Telecommunications i (3.7%) Other (3.3%) Total 278 (57.7%) 144 (29.8%) 37 (7.7%) 14 (2.9%) 9 (1.9%) 482 (100.0%) a b c d e f g h i Internet companies are identified after careful reading of each firm s business description as published in the prospectus Non-internet companies assigned SIC codes starting with 737 (computer programming, data processing and other computer related services) Non-internet companies assigned SIC codes starting with 357 (computer and office equipment), 36 (electronic and other electrical equipment) or 38 (not 384, measuring, analyzing and controlling instruments) Non-internet companies assigned SIC codes starting with 283 (drugs), 384 (surgical, medical and dental instruments and supplies), 80 (health services) or assigned SIC code 8731 (commercial physical and biological research) Non-internet companies assigned SIC codes starting with 73 (business services, not 737), 87 (engineering, accounting, research and management services) or 89 (services, not elsewhere classified) Non-internet companies assigned SIC codes starting with 27 (printing and publishing), 78 (motion pictures), 79 (amusement and recreation services) Non-internet companies assigned SIC codes starting with 20 (food and kindred products), 23 (apparel and textile products), 24 (lumber and wood products), 28 (chemicals and allied products, not 283), 30 (rubber and plastics products), 32 (stone, glass, and concrete products), 33 (primary metal industries), 34 (fabricated metal products), 35 (industrial and commercial machinery, not 357) or 37 (transportation equipment manufacturing) Non-internet companies assigned SIC codes starting with 5 (wholesale and retial trade) Non-internet companies assigned SIC codes starting with 48 (communications) 20

21 Table 3: Summary statistics Percentiles Average Min 25 th 50 th 75 th Max Standard deviation Market capitalization ( 000 euro) a 279,441 9,054 58, , ,240 13,252, ,682 Offer size ( 000 euro) b 51,434 2,641 14,862 27,030 49,788 2,700, ,737 Total assets ( 000 euro) 26, ,003 10,261 21,860 2,311, ,467 Operating cash flow ( 000 euro) c 2,185-24, ,208 3,052 53,497 5,851 Company age (years) Underpricing (%) d Price revision (%) e Volatility (%) f Market return previous 50 days (%) g IPO deal flow (#) h a Number of shares outstanding after IPO times the closing market price on the first trading day b Number of shares sold in IPO times the final offer price c Earnings before interest, taxes, depreciation and amortization during the most recent financial year, as disclosed in the prospectus d Initial offer-to-close return measured as: (first-day closing market price - final offer price)/final offer price e Price revision is measured as: (final offer price midpoint of price range)/midpoint of price range. Price revisions are available for 464 firms. f Standard deviation of the firm s daily stock returns during the 60 trading day interval of 20 to 80 days after its IPO date minus the standard deviation of daily returns to the MSCI index for the country of listing (Germany, France, Italy, Belgium or The Netherlands) during the same period g Return of the MSCI index of the country of listing (Germany, France, Italy, Belgium or The Netherlands) over the 50 trading days preceding the IPO date h Number of companies going public on New European stock markets from 60 trading days before to 10 trading days after the IPO date 21

22 Table 4: Differences across new European stock markets Means across markets Test for difference Neuer Markt Nouveau Marché Nuovo Mercato Neuer Markt Versus Nouveau Marché Neuer Markt versus Nuovo Mercato Nouveau Marché versus Nuovo Mercato t-test z-test a t-test z-test t-test z-test Market capitalization ( 000 euro) 343,767 96, , *** *** * * *** *** Offer size ( 000 euro) 62,368 11,046 44, *** *** * ** *** *** Total assets ( 000 euro) 33,135 13,877 32, ** * *** *** Operating cash flow ( 000 euro) 2,255 1,764 4, * ** 3.19 *** ** Company age (years) Underpricing (%) *** *** ** *** Price revision (%) *** *** ** ** Volatility (%) *** *** ** *** Market return previous 50 days (%) *** *** *** 3.66 *** IPO deal flow (#) *** *** *** ** *** *** a Nonparametric Wilcoxon test 22

23 Table 5: Determinants of underpricing on new European stock markets a Full sample Bookbuilding sample b Neuer Markt sample c Nouveau Marché sample d Nuovo Mercato sample e Market returns previous 50 days (6.669) *** (6.09) *** (5.852) *** (0.974) (1.033) Volatility (2.069) ** (2.224) ** (0.774) (3.301) *** (-0.616) Log(IPO deal flow) (-4.334) *** (-3.973) *** (-3.638) *** (-1.679) * (-0.987) Internet dummy (3.342) *** (3.078) *** (1.793) * (2.562) ** (0.908) Price revision (2.751) *** (5.77) *** (3.935) *** (-0.491) Control variables Log(total assets) (-0.056) (-0.018) (-0.767) (0.896) (-0.776) Company age (1.379) (1.304) (1.143) (-0.671) (-0.755) Neuer Markt dummy (3.569) *** (2.221) ** Nouveau Marché dummy (-2.253) ** (-3.06) *** Intercept (1.954) * (1.989) ** (2.947) *** (-0.921) (1.676) R 2 adjusted F-statistic *** *** *** *** a Cross-sectional regressions are shown, estimated by ordinary least squares with the correction for heteroskedasticity, developed by White (1980). The dependent variable is underpricing. t-statistics are shown in parentheses below the estimated coefficients. * statistically significant at the 10 percent level (two-tailed), ** statistically significant at the 5 percent level (two-tailed) and *** statistically significant at the 1 percent level (two-tailed) b 464 sample firms c 277 sample firms d 140 sample firms e 35 sample firms 23

24 % 225% 200% 175% 150% 125% 100% 75% 50% 25% 0% -25% Nov-00 Jan-01 Mar Jul-00 Sep-00 Mar-97 May-97 Jul-97 Sep-97 Nov-97 Jan-98 Mar-98 May-98 Jul-98 Sep-98 Nov-98 Jan-99 Mar-99 May-99 Jul-99 Sep-99 Nov-99 Jan-00 Mar-00 May-00 Nov-96 Jan-97 Jul-96 Sep-96 Mar-96 May-96 Number of IPOs per month Average underpricing per month Figure 1: Number of IPOs and average underpricing per month Number of IPOs Underpricing