Structuring the IPO: Empirical evidence on the portions of primary and secondary shares

Structuring the IPO: Empirical evidence on the portions of primary and secondary shares

Structuring the IPO: Empirical evidence on the portions of primary and secondary shares by Nancy Huyghebaert Cynthia Van Hulle Katholieke Universiteit Leuven, Belgium* * The authors are, respectively, Assistant Professor, and Full Professor of Finance, Department of Applied Economics, Katholieke Universiteit Leuven, Belgium. The authors thank Vladimir Atanasov, Piet Sercu, Linda Van de Gucht, participants at EFA 2002, Berlin and FMA European Conference 2002, Copenhagen and seminar participants at K.U. Leuven for useful comments on an earlier draft of this paper.

Corresponding author: Nancy Huyghebaert, Department of Applied Economics, Katholieke Universiteit Leuven, Naamsestraat 69, 3000 Leuven, Belgium; tel.: 32-16-32 67 37, fax: 32-16-32 67 32, e-mail:

Structuring the IPO: Empirical evidence on the portions of primary and secondary shares Abstract We empirically study the determinants of the portions of primary and secondary shares offered in IPOs. The data show that young growth firms tend to issue primary shares. Limited internal cash generation and a debt mix that largely consists of bank loans significantly positively affect the size of the primary portion. The data are also consistent with the notion that if financing needs warrant a relatively small primary portion, companies may add secondary shares to the offering to increase the offering size, which enhances post-IPO market liquidity.

Furthermore, these growth firms are also more likely to issue seasoned equity in the aftermarket. Mature firms tend to issue only secondary shares. The diversification motive does not drive the size of the secondary portion but adverse selection costs have an impact. Also, firms including only secondary shares show relatively higher control turnover ex post.

JEL classification: G32, G24 Keywords: IPO structure; primary and secondary shares; motives for going public; liquidity; turnover

1. Introduction The source of the shares sold in an initial public offering is an important but little studied feature in the IPO literature. Nevertheless, this decision has important implications, both for the company and for the initial shareholders, and hence is likely to contain information on the motives for going public. Specifically, when primary (i.e. newly created) shares are offered, the IPO selling proceeds accrue to the firm.

Conversely, original owners are entitled to the proceeds from selling secondary shares (i.e. shares existing before the IPO). While Jegadeesh et al. (1993), Spiess and Pettway (1997) and others use the nature of the shares that are sold in the IPO as an explanatory variable in their research, we are not aware of any study that explicitly examines the portions of primary and secondary shares in IPOs. A potential explanation could be that in the U.S., IPOs traditionally include a notable primary portion whereas pure secondary offerings are scarcely observed. In contrast, in Continental Europe purely primary and secondary offerings have been popular over time, as well as offerings that combine primary and secondary shares.

In this paper, we examine what factors determine the size of the portions of primary and secondary shares relative to the number of shares outstanding pre-IPO. We also study the nature of trade-off between these two fractions and link it to the motives for going public. Finally, we complement our results with ex post information on firm behavior. Pagano et al. (1998) consider the question of why firms go public using data on Italian companies. From comparing publicly listed and privately held firms during 1982-1992, they conclude that the likelihood of an IPO is positively related to the firm’s size and the market-to-book ratio prevailing in the industry.

Also, they find that the limited amount of new equity raised is used to reduce leverage rather than to finance growth. Furthermore, once introduced on the exchange, IPO firms show an abnormal reduction in profitability and a high turnover of the controlling shareholder(s). Similar results are obtained by Rydqvist and Högholm (1995) using Swedish data for the period 1970-1991, and by Goergen (1998) using data on UK and German firms that went public during 1970-1988. Overall, these findings offer a rather pessimistic view on the role played by 3

European stock exchanges; they suggest that the stock market is not used as a mechanism to finance growth, but rather as a way for owners to reduce their firm’s risk – what Pagano et al. (1998) have called a rebalancing of financial structure – and to cut back on their involvement in the company. However, in Continental Europe several changes have taken place since the sample period(s) covered by these earlier studies. Cornelli and Goldreich (1999) and Sherman (2000), for instance, document the global trend towards using the bookbuilding method for selling shares in IPOs, which reduces asymmetric information problems.

Also, in the second half of the nineties, new markets, such as EASDAQ and the alliance of European growth markets Euro.NM, were set up to meet the needs of an increasing number of young and high-growth companies. Simultaneously, consolidation in more traditional sectors enlarged the financing needs of the more established firms. From these observations, it is clear that other motives for going public may have become important. Using a sample of 95 Belgian IPOs over the period 1984-2000, we find that younger and smaller firms with substantial growth opportunities and limited internal cash generation issue a larger fraction of primary shares.

Whereas the overall debt ratio is not significantly related to the size of the primary portion, the fraction of bank debt has a positive impact. Given the decision to go public, market conditions, as measured by the stock market return and issue activity in the pre-IPO year, do not seem to affect the portion of primary shares. Overall, these results indicate that the need for additional financing is the main force determining the size of the primary portion. Furthermore, the evidence suggests that the creation of a liquid market – which is an important concern when the firm wants to tap the stock market in the future –induces firms with growth opportunities and eminent internal cash generation to add secondary shares to a relatively small primary portion to achieve a sufficiently large free float.

By contrast, mature firms with high internal cash generation tend to offering only secondary shares. In determining the size of the secondary portion, the worry to avoid adverse selection costs under the form of underpricing proves to be important. Simultaneously, given the decision to go public, the evidence indicates that windows of opportunity may also have an impact. By contrast, diversification does not seem to be an important motive when owners fix the 4

size of the secondary portion. Furthermore, post-IPO data on market liquidity and seasoned equity offerings support the notion that at IPO-time, growth firms wish to establish a liquid market with a view to realizing seasoned offerings in the future. Data on control turnover indicate that firms issuing only secondary shares in their IPO are more likely to be taken over in the years following their going public operation. Overall, while the relative importance of different offering types varies over time, we find that the underlying relations between firm characteristics and deal structure are largely stable during the sampling period.

The remainder of the paper is organized as follows. Section 2 surveys the main reasons of why companies go public, and the ensuing preference for issuing primary and/or secondary shares. Section 3 describes the sample of 95 Belgian IPOs over the period 1984-2000. Section 4 analyzes the determinants of the portions of primary and secondary shares and investigates the nature of trade- off between primary and secondary shares. Section 5 provides a link between deal structure and post-IPO market liquidity, seasoned equity offerings and control turnover. Section 6 concludes the paper.


Theory and predictions As indicated by Pagano et al. (1998), the decision to go public is too complex to be captured by a single model. Therefore, likewise that paper, we start from the main theories on why companies go public and infer their predictions for the portions of primary (= new shares offered relative to the number of shares outstanding pre-IPO) and secondary shares (= fraction of existing shares divested). The predictions from these models, as summarized in table 1 and discussed below, will allow us to derive empirically what forces drive the size of the primary and secondary portion.

This will give us additional insight in the interaction between motives for going public and observed choices regarding deal structure.


2.1. The costs of going public 2.1.1. Adverse selection costs Adverse selection costs, which occur when (some) investors are less informed than the issuers about the true value of the firms going public, induce underpricing. Not surprisingly, several authors (e.g., Ritter, 1984; Megginson and Weiss, 1991; Chemmanur and Fulghieri, 1999) find these costs to be more serious for young (AGE = logarithm of the firm’s age at IPO-time) and small (SIZE = logarithm of total assets) companies as the latter typically have a limited track record and low visibility.

Considering the case of primary shares first, we expect adverse selection problems to induce firms to offer a smaller portion of primary shares.

For within the firm’s set of investment opportunities, fewer projects become attractive as the price at which shares can be marketed decreases. As this effect holds particularly true for young and small companies, adverse selection is likely to induce a positive relation between the age and size of the firm and the primary portion. Furthermore, ownership structure may interact with adverse selection. Specifically, if high adverse selection costs are to be borne by a few shareholders, one could argue that these owners may prefer to limit the primary portion at IPO-time and issue less than needed.

Then, the firm could issue seasoned equity later on, when more information has become publicly available and adverse selection costs have decreased. However, next to the direct costs of an extra capital increase later on, such action may result in forgone investment opportunities for the company and/or waiting costs. If these costs are important, ownership structure is not likely to affect the size of the primary portion. Nevertheless, to take the possible impact of ownership concentration into account, we add a variable measuring the ownership percentage of block holders owning at least 5% of the shares (= CONC).

Furthermore, to control for the fact that the absolute amount of adverse selection costs may increase with the size of the firm, we also include an interaction term between ownership concentration and firm size.


Conditional upon listing, adverse selection costs may also induce a positive relation between firm age and size and the size of the secondary portion: if adverse selection decreases the price at which securities can be marketed, old owners will be less inclined to part from their shares, ceteris paribus. Likewise the primary portion, ownership structure and adverse selection may also interact. Specifically, Gomes (2000), Habib and Ljungqvist (2001) and others argue that when underpricing imposes considerable costs, large owners may try to maximize their overall proceeds from divesting by limiting the secondary portion at the IPO; afterwards, as more information becomes publicly available, they may continue to sell shares gradually.

As this policy does not hamper the firm’s investment opportunities, we expect this reduction in the size of the secondary portion to be less costly as compared to the reduction in the size of the primary portion. To measure the impact of ownership structure, the same variables as defined above for the primary portion are used. 2.1.2. Other costs Other costs of going public – such as the administrative expenses and fees and the loss of confidentiality – may affect the likelihood of going public, but these are not directly related to the size of the primary or secondary portion. For example, the existence of fixed listing costs reduces the attractiveness of an IPO for small firms (e.g., Ritter, 1987), but, given the decision to go public, we expect no relation with the portions of primary or secondary shares.

Similarly, the information that has to be disclosed upon listing does not depend upon the fractions of primary or secondary shares that are offered to the public. Given the diversity of these costs and the difficulty of measuring them, we do not define variables to test this hypothesis. 7

2.2. The benefits of going public 2.2.1. Overcoming financing constraints When firms go public, they obtain access to an additional source of financing (e.g., Pagano et al., 1998; Allen and Gale, 1999; among others). Especially companies with sizeable financing needs and a largely used debt capacity may value this benefit. Financing needs will be important when firms have valuable growth prospects. High-growth firms often are financially constrained as entrepreneurs have limited personal wealth or dislike investing more of their own resources in the firm. Simultaneously, when there is a sizeable time lag between investments in new projects and cash generation, debt may not be the most suited source of financing.

For these reasons, we expect growth opportunities to be positively related to the size of the primary portion. The firm’s market-to-book ratio (= MARKET/BOOK) is an often-used indicator of growth opportunities. Although it has the advantage that it includes market values and therefore is forward looking, it has the disadvantage that it may also reflect windows of opportunity embedded in too optimistic perceptions by the market. Therefore, in accordance with the literature, we use AGE and SIZE as additional indicators of growth prospects. We expect AGE and SIZE to be negatively related to the size of the primary portion.

Highly levered firms may have fully used their borrowing capacity. According to Myers and Majluf’s (1984) pecking order theory, these firms will raise equity to meet their financing needs. Pagano et al. (1998) indeed find that firms go public to rebalance their financial structure. We therefore expect a positive relation between leverage (LEVERAGE = debt to total assets) and the size of the primary portion. Conversely, firms that generate substantial cash internally have a smaller demand for external financing at IPO-time, ceteris paribus. As a result, we expect a negative relation between internal cash generation (ROA = EBITDA to total assets) and the size of the primary issue.

The wish to overcome financing constraints is unlikely to influence the size of the secondary portion.


2.2.2. Greater bargaining power vis-a-vis banks Rajan (1992) argues that banks with private information about borrowers can extract rents at the time loans have to be extended or rolled over. By going public and disseminating information to the public, a company elicits outside competition to its bank and ensures a lower cost of credit and/or a larger supply of external finance. Pagano et al. (1998) indeed document that firms experience a decrease in the cost of credit after their IPO. We therefore expect that firms whose debt largely consists of bank loans (DEBT MIX = bank debt to total debt) will include a larger primary portion in their offering to reduce bank bargaining power.

The wish to reduce bank bargaining power, however, is unlikely to impact on the size of the secondary portion. 2.2.3. Monitoring Another benefit of being publicly quoted is the managerial disciplining provided by the stock market (Homström and Tirole, 1993; Stoughton et al., 2001). While the danger of a hostile takeover frequently is absent in Belgium – since firms usually introduce less than half of their shares on the stock market – it is still possible to expose managerial decisions to the market’s assessment. This information could then be used to develop managerial compensation schemes.

Again, this motive may affect the likelihood of going public, but we expect no direct relation with the size of the portions of primary and secondary shares.

2.2.4. Windows of opportunity The decision to go public may be affected by perceived windows of opportunity. Specifically, in addition to filling true financing or divestment needs, firms may also try to capitalize on the optimistic perceptions in the market at IPO-time (e.g., De Long et al., 1990; Ritter, 1991; Rajan and Servaes, 1997). 9

During periods of high pre-IPO market returns (MARKET RETURN = the return on the Belgian All Shares Index (BASI) in the year preceding the IPO), firms may increase the size of the primary portion beyond their financing needs.

However this action also entails costs. Next to increasing problems of adverse selection, free cash flow theory shows that building financial slack negatively affects the stock price. Models by Easterbrook (1984) and Jensen (1986), for instance, stress that the need to regularly tap the stock market imposes discipline on firms and improves stock prices. This argument likely holds true especially for young and small firms without a proven record and, in general, for firms that have accumulated substantial financial slack. Another aspect of the use of windows of opportunity is the feedback effect.

In particular, because of the positive market assessment, companies may revise their investment plans upward and correspondingly increase the size of the primary portion. This market feedback effect, however, may also be captured by the firm’s MARKET/BOOK ratio, since, as indicated before, this measure includes company specific expectations. Simultaneously, the clustering of IPOs (VOLUME = the number of IPOs in the preceding year scaled by the total number of IPOs in the sample) may induce information spillovers and hence lower information problems (e.g., Booth and Chua, 1996; Bayless and Chaplinsky, 1996; Hoffmann, 2001).

Reduced information asymmetries allow to market shares at a lower cost and hence may increase the number of attractive projects within the investment opportunity set. As a result, VOLUME is likely to affect positively the size of the primary portion. If capturing windows of opportunity is important in shaping the size of the secondary portion, we expect this variable to be positively related to MARKET RETURN, and to the extent it also reflects windows of opportunity, the firm’s MARKET/BOOK ratio. Also, the reduced information asymmetries during periods of high IPO activity (VOLUME) would allow firms to market more secondary shares at lower cost.


2.2.5. Investor recognition/liquidity In a capital asset pricing model with incomplete information, Merton (1987) shows that stock prices are higher the larger the number of investors aware of the company’s securities. Kadlec and McConnell (1994) find support for this argument: when companies listed elsewhere announce their decision to also list in New York, their stock on average yields a positive abnormal return. This motive mainly affects the likelihood of going public. In the same way, one could also argue that having a sufficiently large number of quoted shares is important, because then more investors (e.g., institutional investors) become interested in the firm’s securities.

This latter argument is closely related to the notion that market liquidity may reduce the cost of capital (e.g., Benveniste and Wilhelm, 1990; Szewczyk et al., 1992; Booth and Chua, 1996; Maug, 1998; Sherman, 2000). Furthermore, this idea has also been supported empirically (e.g., Eckbo and Norli, 2000). Young and small growth firms are likely to benefit most from investor recognition. The reason is that because of their growth prospects, these firms presumably will use the stock exchange for financing their investment projects not only at IPO-time, but also at later dates. It is clear that increasing the size of the primary portion could add to investor recognition.

However, as discussed in section 2.2.4. above, issuing primary shares beyond financing needs entails additional costs of adverse selection and creates free cash flow problems, especially for young and small firms. Alternatively, these companies could add secondary shares to the primary portion to increase liquidity. As the latter action avoids free cash flow problems, it may be a better solution. This is likely to hold true especially if at IPO-time financing requirements are too small to create a sufficiently liquid market, for instance when the company benefits from internal cash generation.

Preceding arguments imply that if investor recognition is important, one would likely observe it in ex post data. Specifically, if it can be achieved by increasing the total size of the offering, we should observe that a measure for the total size of the issue at IPO-time is positively related to liquidity in the aftermarket. Also, firms that go public because of financing needs should be more likely to return to the market with a seasoned equity offering.


The positive price impact of investor recognition may also benefit old owners who wish to cash in on their shares. However, increasing the secondary portion at IPO-time implies incurring the cost of underpricing. As discussed in section 2.1.1., a better policy may be to limit the size of the secondary portion and sell afterwards additional shares as more information has become publicly available. Overall, we do not expect any clear-cut relation between this motive and the size of the secondary portion.

2.2.6. Portfolio diversification Owners may use the IPO as a mechanism to divest shares and diversify their portfolio (Pagano, 1993; Stoughton and Zechner, 1998).

If so, one would expect the IPO to include a secondary portion. Although selling shares to value enhancing block holders in a private sale rather than an IPO could drive up the transaction price, the remaining owners may dislike the monitoring by these new owners (e.g., Brennan and Franks, 1997). Also, Pagano and Roëll (1998) argue that large block holders may even over-monitor the firm, which is not the case when selling to more dispersed shareholders in an IPO. Simultaneously, the liquidity discount applicable to the equity of a private company can be avoided when selling secondary shares in an IPO (e.g., Brau et al., 2000).

If diversification of owner wealth is an important motive in the decision to go public, owners of high-risk firms are likely to divest a sizeable secondary portion. Meulbroek (2000), for instance, finds that in highly volatile Internet-based firms, many owner-managers sell a portion of their shares, or even exercise their stock options with the objective to sell the acquired shares and diversify their portfolio. For young, small firms with considerable growth opportunities, high leverage or limited internal cash generation, the owners will feel the need to diversify to a larger extent, ceteris paribus.

The diversification motive does not imply a need for primary shares and therefore is unlikely to impact on the size of the primary portion.


2.2.7. Change of control Preparing for a turnover in corporate control may be another motivation for selling secondary shares at IPO-time. Zingales (1995) points out that going public can maximize the proceeds from a later sale of the firm. By selling off a minority stake to a widely dispersed shareholder base, owners may be able to increase the surplus they can extract from the future buyer. Mello and Parsons (1998), Pagano and Roëll (1998) and Roëll (1996) similarly argue that an IPO can be used to improve the efficiency of the ownership structure.

Empirical evidence for other European countries indeed indicates increased control turnover activity in the years following the IPO (e.g., Pagano et al. (1998) for Italy, Högholm and Rydqvist (1997) for Sweden). If the transfer of control – whether immediate or remote – is an important motive in the decision to go public, we expect the more established (i.e. larger, older) firms to go public through selling secondary shares, as in these companies the firm- specific investment of the initial owners is less essential. Although this motive has a positive impact on the probability of an IPO through a secondary offering, the impact on the size of the secondary portion is not a priori clear, other than that the original owners are likely to limit its size to keep control and to reduce the loss in pocketing the takeover premium on the portion placed in public.

However, if this motive is important, one would likely observe it in ex post data. Particularly, we expect a higher incidence of control transfers post-IPO when the IPO includes a secondary portion. The change in control motive does not offer a rationale for a positive primary portion. * * Insert Table 1 * * 3. Sample Our sample covers the period 1984-2000 and includes all new listings of Belgian firms on the three main exchanges of the country, i.e. the main market of the Brussels Stock Exchange (now Euronext Brussels), Euro.NM Belgium, and EASDAQ (NASDAQ Europe).1 Our sample does not include unit offerings nor reverse LBOs.

For all companies, we obtained the issue prospectus and the 13

consolidated financial statements over the window two years before until one, two or three years after the introduction, depending on the year of the IPO. For the post-IPO analysis, we collected additional data as indicated in Section 5. Figure 1 reveals a concentration of new listings in the periods 1986-1987 and 1996-2000. Other studies (e.g., Pagano et al., 1998; Arosio et al., 2001) show that in other European countries the issue volume is also peaking during these time periods. As expected, increased IPO activity coincides with periods in which the stock market index BASI is booming. Figure 2 shows that deal structure varies over time, with offerings including a positive primary portion occurring more frequently in the later sampling years.2 * * Insert Figures 1 and 2 * * Table 2 shows the industry distribution of the 95 sample firms.3 Similar to Arosio et al.

(2001) for Italy, financial IPOs are concentrated in the earlier years of the sampling period, while high-tech flotations mainly occurred during the later years. Despite the creation of new markets, an important fraction of the high-tech firms continued to opt for the main market of Euronext. In particular, of the 29 high-tech IPOs since 1996, 14 firms (48.28%) listed on the main market, while 9 and 6 firms listed on EASDAQ, respectively Euro-NM Belgium. * * Insert Table 2 * * Table 3 contains some summary statistics on the IPO firms. Given the length of the sampling period and the high inflation rates during the late eighties and early nineties, all absolute statistics are corrected for inflation.

The number of new shares offered in the IPO relative to the number of shares 1 This approach seems justified as firms can choose where to list and as the listing requirements are not very different. 2 In contrast to some other countries in Continental Europe, where reporting and marketing requirements are more stringent for IPOs that include primary shares, the Belgian regulation does not distinguish between primary and secondary shares.

3 As in Pagano et al. (1998), holding companies that concentrate 75% of their assets in a single industrial company are reclassified as belonging to the corresponding industrial sector. Financial firms are kept in the sample as in Belgium, and many other European countries, these firms represent a relatively important subgroup. However, as financial firms may differ from the other sample firms, we have tested the robustness of our results by removing them from the sample. These results, which show that our conclusions are unaffected, can be obtained upon request. 14

pre-IPO (i.e.

the primary portion) in the median firm is 10.50%. The median of the ratio of secondary shares to the number of shares pre-IPO (i.e. the secondary portion) amounts to 15%. The average amount collected as primary funds equals €31,609,018, which is very similar to the average amount of secondary funds collected (€31,924,896). In total, the free float amounts to 25.79% of the shares outstanding post-IPO. Median underpricing, after correcting for the market return, amounts to 5.31%. This figure is rather low compared to the underpricing reported for many other countries (e.g., Ritter, 1991; Leleux, 1993; Arosio et al., 2001), but may reflect the fact that a large majority of the sample IPOs are firm-commitment offerings (see, e.g., Jegadeesh et al., 1993).

Firm age at flotation varies between zero and 283 years, with a median of 18 years. Firm size in the year preceding the IPO is also dispersed, independent of whether it is measured by total assets or sales. The median firm has a return on assets of 14.52%, but profitability again differs widely across firms. Companies are highly levered: on average, 67.16% of total assets are debt-financed, and bank loans represent 42.53% of total debt. Despite high leverage, an average coverage ratio of 22.05 seems to indicate that firms can easily meet debt obligations. The median coverage ratio, however, is much lower (3.09).

From the observation that some firms have a debt ratio above one and/or a negative coverage ratio, it can be concluded that not all firms are financially sound at IPO- time. The average market-to-book ratio, calculated using the offer price, is 3.94. The median firm experiences a 28.22% increase in total assets and a 21.43% growth in sales in the pre-IPO year. Ownership before and after the IPO is highly concentrated. On average, there are 2.46 block holders per firm; together, they own 93.31% of the shares before the IPO. Afterwards, their number and stake are reduced to 1.99, respectively 64.94%.

In only 16% of the cases, initial ownership decreases below 50 percent because of the IPO, but this does not need to imply that initial block holders lose control once listed. Similar results have been found for Italy (Pagano et al., 1998), Germany (Ljungqvist, 1997; Goergen, 1998) and other European countries. * * Insert Table 3 * * 15

Table 4 presents the median of each variable in Table 3 when firms are sorted according to deal structure (i.e. when firms include only a primary portion, only a secondary portion or a combination of both). Reporting separate statistics may give a first indication of whether the profile of IPO-firms differs across deal structure. Table 4 also contains p-values corresponding to a non- parametric Wilcoxon rank sum test of difference in distribution across pairwise subsamples. Table 4 shows that 24 of the IPOs have only a primary portion, 34 have only a secondary portion whereas 37 combine both portions.

The size of the primary and secondary portion is significantly different across all categories. Interestingly, offerings that combine primary and secondary shares have both a smaller primary and secondary portion than the corresponding pure offering category. The amount of primary and secondary funds collected also differs across categories, except for primary funds collected by firms issuing only a primary portion and firms combining both portions. Consequently, although the percentage of shares placed in public is similar, the total amount of funds collected is significantly higher for the latter type of firms.

Especially as compared to offerings with only a secondary portion, firms issuing only primary shares are younger, smaller and more levered, but they do not differ significantly from firms issuing both types of shares with respect to age, size (when measured by total assets) and leverage. Also, IPOs issuing only primary shares generate significantly less cash internally (ROA) when compared to offerings including a secondary portion and have significantly worse coverage ratios. Furthermore, offerings combining both primary and secondary shares have internal cash generation and coverage ratios similar to those issuing only a secondary portion.

Finally, firms issuing only primary shares do not significantly differ from firms combining both portions in terms of growth prospects; by contrast, growth opportunities of offerings including only a secondary portion are significantly lower, especially when considering the market-to-book ratio and the growth rate of total assets one year before the IPO.

In sum, table 4 suggests that firms issuing a primary portion are younger and smaller growth companies with a subgroup that generates more cash internally, services debt more easily and – 16

consistent with the discussion in section 2.2.5. on investor recognition – adds a secondary portion. By contrast, firms that only issue secondary shares are mature and have limited growth opportunities. Simultaneously, the fact that the total fraction of shares placed in public does not differ across subsets of companies, suggests the existence of trade-offs between the size of the primary and secondary portion.

Such a trade-off is predicted by some of the motivations for going public discussed above. It is also likely to arise, when, as may be expected, several motivations simultaneously shape the data in the sample. From that perspective one can consider companies that only include a primary (secondary) portion to have opted for a limit case in a continuum whereby the secondary (primary) portion is set equal to zero.

* * Insert Table 4 * * 4. Determinants of the flotation structure In this section, we first try to identify the variables that explain the size of the primary, respectively secondary portion in IPOs. Table 5, Panel A presents the results – parameter estimates and p-values – from OLS regressions that explain the size of the portion of primary (column one) and secondary (column two) shares. Then, a two equations system is set up to determine the nature of trade-off between both variables; the results are presented in Panel B of Table 5. To limit the influence of outliers, all variables are winsorized at 5-95%, i.e.

extreme values are replaced by the corresponding percentiles. Also, we have checked for multicollinearity by regressing each explanatory variable on all the others.

4.1. OLS analysis of the determinants of the size of the primary, respectively secondary portion On the basis of the discussion in section 2, we estimate the following OLS-model of the primary portion in Table 5, Panel A, column one: 17

PRIMARY PORTION = F (α1MARKET/BOOK + α2ROA + α3LEVERAGE + α4DEBT MIX + α5AGE + α6SIZE + α7CONC + α8CONC*SIZE + α9CARVEOUT + α10MARKET RETURN + α11VOLUME + γiINDUSTRY) (1) Equation (1) shows that in addition to the explanatory variables discussed above, a dummy indicating a carve-out as well as industry dummy variables are included.

Earlier work (e.g., Michaely and Shaw, 1995; Pagano et al., 1998) suggests that including the former dummy variable is useful as the motivation of corporate owners in an IPO may be different from that of owners in stand-alone firms. Concerning the industry dummy variables, we use the classification by Ritter (1991); their parameter estimates are not reported.

The hypothesis that companies with more need for external financing include a larger primary portion is confirmed. The variables indicating growth opportunities (i.e. MARKET/BOOK as well as AGE, SIZE) are significantly positively related to the size of the primary portion. These results suggest that firms with attractive investment prospects go public because they require additional equity. Conversely, internal cash generation significantly lowers the primary issue, in line with the pecking order argument of Myers and Majluf (1984). Leverage does not significantly influence the size of the primary portion, but for a given level of leverage, firms whose debt largely consists of bank loans issue a larger fraction of new shares at IPO-time.

The latter result is supportive for the argument of Rajan (1992) about bank information monopolies and consistent with the findings in Pagano et al. (1998).

The data show that older and larger companies issue a lower fraction of new shares at IPO- time; simultaneously, we also find that ownership concentration and the interaction term of ownership concentration with firm size lack significance. Hence, the hypothesis that adverse selection costs are important in shaping the size of the primary portion is clearly rejected. Neither the stock market return nor IPO volume in the year preceding the IPO does influence the size of the primary portion. These results, however, do not provide conclusive evidence that 18

firms are not trying to benefit from windows of opportunity.

As discussed earlier, buoyant market conditions may also be captured by the market-to-book ratio and create an important market feedback effect on investment. Furthermore, when market returns or IPO clustering are high, firms may still collect a larger amount of new equity through a higher offer price. However, when using the log of primary funds collected as the dependent variable, the historical stock market return and preceding IPO volume are not significant either (not reported). Our results, therefore, do not support the idea that, except possibly for the market feedback effect, firms try to exploit windows of opportunity when raising new equity at IPO-time.

Likewise the primary portion, the size of the secondary portion is estimated on the RHS of equation (1) above; the results are contained in Table 5, Panel A, column two. The hypothesis that diversification is an important driver of the size of the secondary portion is not supported by the data. For, contrary to the predictions of this hypothesis, owners of more stable and mature firms with large cash generation sell a larger portion of their shares as compared to owners of higher risk companies.4 In line with our findings, Pagano et al. (1998), Goergen (1998), and others report supportive evidence for the idea that European IPOs are mainly used to reorganize ownership structure.

In larger, cash generating firms, the firm-specific investment of initial owners is less essential, which might provide an impetus to rearrange ownership.5 In Section 5, we provide more direct evidence on control turnover following the IPO. 4 When owners time the IPO by selling more shares when cash flows are unexpectedly high or even dress up their firm’s financial statements before going public (e.g., Teoh et al., 1998), the same positive relation between size of the secondary portion and ROA might be observed. To test this argument, the variable internal cash flow generation is split up into an expected part, which is based upon the historical value of EBITDA/total assets, and an unexpected or potentially manipulated part.

We find that only the expected part of internal cash flow generation is significant. These results are not reported, but can be obtained from the authors upon request.

5 As the nature of ownership might influence our conclusions, we have also re-estimated the model using the subsample of firms where a venture capitalist is involved in the ownership structure (25 firms), respectively the subsample of firms where no such investor owns shares. Our conclusion that the diversification motive does not drive the size of the 19

In contrast to the primary portion, the data indicate that in determining the size of the secondary portion, adverse selection costs matter. Specifically, although ownership concentration is not significant, the interaction term between ownership concentration and firm size has a significantly negative coefficient.

Hence, owners worry about underpricing when selling their shares at IPO-time, which is consistent with the earlier mentioned arguments of Gomes (2000) and Habib and Ljungqvist (2001), among others. Also in line with adverse selection is the finding that for the secondary portion, the carve-out dummy variable is significantly positive. For this result supports the argument, as developed in Pagano et al. (1998), that carve-outs are subject to smaller information asymmetries and therefore can divest a larger fraction of existing shares at IPO-time. Market conditions, as reflected by the stock market return, and IPO clustering in the pre-IPO year, do not seem to influence the size of the secondary portion.

Also the MARKET/BOOK ratio lacks significance. However, when using the log of the amount of secondary funds collected as the dependent variable (not reported), the historical stock market return is significantly positively related to the amount owners cash in from selling their shares, suggesting a window of opportunity effect after all. Interestingly, this result is driven by the offerings with only a secondary portion. 4.2. The trade-off between the primary and secondary portion: a simultaneous equations model To take the potential simultaneity of the IPO structuring decision into account, we present the results from a simultaneous equations model in Table 5, Panel B.

Given the explanatory variables used in Panel A, the model is not identified. Therefore, we remove the variables CONC and CONC * SIZE from the equation explaining the primary portion. The variable DEBT MIX is deleted in the secondary portion equation.

The results from the equation explaining the primary portion in column one of Panel B are in line with those from Panel A and show that taking into account the interaction with the size of the secondary portion holds in both subsamples. This result is not surprising since the venture capitalist’s equity stake at IPO-time is relatively small. 20

secondary portion does not add information. By contrast, the significant impact of the size of the primary portion on the secondary portion indicates that the former is an important driver of the latter. Specifically, the data show that if the size of the primary portion is already large, owners divest only a small fraction of their shares.

Furthermore, internal cash generation and size no longer significantly affect the secondary portion once the interaction is accounted for. That is, the primary portion incorporates all relevant information these variables add to the explanation of the size of the secondary portion. This change in significance levels, however, leaves our earlier conclusion that diversification is not an important force in shaping the size of the secondary portion, unaltered. In Panel B, column two historical IPO-volume carries a negative and significant coefficient, which implies that in hot IPO-markets the size of the secondary portion tends to be smaller as compared to cold markets.

However, when using the log of the amount of secondary funds collected as the dependent variable (not reported), this significance disappears. This indicates that the smaller secondary portion likely is the result of a higher issue price during hot IPO-markets. The loss of significance of explanatory variables – except those linked to adverse selection – once the information embedded in the primary portion is accounted for, highlights the importance of the latter in shaping the size of the secondary portion. In fact, the data seem to be generally consistent with a decision process that could be summarized as follows.

In case the company has financing needs, the size of the primary portion is chosen first and is determined by these needs. If this portion is relatively large, no secondary shares are added. However, if financing needs can be filled with a relatively small primary portion, particularly when internal cash generation is substantial, owners divest some of their own shares. In case the company has no financing needs, the primary portion is set equal to zero and a relatively large fraction of secondary shares is sold to the public.

Except for the earlier documented result that firms issuing only secondary shares benefit from windows of opportunity, the findings in Table 5 do not support any motive for selling secondary shares; the data only indicate that owners try to avoid (adverse selection) costs associated with the 21

operation. Consequently, what begs further investigation is the question why owners of firms with financing needs add a secondary portion if the primary portion is small. The analysis in section 2 above, the absence of support for any other motive, the fact that firms adding a secondary portion to the primary issue are young growth companies and the univariate statistics in section 3, all suggest that investor recognition may play an important role.

Second, the benefits associated with going public for firms without financing requirements, need further clarification. The discussion in section 2 as well as the fact that firms issuing only secondary shares are relatively mature, indicate the need to investigate the control turnover motive for these companies. These issues will be addressed in section 5 below; we end this section with a discussion of the robustness checks we implemented. 4.3. Robustness checks The introduction section mentioned that changes occurred in the IPO market during the sampling period. To distinguish between changes in the values of the independent variables, that reflect changes in the characteristics of the firms going public, versus changes in the underlying structure of the IPO market, we checked for structural breaks.

Specifically, we allowed the coefficient of each variable in Table 4 to show a break at the beginning of 1996.6 In a first model, which consumes more degrees of freedom, all variables are allowed to change at once, whereas in a second model, the coefficient of each variable – one by one – could differ before and as of 1996. Using a Chow test, the hypotheses of identical parameter estimates could not be rejected in the equation explaining the primary portion. However, in the secondary portion equation, some of the parameter estimates turned out to be significantly different across both subperiods.

After including year dummy variables, these differences disappeared without affecting the significance of the other explanatory variables. These results indicate that the relations between firm characteristics and the primary, respectively secondary portion have remained stable over time, although institutional features (e.g., 6 Using 1995 as the dividing year results in the same classification, whereas using 1997 leads to fewer differences in parameter estimates. The results of these robustness checks are not reported, but can be obtained from the authors. 22

bookbuilding) and the mixture of firms engaging in an IPO have changed. Section 3, for instance, showed that most financial companies in the sample listed in the earlier years while the high-tech offerings mainly have taken place in the later years. As a last check, the simultaneous regression model has been re-estimated using only the subsample of IPOs that listed in the second half of the nineties. The earlier conclusions prove to be robust. * * Insert Table 5 * * 5. Post-IPO characteristics In this section, we investigate whether post-IPO data support the hypothesis that at the IPO, investor recognition is an important factor driving the size of the secondary portion for growth firms.

Similarly, we investigate whether post-IPO data confirm the idea that subsequent control turnover may be a benefit owners of mature firms bear in mind at IPO-time. Specifically, we examine market liquidity, seasoned equity offerings and control turnover following the IPO and before December 2001. A major problem we encountered is that for many firms in the sample we do not have information that extends a long period beyond the IPO. For an examination of market liquidity, the problem is limited though. Listings with a history shorter than one, respectively two years are simply disregarded, depending upon the horizon under consideration.

In line with the literature, we use a window of three, respectively five years to analyze seasoned equity offerings respectively takeovers. Therefore, we especially consider our findings with respect to takeovers as preliminary evidence. 5.1. Market liquidity Presently, we investigate the hypothesis that securing market liquidity is an important consideration for firms at the time of the IPO. Post-IPO market liquidity is hereby defined as the number of shares traded during a horizon of one, respectively two years starting one month after the IPO divided by the number of shares outstanding after the IPO (see also Eckbo and Norli, 2000).

The first post-IPO 23

month is disregarded to correct for the fact that early liquidity may be affected by the adopted distribution rules. To determine whether free float influences post-IPO market liquidity, we calculate total funds collected as the log of the number of shares placed in public times the offer price. To control for firm age and size, the log of firm age at the IPO, respectively the log of total assets post-IPO are included. Following the specification of Brennan and Subrahmanyam (1995), Chordia et al. (2001) and others, we control for additional factors that may impact on market liquidity: our earlier measure for growth opportunities, a dummy that equals one when the firm lists on a market for innovative growth companies (i.e.

EASDAQ or Euro.NM Belgium), a dummy that equals one when at least one market maker is appointed, the historical stock market return and our measure for IPO-activity. Again, we include industry dummy variables using Ritter’s (1991) classification. The model of post- IPO market liquidity thus looks as follows: LIQUIDITY = F (α1FUNDS COLLECTED + α2MARKET/BOOK + α3AGE + α4SIZE + α5MARKET TYPE + α6MARKET MAKER + α7MARKET RETURN + α8VOLUME + γiINDUSTRY (2) The results shown in Table 6 are in line with the empirical findings in the literature. However, the main piece of evidence for this research is the fact that increasing free float (i.e.

funds collected) at IPO-time significantly increases market liquidity in the years following the IPO. Hence the data are consistent with the idea that complementing a relatively small primary issue with a secondary portion significantly helps to secure a sufficient free float and hence liquidity. Next, we find evidence of different levels of market liquidity depending upon the exchange on which the firm lists. Hence, as pointed out by Corwin and Harris (2001), the selection of the appropriate stock market is an important consideration for IPO candidates. Appointing a market maker also significantly increases market liquidity, especially over longer horizons.

The latter result is not surprising as the task of a market maker mainly consists of guaranteeing market liquidity. 24