Going public in bear markets* Pengda Fan

Graduate School of Economics, Kyushu University 6-19-1, Hakozaki, Higashiku Fukuoka 812-8581 JAPAN Konari Uchida**

Faculty of Economics, Kyushu University 6-19-1, Hakozaki, Higashiku Fukuoka 812-8581 JAPAN

Abstract We find that Japanese private firms in industries characterized by keen competition and small variation in production costs are more likely to go public in bear markets than do companies in industries without those characteristics. We also find that firms relying on bank finance are more likely to go public in bear markets than those with outstanding bonds. Firms going public in bear markets (bear market IPOs) have smaller proceeds, lower offering price, and stockpile less cash from IPO proceeds than other IPOs do. Besides, bear market IPOs have greater sales growth from the pre-IPO level, show less financing constraints, and make more efficient investment decisions during the post-IPO period than do companies going public in non-bear markets. Those results suggest that firms which put relatively low priority on equity issues as a purpose of IPO are willing to go public even in bear markets. Key words: IPO; Bank; Bear market; Product market; Investment; JEL classification code: G21; G30; G31

*

An early version of this paper was presented at the Japan Finance Association conference and finance seminar at Nagoya University. We thank Kotaro Inoue, Hideaki Kato, Hideo Okamura, Katsutoshi Shimizu, and Hidenori Takahashi for their helpful comments. We are grateful for financial support provided by JSPS KAHENHI Grant Number 15H03367. ** Corresponding author. Faculty of Economics, Kyushu University 6-19-1, Hakozaki, Higashiku, Fukuoka 812-8581 JAPAN. Tel.: +81-92-642-2463 E-mail: [email protected] 1

1. Introduction This paper explores characteristics of firms going public in bear markets. Previous studies stress that initial public offerings (IPOs) are disproportionally distributed in the period following bull market (e.g., Ritter, 1984; Loughran et al.,1994; Helwege and Liang, 2004; Lowry and Schwert, 2002; Alti, 2005; Pastor and Veronesi, 2005; Yung et al., 2008; Chemmanur and He, 2011). However, the stock market can show long-run slump in the recent matured economy. For instance, CAC 40 (representative French stock price index) showed 30 month reversal after it recorded its historical high (6,922.33) on September 4, 2000, during which the index declined to 2,618.46 on March, 2003. As of August 2015, CAC 40 index is approximately two-thirds of the historical high. When firms desire to go public for various purposes such as reputation and market share increases, they may need to consider IPO in bear markets. Indeed, more than one-fourth of our sample companies (1,793 Japanese firms that went public during the period from 1997 to 2014) decided to go public after a significant decline of the Tokyo Stock Price Index (TOPIX) (specifically, after -10% or worse six-month index return). This figure is unexpectedly high, given that firms can withdraw IPOs and wait for good market conditions (Busaba et al., 2001; Dunbar and Foerster, 2008). It is important to uncover what firms go public in bear markets to understand IPOs in recent matured economy. Exploring bear market IPOs also highlights objectives of IPOs other than equity issuance. To the best of our knowledge, however, there are only few studies to address the issue. This paper attempts to fill this void. Generally, financing in securities markets are highly vulnerable to market conditions. Kim and Weisbach (2008) indicate that companies tend to time the market in IPO and seasoned equity offerings (SEOs) to raise cash more than their immediate needs. It is especially the case for young companies, and thus firms should prefer to go public in good market conditions if they put a high weight on equity financing as a motive of IPO. Meanwhile, previous studies suggest various purposes of IPOs such as increases of visibility and market share. We premise that firms with relatively low priority on equity financing go public irrespective of market conditions (thus they are willing to go public even in bear markets). Specifically, we pay attention to market share improvements as a non-equity financing objective of IPOs, given that Stoughton et al. (2001) and Chemmanur and He (2011) argue that IPOs increase firms’ publicity and then increase market share of the firm. Since the motive should be especially strong for firms which are subject to fierce competition and difficulty in product differentiation, we examine the relation between industry competitiveness, variation in gross profit margin, and the probability of firms going public in bear markets (Stoughton et al., 2001). 2

We next examine whether the existence of alternative financing sources decreases IPO firms’ priority on equity financing, and in turn affects firms’ decisions to go public. Previous studies commonly suggest that banks tend to keep long-term relations with borrowing companies, which significantly decrease renegotiation costs and information asymmetry. Although venture capitals (VCs) are generally viewed as important finance providers for private companies, bank debt is also important for private companies (Berger and Udell, 1995, 2002). A distinct difference between those two financing sources is that VC-firm relation generally terminates at the firm’s IPO while bank-firm relation can continue even after IPO. It is plausible to presume that relationships with banks which young companies have established before their IPOs enable them to timely finance their projects even after the IPO. Although bond issues are also available to private companies, bond issues are generally one-time financing activities in which suppliers of funds are not supposed to continuously provide credits to the firm. Therefore, companies which prefer securities financing to bank finance should consider IPOs as an important opportunity to raise external capital for future investments, and thus favor bull markets for their IPOs. Taken together, we hypothesize that firms relying on bank debt put relatively low priority on equity issues as a motive for IPO, and are more likely to go public in bear markets than those relying less on bank finance. We address those issues by using Japanese IPO data. Since 1990s, the Japanese stock market has shown a long-term downturn trend. After the Tokyo Stock Price Index (TOPIX) recorded its historical high price (2,884.80) on December 18, 1989, it sharply declined during 1990s, and the current price is less than half of its peak. This fact suggests that many Japanese private companies need to consider IPO in bear markets to pursue non-equity finance objectives. Another advantage of Japanese data is that Japanese private firms are likely to rely on bank debt, given the well-known fact that Japanese companies establish long-term relations with their main banks. Our main analysis defines bear market IPOs as firms going public after the TOPIX records six month buy-and-hold return lower than -10%. Investigating Japanese IPOs during the period from 1997 to 2014, we find that firms in industries with sever competition and small variation in firms’ gross profit margin are more likely to go public in bear markets than those from other industries. This finding supports our hypothesis that firms with a desire to increase visibility and market shares are willing to go public in bear markets. We also find that the ratio of bank loans to debt, bank ownership, and director appointment from a bank are positively associated with the probability of firms going public in bear markets. In contrast, firms which have relatively large outstanding bonds are less likely to go public in bear markets. Taken together, the results suggest that firms are willing to go public even in bear markets if 3

they put relatively low priority on equity financing as an objective of IPO. We implement various additional analyses. Consistent with our view, we find that IPOs in bear markets have significantly smaller proceeds, lower offering price, and stockpile cash less than do IPO companies in non-bear markets. Bear market IPOs show significantly superior long-term post-IPO stock price performance than other IPOs, suggesting that bear market IPOs are less overvalued. We also find that Bear market IPOs increase sales from the pre-IPO level significantly more than do Non-BEAR IPOs for a few years after IPO. This result reinforces our first hypothesis that BEAR IPOs go public to increase market share. Bear market IPO firms also show significantly smaller investment-to-cash flow sensitivity and greater investment-Tobin’s Q sensitivity during the post-IPO period than non-bear market IPO firms. Those results are consistent with our second hypothesis, suggesting that firms going public in bear markets can timely finance projects since they have a stable financing source. This paper makes significant contributions to the literature. To the best of our knowledge, this is the first research to examine characteristics of companies going public in bear markets. Previous studies have paid particular attention to hot market IPOs. For instance, Ritter (1984) indicates that risky companies such as natural resource firms tend to go public in hot markets. Yung et al. (2008) and Chemmanur and He (2011) suggest that low quality firms go public in good market conditions. In addition to exploring a non-negligible aspect of IPO, our analyses show new evidence that product market characteristics and reliance on bank financing significantly decreases firms’ incentives to go public in good market conditions. Stoughton et al. (2001) suggest firms in industries characterized by small differences in marginal production costs tend to go public in hot markets to improve reputation. Chemmanur and He (2011) argue that IPO waves tend to occur in competitive industries. Our analyses on product market characteristics reinforce those arguments by examining determinants of bear market IPOs. Secondly, it would be a novel finding that market condition at the IPO is correlated with financing constraints and investment efficiency during the post-IPO period. We argue that the correlation exists since firms with stable and timely financing sources do not need to care on market conditions at the time of IPO. Finally, we present new evidence of roles of banks for young private companies. Although previous studies have stressed that VCs support young companies in finance and management, we argue that banks provide young firms with a stable financing channel which continues even after the IPO, and thereby enable firms to go public even in bear markets for product market purposes. Overall, our findings suggest that non-equity issue motives are important factors associated with firms’ IPO decisions. 4

The reminder of the paper is organized as follows. Section 2 presents literature review and hypotheses. Section 3 descries sample selection, data, and definition of bear market IPOs. Section 4 presents empirical results. Section 5 is a brief summary and conclusion of this research. 2. Literature review and hypothesis It is well-documented that initial public offerings are disproportionally concentrated in the period after significant stock price increases. Large body of literature attempts to explain the phenomenon of hot IPO market. Ritter (1984) indicates that risky companies tend to go public during a specific period (hot market), in which significant underpricing exists due to the high risk nature of IPO companies. Chemmanur and He (2011) argue that positive economic shocks motivate less productive firms to go public. Yung et al. (2008) find firms going public in hot markets have high average initial return, large cross-sectional variance in long-term stock returns, and high probability of post-IPO delisting. Lowry and Schwert (2002) report a positive lead-lag relation between initial return and IPO volume, which is likely attributable to information on investors’ sentiment learned during the registration period. Pastor and Veronesi (2005) argue that clustering of IPOs during certain periods is attributable to firms’ market timing behaviors. IPO is one of important financing activities in corporate life, in which firms raise substantial equity capital. Generally, firms can raise funds in good conditions from the securities market when the market goes well. Indeed, many previous studies show evidence that managers time the market to conduct seasoned equity offerings (Baker and Wurgler, 2002; Graham and Harvey, 2001). That should be the case for IPOs, since the book-building system determines offering price in consideration of concurrent investors’ demands. In terms of equity issues, firms will have an incentive to time the market in their IPO decisions to raise as large lump sum funds as possible for future investments (Kim and Weisbach, 2008).1 IPOs provide shareholders (e.g., founder, venture capitalist) with an opportunity to realize capital gains, which also incentivizes them to time the market (Lerner, 1994). Loughran and Ritter (1995) show that companies issuing stocks during 1970 to 1990 (IPO and seasoned equity offering) significantly underperform non-issuing firms for five years after the offering date. Using a sample of 350 privately held venture-backed biotechnology firms going public between 1978 and 1992, Lerner (1994) shows that firms go public (employ private financings) when equity valuations are high (low). Kim and Weisbach (2008) 1

Pagano, Panetta, and Zingales (1998) show evidence that the likelihood of firms’ going public increases with the market-to-book ratio in the firm’s industry. However, they argue that firms go public to rearrange capital structure rather than to finance future investments. 5

find that high market to book firms tend to save more cash from proceeds of IPOs and offer a higher fraction of secondary shares in seasoned equity offerings (SEOs) than low market to book firms. A complexity of IPO is that firms may go public for various purposes other than equity issues and capital gain realization. IPO can significantly increase the firm’s reputation (Stoughton et al., 2001) and market share in the product market (Chemmanur and He, 2011) as well as managers’ nonpecuniary utility (such as self-respect and sense of achievement). A survey of 336 CFOs by Brau and Fawcett (2006) suggests that an important motivation for IPO is to facilitate potential takeover transactions. IPO reduces valuation uncertainty and allows firms to pursue a more efficient acquisition strategy (Hsieh et al., 2011). To the degree that firms believe bull market will come soon, they are likely to wait and time the market to conduct IPO. Indeed, Dunbar and Foerster (2008) indicates 1,473 (approximately 20%) of 7,442 US firms which filed IPOs during 1985 to 2000 withdraw their IPOs before the completion. In the recent matured economy, however, stock markets can be slumping over relatively long period and it is not easy to predict when bull market will come. Such a situation makes private firms, which desire to go public for reasons other than equity financing and capital gain realization, encounter a trade-off problem. By going public in bear markets, firms can immediately achieve non-equity issue objectives of IPOs while they need to accept poor conditions for equity issues. A plausible prediction is that firms go public irrespective of market conditions when they do not put high priority on equity issues (and capital gain realization) as a purpose of IPO. We focus on two factors which are likely to decrease relative importance of equity financing as a motive of IPO. Firstly, we premise that firms do not care on market conditions when they wish to go public mainly for reputation in their product markets. Generally, IPOs are expected to increase firms’ visibility and in turn market shares in product markets (Chemmanur and He, 2011). This nature should bring significant benefits to companies in competitive industries. The increased publicity is also advantageous when firms find it difficult to differentiate themselves from competitors in production process. Stoughton et al. (2001) suggest that firms in industries characterized by small differences in marginal production costs have an incentive to go public for reputation improvement. Those ideas give rise to the following hypothesis: Hypothesis 1: Private firms in industries with fierce competition and small variance in marginal production costs are more likely to go public in bear markets than do firms from less competitive industries with large heterogeneity in marginal 6

production costs. Next, we posit that firms’ financing patterns preceding IPOs are associated with importance of IPO as a place of equity issuance. Although it is commonly described that VCs are important suppliers of funds for young private companies, those companies also use bank borrowings as an important financing source (Berger and Udell, 1995, 2002). One might argue that banks hesitate to provide credits to immature private firms, but previous studies find that investments of bank-affiliated VCs significantly increase firms’ access to bank debt (Hellmann et al., 2008; Sun and Uchida, 2016). Sun and Uchida (2016) also find that Japanese banks provide credits to large IPO companies with substantial tangible assets. A distinctive feature of bank finance is long-term relation between lenders and borrowers, which lasts even after IPO.2 Besides, banks tend to provide financial supports (Hoshi et al., 1990) and privately restructure debt for financially distressed companies (Gilson et al., 1990). These facts suggest that banks can serve as a stable finance provider for IPO companies once a relationship is established. Accordingly, we predict that firms do not need to raise funds as much as possible in good conditions by IPO, if they mainly use bank finance. In contrast, VCs generally exit from investee firms after the IPO (VCs do not provide capital after the IPO). Private companies also issue bonds, but bond issues are generally viewed as one-time financing transactions which do not suppose continuous trading between lenders and borrowers. Besides, issues of securities are susceptible to market conditions, and inherently subject to uncertainty.3 These facts will increase importance of IPOs as a financing vehicle for firms which do not rely on bank finance. These discussions give rise to the following prediction. Hypothesis 2: Private firms relying on bank finance are more likely to go public in bear markets than those relying less on bank finance. We test the hypotheses by using Japanese IPO data. As mentioned, the Japanese stock market has suffered from long-term slump since the 1990s. In addition, it is well-documented that Japanese companies establish long-term relations with their 2

Sun and Uchida (2016) find that banks do not significantly decrease lending and equity stakes to companies even after the IPO, although their subsidiary VCs sell most of shares in a few years after the IPO. 3 Recent studies on the global financial crisis also argue that banks’ lending behaviors are affected by financing structures. Ivashina and Scharfstein (2010) and Cornett et al, (2011) show that banks relying more on core deposit and equity capital financing, which are stable sources of financing, continued to lend relative to other banks during the global financial crisis. 7

main banks. Hoshi et al. (1991) show evidence that Japanese firms affiliated with large banks show significantly smaller investment-to-cash flow sensitivities than non-affiliated firms do, which is commonly interpreted that main banks mitigate information asymmetry and release firms from financial constraints. The relationship lending is also evident for young private companies in Japan. Bank-affiliated VCs are one of predominating forms in the Japanese VC industry, and their investments help young companies build relation with parent banks (Sun and Uchida, 2016). Takahashi (2015) argues that banks establish lending relations with start-up companies even before equity investments by their subsidiary VCs. These facts suggest that Japanese IPO data are advantageous to address the hypothesis. 3. Sample selection and data We collect information of firms that went public in the Japanese stock market during the period of 1997 to 2014 from Nikkei NEEDS. We start our sample period at 1997, since the book-building system was introduced at that year. 4 For those companies, we manually collect detailed IPO information from Prof. Takashi Kaneko’s website (offering price, proceeds, firm age, and so on) and Japanese IPO White Papers (the number of primary and secondary shares, offering price, first trading date opening price, firm age, lead underwriter, banks which have business relationship with the IPO firm, main bank and its ownership, VC ownership, and so on).5 When the Prof. Kaneko’s data are inconsistent with the data from White Papers, we adopt data of White Papers. The IPO data are merged with financial and stock price data, which are available from Nikkei NEEDS Financial Quest and Portfolio Master. Financial institutions and utilities are removed from the sample. As a result, our entire sample consists of 1793 IPO companies. When a sample firm has VC shareholders, we manually identify affiliations of the VCs by using the Handbook of Venture Capital (issued by Venture Enterprise Center, a Japanese institute of venture capitalists) to compute ownership of VCs affiliated with the firm’s main bank. We also hand-collect data of directors appointed by banks from the firms’ prospectus, which is available only from 2001. To identify firms that decided to go public in bear market, we calculate buy-and-hold return (BHR) of TOPIX during the 126 trading days ending at 22 days before the first trading day (from day -147 to day -22, where day 0 is the first trading day) for each of sample companies. Previous studies commonly identify hot market 4

Auction method is also allowed for offering price determination in Japan. However, no IPOs have adopted the auction method since the introduction of book-building system. 5 Prof. Kaneko is a Japanese IPO researcher. The web-site URL is: http://www.fbc.keio.ac.jp/~kaneko/KP-JIPO/top.htm. 8

IPOs by using market condition variables (index return, underpricing, and the number of IPOs) during several months before listing date. However, this method may not accurately capture IPO decisions in bull markets, because approximately one month interval exists between the submission date of the first prospectus, which is also the date of shareholder meeting for IPO approval, to the first trading date. We do not include data during one month preceding the first trading date for index return calculation to prevent unpredictable market condition changes in the one month interval from contaminating the identification variable (we assume that one month has 21 trading days). Specifically, our main analysis defines as bear market IPOs (BEAR IPOs) those with the 126 day BHR lower than -10%. All the other sample companies are classified as non-bear market IPOs (Non-BEAR IPOs). We also employ various definitions of BEAR IPOs as robustness checks. For instance, we define BEAR IPOs as IPO firms with 3-month TOPIX return (from day -84 to day -22) lower than -5% and those with 12 month TOPIX return (from day -273 to day -22) lower than -20%. We also replicate the analyses by deleting BEAR IPOs with positive (or larger than 1% or 2%) market returns during day -21 to day -1 to exclude firms going public in improved market conditions from BEAR IPOs. We also use monthly stock return data (from month -7 to month -2, where month 0 is the month of going public) instead of daily data. Finally, we adopt JASDAQ index (JASDAQ is a main IPO market in Japan) rather than TOPIX. These analyses present qualitatively same results (results become stronger in some analyses). The following part of the paper presents results when we define BEAR IPOs as IPO companies with the 126 trading day BHR lower than -10%. Table 1 presents sample distribution by year as well as the closing price of TOPIX for the corresponding year. It is noticeable that about one-fourth of sample companies are classified as BEAR IPOs. Under the recent Japanese market condition, it is not rare that companies decide to go public when the market is going down. In other words, bear market IPO is a non-negligible aspect of IPO. Table 1 also indicates significant variation of the proportion of BEAR IPOs across years. For instance, BEAR IPOs dominate Non-BEAR IPOs in frequency for year 1998, 2001, 2002, and 2008, when stock prices went down. In contrast, we find no BEAR IPOs for years of strong stock price movements (year 2005, 2006, 2013, and 2014). [Insert Table 1 about here] To test Hypothesis 1, we compute the Harfindahl-Hirschman index (HHI) for the IPO firm’s industry at the year before IPO by using all listed firms’ data (see Appendix for definition of variable). For similarity of marginal production costs, we compute the standard deviation of gross profit margin for all listed firms in the industry at the year before IPO (SD_Margin). Industry classification by Tokyo Stock 9

Exchange is employed for the variable computations. As for Hypothesis 2, leverage is a potential proxy variable for bank-firm relation since firms relying on bank debt tend to have high leverage, and firms which have mainly raised funds from VCs will have low leverage. We compute leverage by total liabilities over assets (LEVERAGE). However, we do not employ LEVERAGE as our key proxy variable for bank reliance, since highly leveraged firms have an incentive to raise substantial equity capital at IPO to rearrange their capital structures (Pagano, Panetta, and Zingales (1998)). Instead, we employ the ratio of bank loans to total debt (LOANDR), which is commonly used as a measure of firms’ reliance on bank debt in the literature of main bank (Morck and Nakamura, 1999; Kang and Stulz, 2000; Kang et al., 2000). LOANDR becomes small as firms choose bond issues rather than bank debt. We also compute ratios of loans and bonds (including commercial papers) respectively over assets (LOANAR and BONDAR) to test our hypothesis. Japanese banks are allowed to hold up to 5% of outstanding shares of companies. Sun and Uchida (2016) find that bank lending to IPO companies is significantly associated with percentage ownership of the bank. We adopt percentage ownership by banks (BANKOWN), which have business relationships with the firm in the IPO White Paper. Given that Japanese banks send directors to borrowing firms (Kaplan and Minton, 1994; Sun and Uchida, 2016), a dummy variable indicating existence of directors appointed from banks is also adopted (BDIRECD). BDIRECD is available only from 2001. We winsorize all continuous variables at the top and bottom one percent values. [Insert Table 1 about here] Table 2 presents summary statistics separately for BEAR IPOs and Non-BEAR IPOs. For financial variables, data for the year before IPO are presented while BDIRECD and ownership variables are from IPO prospectus and White Papers. Table 2 shows BEAR IPOs have significantly smaller HHI, which indicate more fierce competition, than Non-BEAR IPOs (0.118 vs 0.193) do. Although we do not find a significant difference in SD_Margin between the two groups, the HHI result is consistent with Hypothesis 1. As for Hypothesis 2, the mean LOANDR is 80.2% for BEAR IPOs, which is significantly higher than that of Non-BEAR IPOs (74.3%). The mean BONDAR is 1.6% for Non-BEAR IPOs, which is significantly greater than that of BEAR IPOs (0.9%). Although the median BONDAR is zero for the two subsamples, the presented findings suggest that Non-BEAR IPOs tend more to issue bonds than do BEAR IPOs. Table 2 also suggests that BEAR IPOs have strong relationship with banks in terms of shareholdings. BANKOWN is about 2.6% for BEAR IPOs, which is significantly greater than that for Non-BEAR IPOs (1.9%). We also examine ownership by main 10

bank (MBOWN), which is identified in the With Papers. Table 2 suggests that MBOWN is also significantly greater for BEAR IPOs than for Non-BEAR IPOs. Panel B of Table 2 also indicates that the BEAR IPOs have significantly greater likelihood to have positive MBOWN and MBVCOWN (percentage ownership by main bank and its affiliated VC). Besides, Panel B of Table 2 indicates that BEAR IPOs have significantly greater probability of having directors appointed from banks than Non-BEAR IPOs do (8% versus 4.4%). Overall, the univariate analyses support Hypothesis 2 that private firms relying on bank finance are willing to go public in bear markets. [Insert Table 2 about here] Previous studies commonly find that hot market IPOs are accompanied by high underpricing. We also find that BEAR IPOs have significantly lower initial return (opening price of first listing day less offering price divided by offering price) than Non-BEAR IPOs do (36.5% versus 70.0%). With regard to other variables, BEAR IPOs are significantly larger, and the median firm age is higher for BEAR IPOs. This fact is in spirit consistent with our hypothesis, given the conventional wisdom that banks tend to provide credits to larger and matured firms.6 As for performance variables, we find a significant difference in median ROA (operating income over assets) between the two groups, but no significant difference exists in the sales growth ratio (SGR). 4. Empirical results 4.1 What firms do go public in bear markets? This section implements logit regressions, in which the dependent variable takes a value of one for BEAR IPOs and zero for Non-BEAR IPOs (BEARIPOD), to test our hypothesis with controlling for various factors. As with the univariate analysis, data before the IPO are employed for financial variables while prospectus and White Papers data are used for BDIRECD and ownership variables. We do not include year dummies due to inherent high correlation between BEARIPOD and year dummies. Instead, we add the natural logarithm of TOPIX at the IPO date (Ln_TOPIX) to capture the macro-economic conditions at the time of IPO. All estimations include industry dummies to control for industry characteristics other than the competitiveness and similarity of production costs. Results are shown in Table 3. [Insert Table 3 about here] Panel A employs HHI as an industry characteristic variable. Consistent with 6

Sun et al. (2012) also find that Japanese IPOs owned by finance-affiliated VCs are more matured than those owned by independent VCs. 11

Hypothesis 1, HHI has a negative and significant coefficient in all models except for Models (5) and (6), suggesting that firms in more competitive industries are more likely to go public in bear markets. The industry competition is an economically significant determinant of bear market IPOs. The presented coefficient suggests that a one standard deviation increase in HHI decreases the likelihood of conducting BEAR IPOs by 20%, holding all other variables at their mean values. It would be noteworthy that Models (5) and (6) also carry a significant coefficient on HHI when we delete industry dummies. Panel B of Table 3 adopts SD_Margin as a product market characteristic variable. Regression results carry a negative and significant coefficient on the variable except for Models (5) and (6). The result indicates that firms in homogeneous industries in production costs are willing to go public in bear markets. Again, the SD_Margin has an economically significant impact on the probability of going public in bear markets. Estimated coefficients suggest that a one standard deviation increase in the variable decreases the likelihood by 8.8%. As a robustness check, we also compute those industry characteristics variables by using Nikkei middle industry classification. Results carry a negative and significant coefficient on SD_Margin, while HHI has a marginally significant coefficient (10% significance level). Regarding Hypothesis 2, Model (1) of both panels engenders a positive and significant coefficient on LOANDR, suggesting that firms relying on bank loans are likely to go public in bear markets. Estimated coefficients suggest that a one standard deviation increase in LOANDR increases the likelihood of conducting BEAR IPOs by 4.07%. The economic impact is significant, given that the unconditional frequency of BEAR IPO is 25.6%. Meanwhile, LEVERAGE has a negative coefficient. We interpret that highly-leveraged IPO firms avoid bear markets since they need substantial equity capital to adjust capital structures. Model (2) replaces LOANDR by LOANAR, BONDAR, and a dummy variable which takes a value of one for firms with positive BONDAR (BONDD). We include BONDD, since BONDAR takes a value of zero for many observations. The estimation carries a negative and significant coefficient on BONDAR, indicating that firms with more outstanding bonds are less likely to go public in bear markets. The insignificant coefficient on LOANAR is attributable to its high correlation with LEVERAGE (correlation coefficient is 0.6). That is, LOANR is likely to absorb effects of leverage and bank debt reliance on bear market IPO. The presented coefficient suggests that a one standard deviation increase in BONDAR decreases the likelihood of conducting BEAR IPOs by 4.8%. We also examine whether firms highly using bonds tend less to go public in bear markets. Specifically, we make a dummy variable which takes a value of one for firms with BONDAR greater than 10% and 12

zero otherwise (BONDHIGHD). 7 Model (3) carries a negative and significant coefficient on BONDHIGHD, suggesting that the probability of going public in bear markets significantly declines when a firm has outstanding bonds beyond a certain level. Given that bond issues generally do not build long-term relations with lenders, those results suggest that firms using non-relationship-based financing tend less to go public in bear markets. We also adopt non-capital structure variables as a proxy for reliance on bank finance. Model (4) carries a positive and significant coefficient on BANKOWN. Besides, Model (5) employs existence of directors appointed from banks (BDIRECD) as a measure of reliance on bank finance (sample size declines to 1181 since director appointment data are only available since 2001). Both panels engender a positive coefficient on BDIRECD, although the coefficients are not statistically significant. A potential reason for the insignificant coefficient on BDIRECD is that banks do not frequently appoint directors to firms in good economic conditions. Indeed, BDIRECD has a positive and significant coefficient once we exclude Ln_TOPIX. Finally, Model (6) includes all the proxies for reliance on bank finance and those variables are still statistically significant in predicted signs. Taken all together, results in Table 3 are generally consistent with our view that private firms go public irrespective of market conditions when they put low priority on equity financing as a motive of IPO. With respect to control variables, large firms have greater probability of going public in bear markets. We interpret that large companies do not put high priority on equity issuance as a motivation of IPOs since they have stable financing sources. Ln_TOPIX has a significantly negative coefficient probably because the stock price index tends to be low in bear markets. We do not find significant coefficients for other variables. 4.2 Is equity financing less important for bear market IPOs? Our hypotheses stand on the view that BEAR IPOs put relatively low priority on equity financing. Indeed, the univariate analysis (Table 2) suggests that the median primary proceeds over assets for BEAR IPOs are almost half of that for Non-BEAR IPOs (8.7% versus 15.2%). Following Kim and Ritter (1999), we also examine offering price to operating income ratio, and find that the ratio is significantly smaller for BEAR IPOs than for Non-BEAR IPOs. Untabulated analyses also compute other IPO valuation ratios (offering price to sales ratio and offering price to net income ratio), and find the qualitatively same result. In addition, we implement regression analyses of those valuation ratios as well as primary proceeds over assets with 7

For robustness checks, we also employ different cut points (5%, 12%) and obtain qualitatively the same result (untabulated). 13

controlling for various factors including Ln_TOPIX, which represents market condition at the time of IPO. Untabulated results engender a negative and significant coefficient on BEARIPOD, suggesting that BEAR IPOs accept poor equity financing conditions at the IPO. Given that firms have an option to withdraw IPO (Busaba et al., 2001; Dunbar and Foerster, 2008), the result is consistent with our view that BEAR IPOs put relatively low weight on equity financing as a purpose of IPO. Nevertheless, one can raise a concern that the aforementioned results are simply attributable to demand side factors (e.g., investor sentiment) rather than to firms’ purposes of IPO. To further examine our underlying view, we examine cash stockpiling behaviors of sample companies. Kim and Weisbach (2008) show evidence that firms tend to increase cash holdings by 49.0 cents before and after IPO for every dollar raised by IPO, suggesting that IPO firms generally raise funds greater than their urgent investment needs. Meanwhile, firms should stockpile cash less if they put low priority on equity financing. Table 4 presents cash holdings variables. Although cash holdings scaled by assets are not significantly different at the year before IPO (Year -1) between the two subsamples, BEAR IPOs have significantly smaller cash holdings at the end of IPO year (Year 0) than Non-BEAR IPOs do (25.5% versus 28.8% in mean values). Accordingly, BEAR IPOs show significantly smaller increases in cash (scaled by assets at Year -1) surrounding IPO than Non-BEAR IPO firms do. For instance, the median of one year increase in cash (scaled by total assets at the year before IPO) for Non-BEAR IPOs is more than double of that for BEAR IPOs (9.4% versus 3.7%). However, those results might be attributable to the fact that BEAR IPOs could not raise large funds due to poor market condition. To address the concern, we examine the proportion of retained cash over proceeds by IPO (Kim and Weisbach, 2008). Table 4 suggests that the median BEAR IPOs hold approximately 43% of IPO proceeds at the end of IPO year while the median Non-BEAR IPOs save about 64% of its proceeds. Although there is no significant difference in the change from Year -1 to Year 1, the presented result is consistent with our premise that BEAR IPOs have weaker incentives to raise lump sum funds by IPO for future investment projects. [Insert Table 4 about here] Given that IPOs in bear markets should be also disadvantageous for secondary offerings, Table 2 presents the secondary proceeds over assets at the year before IPO. The result clearly suggests that BEAR IPOs have significantly smaller secondary proceeds than Non-BEAR IPOs do. Untabulated regression analyses also present the qualitatively same result after controlling for various factors. This finding is consistent with Stoughton et al. (2001), which indicates that IPOs motivated by product market benefits are characterized by smaller fractions of insiders’ cash-out at 14

the time of IPO. Overall, BEAR IPOs receive smaller benefits from both primary and secondary equity offerings. BEAR IPOs are less likely to choose good time for equity financing, to the degree that they put relatively low priority on equity issues. It is well-documented that US firms experience long-term stock price underperformance during the post-SEO period since they conduct SEOs when their stocks are overvalued (Loughran and Ritter, 1995; Spiess and Affleck-Graves, 1995). To examine whether BEAR IPOs are less overvalued at the time of IPO, we investigate long-term stock price performance. In this analysis, control firm is selected from listed companies in same industry whose size (market value of equity) ranges between 70% and 130% of the sample firm in IPO year. For those companies, a firm which is closest in the M/B ratio (market value of equity scaled by the book value of equity at the IPO year) is adopted as a control firm. Control firms are also required not to issue new shares during three years ending at the IPO year of the sample firm. The industry average is spliced when a matching firm is delisted (for the delisting year and onward). Table 5 clearly indicates that BEAR IPOs significantly outperform Non-BEAR IPOs. This result suggests that Non-BEAR IPOs are more overvalued at the time of IPO than BEAR IPOs. Consistent with our view, BEAR IPOs do not time the market for equity financing. [Insert Table 5 about here] 4.3 Post-IPO sales growth Hypothesis 1 predicts that firms from competitive and homogeneous industries in production costs are likely to go public for increased visibility and market shares. To further examine the hypothesis, Table 6 reports firms’ sales growth from Year -1 ((concurrent year sales – sales for the year before IPO)/ sales for the year before IPO). Panel A of Table 6 indicates that the median BEAR IPO achieves 56.8% sales increases from Year -1 to Year 5. To control for the macro-economic and industry conditions, we also present the adjusted value, which is computed by subtracting the industry median or the value of the control firm from the sample firm’s value. For each of sample companies, the listed firm which is closest in sales for Year -1 is employed as a control firm. Control firms are also required not to conduct IPO and SEOs during five years ending at the IPO year of the sample firm. When a matching firm is delisted, we splice the industry average (for the delisting year and onward) with the controlling firm data to avoid sample size reduction due to lack of control firm data. Panels B and C indicate that BEAR IPOs achieve significantly greater sales growth from the pre-IPO level than do their industry peers. Besides, BEAR IPOs achieve significantly larger sales growth at Year 3 and afterwards than do Non-BEAR IPOs. 15

Although we do not find significant differences in the sales growth immediately after the IPO, those results suggest that BEAR IPOs can achieve long-term increases of market share, which is consistent with Hypothesis 1. We also examine post-IPO ROA of sample firms. Untablulated results suggest that BEAR IPOs significantly outperform BEAR IPOs in control firm-adjusted ROA for Year 2 and afterwards. [Insert Table 6 about here] 4.4 Do BEAR IPOs have stable and timely financing sources? Hypothesis 2 assumes that firms relying on bank debt put low priority on equity financing at the IPO since those firms have stable and timely financing sources. If this presumption is the case, BEAR IPOs should show less financing constraints during the post-IPO period. Hoshi et al. (1991) show evidence that firms affiliated with a large Japanese bank have significantly smaller investment-to-cash flow sensitivities than do unaffiliated firms. Previous studies also investigate the investment-to-Tobin's Q sensitivity to examine firms' investment efficiencies (Chen et al. 2007; Jiang et al., 2011). Given that the finance literature commonly employs Tobin's Q as a proxy for investment opportunities, high sensitivity of capital expenditures to Tobin's Q represents efficient investment decisions. Since investment efficiency should be affected by availability of timely financing, our idea also predicts that BEAR IPOs show greater investment-to-Tobin’s Q sensitivities than Non-BEAR IPOs do. We run regression of firms' investments by using data during five years following IPO (we exclude firms from the analysis, for which less than 3 years data are available). The dependent variable is computed by the change of PPE plus depreciation scaled by book value of assets at the previous year (CAPX). Cash flow is calculated as net income plus depreciation scaled by the one-year lagged assets (CASHFLOW). Tobin's Q is defined as the market value of equity plus book value of total liabilities divided by the book value of asset (one-year lagged data are used). To test our prediction, the regression includes the interaction terms of BEARIPOD with CASHFLOW and Tobin's Q. All estimations adopt firm-fixed effects models with year dummies. Panel A of Table 7 presents CAPX by relative year to IPO, suggesting that sample companies undertake fixed investments which amount to about 5% of existing assets. Non-BEAR IPOs conduct significantly more fixed investments at the year after IPO (Year 1), but the significant difference disappears at Year 3 and after. Panel B of Table 7 suggests that the median CAPX is much smaller than median CASHFLOW (2.9% versus 5.7%). IPO companies tend to conduct capital investments within their cash flow levels. Panel C of Table 7 presents regression results. Consistent with previous studies, 16

both CASHFLOW and Tobin’s Q have a positive and significant coefficient. Importantly, Models (1) and (3) engender a negative and significant coefficient on the interaction term of CASHFLOW and BEARIPOD, suggesting that post-IPO investments of BEAR IPOs are less susceptible to the level of cash flow compared to those of Non-BEAR IPOs. Noticeably, the estimated coefficients in Model (1) suggest that the investment-to-cash flow sensitivity for BEAR IPOs is 0.028 whereas it is 0.093 for Non-BEAR IPOs. In conjunction with the logit regression result, we argue that firms going public in bear markets have stable access to bank financing. Availability of stable financing source should enable efficient investment decisions. Indeed, Models (2) and (3) of Table 7 engenders a positive and significant coefficient on the interaction term of Tobin’s Q and BEARIPOD. The estimated coefficients suggest economically significant difference exists in the investment efficiency between the two subsamples. Specifically, the investment-to-Tobin's Q sensitivity for BEAR IPOs is double of the sensitivity for Non-BEAR IPOs (0.011 versus 0.005 in model (3)). [Insert Table 7 about here] As an additional analysis, we simultaneously include concurrent and one-year lagged Tobin’s Q to address the concern that CASHFLOW may contain information about future investment opportunities which is not incorporated in one-year lagged Tobin’s Q (Hoshi et al., 1991). The additional analysis generates the qualitatively same results. Overall, our analyses show robust evidence that firms going public in bear markets have access to stable financing sources. The result supports our view that those firms put relatively low priority on fund raising as a motive of IPO. 4.5 Additional analyses We have argued that firms with relatively low priority on equity financing are willing to go public even in bear markets. A potential alternative explanation for bear market IPOs is that firms with near-term cash needs go public even in bear markets. Investigating US SEOs, DeAngelo et al. (2010) show evidence that around 60% of SEO firms would have fallen into negative cash holdings without SEO proceeds. We follow DeAngelo et al. (2010) to examine pro forma cash, which is hypothetical cash holdings when we assume that IPO firms received no primary proceeds with holding all other variables unchanged. Table 8 suggests that both BEAR and Non-BEAR IPOs have pro forma cash greater than 10% of the assets at the end of IPO year. Besides, the median pro forma cash (scaled by assets) is significantly greater for BEAR IPOs at Year 1 and 2 than for Non-BEAR IPOs. The proportion of negative pro forma cash is also smaller for BEAR IPOs (the difference is statistically significant for Year 1 and 2). Those results do not support the idea that firms with urgent cash needs go 17

public in bear markets. [Insert Table 8 about here] Another possible explanation for bear market IPOs is that they have urgent needs to finance large projects. If this explanation is true, BEAR IPOs should conduct substantial capital expenditures immediately after the IPO. However, Panel A of Table 7 shows that BEAR IPOs conduct less fixed investments (CAPX) at Year 1 than Non-BEAR IPOs do, and no significant difference exists in the subsequent years. We also find that the change in capital expenditures surrounding IPO is greater for Non-BEAR IPOs than for BEAR IPOs. In sum, we cannot support the alternative explanation that BEAR IPOs go public in bear markets because they have urgent financing needs for investment projects. One can raise a concern that our result is attributable to the fact that firms going public in bear markets had to rely on bank debt due to poor market condition. As mentioned, however, results are qualitatively unchanged even when we define BEAR IPOs by using TOPIX return in shorter (3-month) period, which should decrease the number of firms that began to heavily rely on bank debt due to poor index return (remember that we use financial data at the year before IPO to compute financial variables). Another potential concern is that banks keep relations with private companies with high creditor value whereas those companies do not need to care on market conditions. To address the endogeneity concern, we implement a two-stage probit regression, which treats LOANDR as an endogenous variable, by using lagged LOANDR as an instrument variable. This analysis also suggests that firms relying on bank debt are more likely to go public in bear markets. Table 1 indicates that BEAR IPOs cluster around 2000, during which the Japanese banking sector suffered from non-performing loan problems and contracted lending. To test whether our finding is attributable to the banking crisis, we rerun the logit regressions of Table 3 by including ROA and BIS capital ratio of the firm’s main bank. Regression results carry an insignificant coefficient on those variables, while our main results remain unchanged. 5. Conclusion Non-trivial number of firms go public in bear markets. Indeed, approximately one-fourth of Japanese IPO firms during the past 17 years went public after the market index significantly declined. This paper examines characteristics of those companies to highlight an aspect of IPO, which has not been well-documented in the literature. We find that firms from industries characterized by keen competition and small 18

variation in production costs are more likely to go public in bear markets than those in less competitive industries which consist of heterogeneous companies in production costs. We also find that firms relying on bank finance are more likely to go public in bear markets than those with outstanding bonds. Firms going public in bear markets raise significantly smaller proceeds, sell stocks for smaller prices, and stockpile cash less than other IPO firms do. IPO firms in bear markets also show significantly superior long-term stock price performance during the post-IPO period. Those results suggest that private firms which put a relatively low weight on equity financing as a motive of IPO are willing to go public even in bear markets. We also find that firms going public in bear markets achieve significantly greater sales increases from the pre-IPO level than do other IPOs for a few years after the IPO. Also, bear market IPO firms suffer less from financing constraints, and make more efficient investment decisions. Those results indicate that bear IPO firms put a relatively low priority on equity financing at the time of IPO since they go public to increase market shares and have access to stable and timely financing sources. This paper makes significant contributions to the literature. To the best of our knowledge, this is the first research to examine characteristics of companies going public in bear markets. Differently from previous studies on hot IPO (Chemmanur and He, 2011; Lowry and Schwert, 2002; Pastor and Veronesi, 2005; Ritter, 1984), our research highlights product market conditions and reliance on bank finance as a determinant of firms’ IPO timing. We also show new evidence that stable access to bank finance significantly decreases firms’ incentives to go public in good market conditions. It would be also a novel finding that market condition at the IPO is correlated with financing constraints, investment efficiency, and long-term performance during the post-IPO period. References Alti, A., 2005. IPO market timing. Review of Financial Studies 18, 1105-1138. Baker, M., and J. Wurgler. 2002. Market Timing and Capital Structure. Journal of Finance 57, 1–32. Berger, A. N., Udell, G. F., 1995. Relationship lending and lines of credit in small firm finance. Journal of Business 68, 351-382. Berger, A. N., Udell, G. F., 2002. Small business credit availability and relationship lending: The importance of bank organizational structure. The Economic Journal 477, 19

32-53. Brau, J. C., Fawcett, S. E., 2006. Initial public offerings: An analysis of theory and practice. Journal of Finance 61, 399-436. Busaba, W. Y., Benveniste, L. M., Guo, R.-J., 2001. The option to withdraw IPOs during the premarket: Empirical analysis. Journal of Financial Economics 60, 73–102. Chemmanur, T. J., and He, J., 2010. IPO waves, product market competition, and the going public decision: Theory and evidence. Journal of Financial Economics 101, 382-412. Chen, Q., Goldstein, I., Jiang, W., 2007. Price informativeness and investment sensitivity to stock price. Review of Financial Studies 20, 619-650. Cornett, M. M., McNutt, J. J., Strahan, P. E., Tehranian, H., 2011. Liquidity risk management and credit supply in the financial crisis. Journal of Financial Economics 101, 297-312. Dunbar, C. G., Foerster, S. R., 2008. Second time lucky? Withdrawn IPOs that return to the market. Journal of Financial Economics 87, 610-635. DeAngelo, H., DeAngelo, L., Stulz, R. M., 2010. Seasoned equity offerings, market timing, and the corporate lifecycle. Journal of Financial Economics 95, 275-295. Graham, J. R., Harvey, C. R., 2001. The theory and practice of corporate finance: evidence from the field. Journal of Financial Economics 60, 187-243. Gilson, S., John, K., Lang, L., 1990. Troubled debt restructuring: An empirical study of private renegotiations of firms in default. Journal of Financial Economics 27, 315-353. Hoshi, T., Kashyap, A., Scharfstein, D., 1990. The role of banks in reducing the costs of financial distress in Japan. Journal of Financial Economics 27, 67-88. Hoshi, T., Kashyap, A., Scharfstein, D., 1991. Corporate structure, liquidity, and investment: Evidence from Japanese industrial groups. Quarterly Journal of Economics 106, 33-60. 20

Helwege, J., Liang, N., 2004. Initial public offerings in hot and cold markets. Journal of Financial and Quantitative Analysis 39, 541-569. Hellmann, T., Lindsey, L., Puri, M., 2008. Building relationships early: banks in venture capital. Review of Financial Studies 21, 513-541. Hsieh, J., Lyandres, E., Zhdanov, A., 2011. A theory of merger-driven IPOs. Journal of Financial and Quantitative Analysis 46, 1367-1405. Ivashina, V., Scharfstein, D., 2010. Bank lending during the financial crisis of 2008. Journal of Financial Economics 97, 319-338. Jiang, L., Kim, J.-B., Pang, L., 2011. Control-ownership wedge and investment sensitivity to stock rice. Journal of Banking and Finance 35, 2856-2867. Kaplan, S. N., Minton, B. A., 1994. Appointments of outsiders to Japanese boards: Determinants and implications for managers. Journal of Financial Economics 36,225-258. Kim, M., Ritter, J.,1999. Valuing IPOs. Journal of Financial Economics 53, 409–437. Kang, J.-K., Stulz, R. M., 2000. Do banking shocks affect borrowing firm performance? An analysis of the Japanese experience. Journal of Business 73, 1-23. Kang J.-K., Shivdasani, A., Yamada T., 2000. The effect of bank relations on investment decisions: An investigation of Japanese takeover bids. Journal of Finance 55: 2197-2218. Kaplan, S., Stromberg, P., 2003. Financial contracting theory meets the real world: Evidence from venture capital contracts. Review of Economic Studies 70, 281-315. Kim, W., Weisbach, M., 2008. Motivations for public equity offers: An international perspective. Journal of Financial Economics 87, 281-307. Lerner, J., 1994. Venture capitalists and the decision to go public. Journal of Financial Economics 35, 293-316.

21

Loughran, T., Ritter, J., Rydqvist, K., 1994. Initial public offerings: International insights. Pacific-Basin Finance Journal 2, 165-199. Loughran, T., Ritter, J., 1995. The new issues puzzle. Journal of Finance 50, 23-51. Lowry, M., Schwert, G. W., 2002. IPO market cycles: Bubbles or sequential learning? Journal of Finance 57, 1171-1200. Morck, R., Nakamura, M., 1999. Banks and corporate control in Japan. Journal of Finance 54, 319–339. Pagano, M., Panetta, F., Zingales, L., 1998. Why do companies go public? An empirical analysis. Journal of Finance 53, 27-64. Pinkowitz, L., Williamson, R., 2001. Bank power and cash holdings: Evidence from Japan. Review of Financial Studies 14, 1059-1082. Pastor, L., Veronesi, P., 2005. Rational IPO waves. Journal of Finance 60, 1713-1757. Ritter, J. R., 1984. The 'hot issue' market of 1980. Journal of Business 32, 215-240. Spiess, D. K., Affleck-Graves, J., 1995. Underperformance in long-run stock returns following seasoned equity offerings. Journal of Financial Economics 38, 243–267. Stoughton, N. M., Wong, K.-P., Zechner, J., 2001. IPOs and product quality. Journal of Business 74, 375-408. Sun, Y., Uchida, K., 2016. The role of bank-affiliated venture capital for parent banks in Japan: New evidence. Asia-Pacific Journal of Financial Studies, forthcoming. Takahashi, H., 2015. Dynamics of bank relationships in entrepreneurial finance. Journal of Corporate Finance 34, 23-31. Yung, C., Çolak, G., Wang, W., 2008. Cycles in the IPO market. Journal of Financial Economics 89, 192-208.

22

Appendix Variable definition Variable BEARIPOD HHI

SD_Margin LOANDR LOANAR BONDAR BONDD BONDHIGHD BANKOWN MBOWN MBVCOWN BDIRECD Ln(Assets) Firm age ROA SGR LEVERAGE CAPX CASHFLOW Tobin's Q Ln_TOPIX

Definition Dummy variable which takes on a value of one for BEAR IPOs and zero for Non-BEAR IPOs. Herfindahl-Hirschman index (HHI) index is calculated by squaring the market share of each firm competing in an industry and in a given year, and then summing the resulting numbers by industry and year. Standard deviation of gross margin by industry, where gross margin is calculated as gross profit divided by sales. The ratio of total bank loan (including both short term and long term) to total debt The ratio of total bank loan (including both short term and long term) to total asset Total issuance of bond, convertible bond and commercial paper, divided by total asset A dummy variable takes a value of one for firms with positive BONDAR A dummy variable which takes a value of one for firms with BONDAR greater than 10% and zero otherwise Percentage ownership by banks, which have business relationships with the firm in the IPO White Paper Percentage ownership by main bank Total percentage ownership by main bank and main bank affiliated venture capitalist A dummy variable indicating existence of directors appointed from banks is also adopted Natural logarithm of the book value of total asset Firm age at time of IPO Operating income divided by total assets Percentage sales growth ratio from previous year Total liability, divided by total asset The change of PPE plus depreciation divided by one-year lagged assets Cash flow, which is calculated as net income plus depreciation, divided by one-year lagged total assets Market value of equity plus book value of total liabilities divided by the book value of assets Natural logarithm of TOPIX, stock price index of the Tokyo Stock Exchange

23

Table 1 Sample year distribution This table indicates year distribution of our sample IPOs. The table also indicates the market index (TOPIX) at the end of corresponding year. Year 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Total

BEAR IPOs 40 41 9 68 97 66 46 7 0 0 15 33 7 4 17 8 0 0 458

Non-BEAR IPOs 95 40 94 122 67 48 70 154 151 176 99 14 12 15 18 36 52 72 1335

24

TOPIX (closing price) 1147.87 1088.83 1712.27 1291.65 1013.73 849.25 1026.24 1139.41 1663.75 1678.91 1499.94 854.44 907.59 898.80 728.61 859.80 1302.29 1407.51

Table 2 Summary statistics This table reports summary statistics. Panel A presents mean (median in the bracket) of continuous variables. Data preceding IPO are used for financial variables while ownership data are obtained from IPO White Papers. All continuous variables are winsorized at the top and bottom one percent values. P-values in Panel A are for mean (median) difference test between BEAR and Non-BEAR IPOs. Panel B presents the proportion of observations which take a value of one for the dummy variable. P-values in Panel B are for the proportion difference test. See Appendix for definition of variable. BEAR IPOs Panel A: Non-dummy variables HHI SD_Margin LOANDR BONDAR BANKOWN MBOWN Initial return Ln(Assets) Firm age ROA SGR LEVERAGE Primary proceeds / Total assets at the year before IPO Secondary proceeds / Total assets at the year before IPO Offering price / Operating Income Panel B: Dummy variables BDIRECD Dummy for MBOWN > 0 Dummy for MBVCOWN > 0

Non-BEAR IPOs

P-value

0.118 [0.053] N=455 0.137 [0.147] N=457 0.802 [1.000] N =455 0.009 [0.000] N=455 2.550 [0.810] N=458 1.156 [0.380] N=457 0.365 [0.104] N=447 8.746 [8.723] N=455 23 [20] N=428 0.114 [0.097] N=455 0.421 [0.165] N=449 0.600 [0.645] N=455 0.250 [0.087] N=445 0.156 [0.054] N=445 15.540 [4.845] N=429

0.193 [0.071] N=1323 0.137 [0.145] N=1334 0.743 [1.000] N =1323 0.016 [0.000] N=1320 1.943 [0.000] N=1335 0.952 [0.000] N=1334 0.700 [0.319] N=1322 8.570 [8.451] N=1320 22 [17] N=1269 0.124 [0.103] N=1318 0.430 [0.173] N=1312 0.598 [0.628] N=1320 0.418 [0.152] N=1303 0.262 [0.091] N=1303 22.980 [7.532] N=1264

0.000*** [0.000***] 0.858 [0.093*] 0.006*** [0.002***] 0.000*** [0.049**] 0.000*** [0.000***] 0.008*** [0.002***] 0.000*** [0.000***] 0.023** [0.012***] 0.106 [0.055*] 0.075* [0.032**] 0.848 [0.514] 0.824 [0.705] 0.000*** [0.000***] 0.022** [0.000***] 0.049** [0.000***]

0.080 N=287 0.527 N=457 0.661 N=457

0.044 N=942 0.441 N=1334 0.581 N=1334

0.015**

***: Significant at the 1% level; **: Significant at the 5% level; Significant at the 10% level

25

0.001*** 0.003***

Table 3 Logit regression results This table shows logistic regression results. Panel A uses HHI (Herfindahl-Hirschman index index) as an industry characteristic variable while Panel B uses SD_Margin, which is the standard deviation of the gross margin in the industry. The dependent variable (BEARIPOD) takes a value of one for BEAR IPOs and zero for Non-BEAR IPOs. Data preceding IPO are used for most of independent variables while ownership data are obtained from IPO White Papers. All continuous variables are winsorized at the top and bottom one percent values. Z-statistics computed by using heteroskedasiticity-consistent standard errors are in parentheses. All estimations include industry dummies (not-reported). See Appendix for definition of variable. Model (1) Model (2) Panel A: Use HHI for industry characteristics HHI -27.662*** -26.417*** (-3.07) (-2.94) LOANDR 0.584*** (2.93) BONDAR -9.582*** (-2.94) BONDD 0.143 (0.67) LOANAR 0.225 (0.60) BONDHIGHD

Model (3)

Model (4)

Model (5)

-26.758*** (-2.96)

-24.578*** (-2.68)

-16.130 (-1.51) 0.485** (2.23) -5.159 (-1.35) -0.059 (-0.21)

0.069*** (3.13)

BDIRECD

Ln(Assets) Firm age ROA SGR Ln_TOPIX Constant Pseudo R2 N

0.458 (1.37) -0.510 (-1.30) 0.116** (2.01) 0.000 (0.04) -0.163 (-0.22) 0.053 (0.75) -4.167*** (-14.61) 39.570*** (8.36) 0.162 1649

0.104* (1.83) 0.002 (0.41) -0.464 (-0.61) 0.041 (0.59) -4.180*** (-14.35) 39.536*** (8.28) 0.165 1649

0.762*** (3.16)

0.185 (0.50) -1.077*** (-2.71)

BANKOWN

LEVERAGE

Model (6)

0.109* (1.94) 0.001 (0.30) -0.389 (-0.52) 0.042 (0.59) -4.175*** (-14.45) 39.607*** (8.31) 0.162 1649

26

-0.087 (-0.24) 0.109* (1.86) -0.003 (-0.65) -0.283 (-0.39) 0.051 (0.73) -4.239*** (-4.47) 39.178*** (7.75) 0.162 1649

0.040 (0.56) 0.000 (0.08) 0.723 (0.81) -0.011 (-0.14) -5.060*** (-12.07) 40.362*** (7.32) 0.219 1181

0.080** (2.28) 0.401 (1.24) -1.235** (-2.50) 0.079 (1.07) -0.006 (-0.96) 0.847 (0.96) -0.016 (-0.19) -5.066*** (-12.09) 40.075*** (7.41) 0.225 1181

Table 3 (Continued) Model (1) Model (2) Model (3) Panel B: Use SD_Margin as industry characteristics SD_Magrin -18.387*** -17.494*** -17.479*** (-2.80) (-2.63) (-2.64) LOANDR 0.585*** (2.94) BONDAR -9.872*** (-3.01) BONDD 0.169 (0.78) LOANAR 0.206 0.168 (0.56) (0.46) BONDHIGHD -1.078*** (-2.74) BANKOWN BDIRECD

Model (4)

Model (5)

Model (6)

-18.156*** (-2.76)

-13.516 (-1.64) 0.466** (2.15) -5.150 (-1.35) -0.048 (-0.17)

-15.284* (-1.86) 0.752*** (3.13)

0.076*** (3.46) 0.427 (1.24)

LEVERAGE

-0.565 -0.144 (-1.44) (-0.39) Ln(Assets) 0.131** 0.115** 0.121** 0.120** 0.049 (2.31) (2.09) (2.21) (2.09) (0.70) Firm age -0.000 0.002 0.001 -0.004 0.000 (-0.04) (0.33) (0.23) (-0.77) (0.04) ROA -0.239 -0.517 -0.441 -0.350 0.599 (-0.33) (-0.69) (-0.59) (-0.48) (0.67) SGR 0.058 0.046 0.047 0.056 -0.005 (0.83) (0.68) (0.68) (0.82) (-0.07) Ln_TOPIX -4.092*** -4.106*** -4.102*** -4.183*** -5.031*** (-14.36) (-14.16) (-14.24) (-14.24) (-11.98) Constant 30.177*** 30.537*** 30.465*** 31.183*** 35.670*** (11.97) (11.92) (11.97) (12.14) (10.51) Pseudo R2 0.161 0.165 0.161 0.163 0.220 N 1649 1649 1649 1649 1181 ***: Significant at the 1% level; **: Significant at the 5% level; Significant at the 10% level

27

0.081** (2.28) 0.363 (1.08) -1.294*** (-2.63) 0.092 (1.24) -0.006 (-0.99) 0.701 (0.79) -0.011 (-0.13) -5.039*** (-11.99) 36.037*** (10.59) 0.226 1181

Table 4 Cash holdings This table presents mean (median in brackets) of cash holdings variables. Year 0 indicates the IPO year. The most right-hand column presents p-values for mean difference test (median difference test in brackets). All variables are winsorized at the top and bottom one percent values. BEAR IPOs

Non-BEAR IPOs

Cash and cash equivalents over total assets 0.254 [0.201] 0.260 [0.213] Year -1 N=455 N=1319 0.255 [0.196] 0.288 [0.240] Year 0 N=480 N=1377 0.238 [0.184] 0.246 [0.198] Year 1 N=472 N=1305 Change of cash and cash equivalents over total assets at the year before IPO 0.159 [0.037] 0.315 [0.094] Year[-1, 0] N=454 N=1319 0.220 [0.043] 0.383 [0.071] Year[-1, 1] N=446 N=1248 Change of cash and equivalents over proceeds 0.507 [0.426] 0.720 [0.638] Year[-1, 0] N=437 N=1281 0.903 [0.467] 0.875 [0.530] Year[-1, 1] N=429 N=1212

P-value 0.529 [0.400] 0.001*** [0.001***] 0.370 [0.402]

0.000*** [0.000***] 0.000*** [0.000***]

0.003*** [0.000***] 0.816 [0.226]

***: Significant at the 1% level; **: Significant at the 5% level; Significant at the 10% level

28

Table 5 Post-IPO stock return This table indicates buy-and-hold returns (BHRs) during the post-IPO period. The adjusted BHR is the sample firm’s BHR less control firm’s BHR. Control firms are selected from listed companies in same industry whose size (market value of equity) ranges between 70% and 130% of the sample firm in IPO year. For those companies, a firm which is closest in the M/B ratio (market value of equity scaled by the book value of equity at the IPO year) is adopted as a control firm. Control firms are also required not to issue new shares during three years ending at the IPO year of the sample firm. The industry average is spliced when a matching firm is delisted (for the delisting year and onward). All variables are winsorized at the top and bottom one percent values.

Raw BEAR IPOs Non-BEAR IPOs P-value (Diff. test) Adjusted BEAR IPOs Non-BEAR IPOs P-value (Diff. test)

1-year BHR

2-year BHR

3-year BHR

0.069 [-0.152] N=483 -0.251 [-0.402] N=1333 0.000***[0.000]***

0.026 [-0.220] N=483 -0.304 [-0.548] N=1333 0.000***[0.000]***

0.053 [-0.240] N=483 -0.411 [-0.626] N=1333 0.000***[0.000***]

0.028 [-0.085] N=458 -0.172 [-0.220] N=1283 0.000***[0.000***]

-0.074 [-0.073] N=458 -0.297 [-0.249] N=1283 0.000***[0.000***]

-0.085 [-0.125] N=458 -0.234 [-0.246] N=1283 0.006***[0.000***]

***: Significant at the 1% level; **: Significant at the 5% level; Significant at the 10% level

29

Table 6 Post-IPO sales growth This table reports mean (median in brackets) of sales growth ratio from the Year -1 ((concurrent year sales – sales for Year -1) / sales for Year -1). The ratio is winsorized at the top and bottom one percent values. Year -1 is the year before IPO. Adjusted values are computed by subtracting the industry median or the value of control firm from the sample firm’s value. For each of sample companies, control firms are selected from listed companies in the same industry which is closest in sales for Year -1. Controlling firms are also required not to conduct IPO and SEOs during five years ending at the IPO year of the sample firm. P-values are for the null hypothesis that the mean (median) is zero. P-values (Diff. test) are for mean difference test (median difference test in brackets) between BEAR IPOs and Non-BEAR IPOs. Year 0 Panel A: Raw sales growth BEAR IPOs 0.274 [0.123] P-value 0.000***[0.000***] N 456 Non-BEAR IPOs 0.266 [0.148] P-value 0.000***[0.000***] N 1316 P-value (Diff. test) 0.727 [0.080*]

Year 1

Year 2

Year 3

Year 4

Year 5

0.453 [0.181]

0.732 [0.239]

1.104 [0.335]

1.451 [0.422]

1.800 [0.568]

0.000***[0.000***]

0.000***[0.000***]

0.000***[0.000***]

0.000***[0.000***]

0.000***[0.000***]

451 0.493 [0.249]

444 0.738 [0.301]

427 0.933 [0.348]

401 1.129 [0.384]

393 1.323 [0.419]

0.000***[0.000***]

0.000***[0.000***]

0.000***[0.000***]

0.000***[0.000***]

0.000***[0.000***]

1247

1186

1116

1061

1015

0.365 [0.004***]

0.931 [0.070*]

0.114 [0.449]

0.026**[0.122]

0.009*** [0.003***]

0.724 [0.232]

1.081 [0.315]

1.406 [0.364]

1.743 [0.521]

0.000***[0.000***]

0.000***[0.000***]

0.000***[0.000***]

0.000***[0.000***]

444 0.702 [0.279]

427 0.902 [0.317]

401 1.104 [0.358]

393 1.300 [0.388]

0.000***[0.000***]

0.000***[0.000***]

0.000***[0.000***]

0.000***[0.000***]

1186

1116

1061

1015

0.766 [0.436]

0.093*[0.277]

0.034** [0.256]

0.014**[0.018**]

0.544 [0.221]

0.793 [0.236]

1.187 [0.293]

1.307 [0.327]

0.000***[0.000***]

0.000***[0.000***]

0.000***[0.000***]

0.000***[0.000***]

442 0.279 [0.206]

426 0.373 [0.185]

403 0.432 [0.180]

390 0.356 [0.205]

0.000***[0.000***]

0.000***[0.000***]

0.000***[0.000***]

0.000***[0.000***]

1188

1110

1061

1014

0.011** [0.188]

0.014** [0.062*]

0.007*** [0.022***]

0.001*** [0.016**]

Panel B: Industry-median adjusted sales growth BEAR IPOs 0.269 [0.123] 0.455 [0.184] P-value 0.000***[0.000***] 0.000***[0.000***] N 456 451 Non-BEAR IPOs 0.248 [0.131] 0.460 [0.213] P-value 0.000***[0.000***] 0.000***[0.000***] N 1316 1247 P-value (Diff. test) 0.365 [0.647] 0.897 [0.276] Panel C: Control firm-Adjusted sales growth BEAR IPOs 0.245 [0.125] 0.330 [0.142] P-value 0.000***[0.000***] 0.000***[0.000***] N 454 451 Non-BEAR IPOs 0.186 [0.112] 0.243 [0.142] P-value 0.000***[0.000***] 0.000***[0.000***] N 1315 1250 P-value (Diff. test) 0.088* [0.193] 0.125 [0.894]

***: Significant at the 1% level; **: Significant at the 5% level; Significant at the 10% level

30

Table 7 Regression of capital expenditures Panel A of this table presents CAPX (change of PPE plus depreciation expenses over one-year lagged assets) of sample companies by year (we delete firms for which less than 3 years post-IPO data are available). Year 1 indicates the year after the IPO. Panel B shows descriptive statistics of variables during the five year following the IPO (Year 1 to Year 5). Panel C presents regression results of CAPX. The analysis uses data during five years following the IPO. Tobin’s Q is computed by the market value of equity plus book value of total liability divided by the book value of total asset (one-year lagged data are used for estimation). CASHFLOW is calculated as net income plus depreciation, divided by one-year lagged assets; Firm-fix effects models with year dummies are used for estimation. BEARIPOD takes a value of one for BEAR IPOs and zero for Non-BEAR IPOs. All the continuous variables are winsorized at the top and bottom one percent values. Heteroskedasticity-consistent standard errors are shown in parentheses. Panel A: CAPX Year 1 BEAR IPOs 0.051 [0.028] N 425 Non-BEAR IPOs 0.065 [0.038] N 1103 P-value (Diff. 0.002***[0.000***] test) Panel B: Descriptive statistics Mean CAPX 0.050 Tobin’s Q 1.407 CASHFLOW 0.056 BEARIPOD 0.280

Year 2 0.049 [0.027] 432 0.055 [0.032] 1107 0.103 [0.046**]

Year 3 0.049 [0.030] 437 0.048 [0.027] 1124 0.800 [0.744]

Year 4 0.046 [0.025] 405 0.044 [0.026] 1056 0.584 [0.948]

Year 5 0.038 [0.025] 395 0.042 [0.025] 997 0.304 [0.395]

SD 0.066 0.947 0.080 0.449

Minimum -0.070 0.474 -0.201 0

Median 0.029 1.079 0.057 0

Max 0.311 5.170 0.249 1

Panel C: Regression results

CASHLOW CASHLOW*BEARIPOD

Model (1)

Model (2)

Model (3)

0.093*** (6.06) -0.065** (-2.28) 0.007*** (5.49)

0.075*** (5.72)

0.096*** (6.23) -0.075*** (-2.60) 0.005*** (3.50) 0.006** (2.30) 0.051*** (4.69)

Constant

0.020 (1.64)

0.006*** (3.72) 0.005* (1.93) 0.050*** (4.61)

R2

0.041

0.041

0.042

N

5889

5889

5889

Tobin’s Q Tobin’s Q*BEARIPOD

***: Significant at the 1% level; **: Significant at the 5% level; Significant at the 10% level

31

Table 8 This table presents mean (median in brackets) of the hypothetical cash holdings (pro forma Cash), which is calculated under the assumption that sample firms received no IPO proceeds. Percentage values in parentheses present the percentage of observations which have negative value. Year 0 indicates the firm’s IPO year. P-value (Diff. test) is for mean difference test (median difference test in brackets) between BEAR IPOs and Non-BEAR IPOs. P-value (Proportion diff. test) is for the null hypothesis that the proportion of observations which have negative value is identical between BEAR and Non-BEAR IPOs. Pro forma cash variables are winsorized at the top and bottom 1% values. Year 0 Year 1 0.156 [0.127] 0.147 [0.122] N= 480 (Negative: 14.8%) N=472 (Negative: 16.4%) Non-BEAR IPOs 0.167 [0.131] 0.130 [0.102] N= 1377 (Negative: 17.7%) N=1305 (Negative: 23.8%) P-value (Diff. test) 0.315 [0.653] 0.109 [0.020**] P-value (Proportion diff. test) 0.153 0.001*** ***: Significant at the 1% level; **: Significant at the 5% level; Significant at the 10% level Bear IPOs

32

Year 2 0.154 [0.122] N=467 (Negative: 15.0%) 0.130 [0.100] N=1243 (Negative: 22.7%) 0.024** [0.002***] 0.000***