IPO Underpricing and Information Disclosure Laura Bottazzi (Bologna and IGIER) Marco Da Rin (Tilburg, ECGI, and IGIER)
!! Work in Progress!!
Motivation IPO underpricing (UP) is a pervasive feature of primary equity markets. It has implications for investors, companies, and regulators. It is the object of a large literature, both theoretical and empirical.
Motivation Asymmetric information is at the core of all theories of IPO underpricing. Asymmetric information may exist among all market participants: Founders (sellers) may know more about the firm than investors (buyers), leading to a lemons problem. UP is then a credible signal of quality.
Motivation Some investors may know more about the firm than the others, leading to a winner s curse. UP then induces participation by the uninformed. In the same situation, the repeated interaction of underwriting banks with institutional investors through the bookbuilding process may generate UP as a compensation to informed investors revelation of information.
Motivation If it is the underwriting bank to be better informed, and if its marketing efforts are non contractible, then UP provides a viable second best solution by compensating the underwriter. If investors care about liquidity in the secondary market, then UP is a compensation for the risk of future illiquidity. Such risk depends on post-ipo asymmetric information.
Motivation Despite its central role in explaining UP, asymmetric information is very difficult to measure. Typically, studies resort to indirect proxies: Company characteristics: age, size, sector, venture backing or auditor reputation. Issue characteristics: bookbuilding, underwriter reputation, insiders secondary sales After-market characteristics: return volatility
Our contribution We provide a direct measure of asymmetric information, which is gathered from the IPO prospectus. We build an index of the quantity of information provided in the prospectus. This is information which is provided voluntarily, not mandated by law. We measure voluntary information disclosure through a direct measure gathered from the IPO prospectus.
Our contribution We find three main (preliminary) results: Disclosure negatively affects UP. The effect of disclosure is asymmetric: low disclosure increases UP, high disclosure has virtually no effect. The effect of disclosure appears to be robust to controls for endogeneity of the regressors.
The Data Hand-collected database with information from IPO prospectuses of all 515 companies which went public on Neuer Markt, Nouveau Marché, Nuovo Mercato until the end of 2001. Disclosure Index Balance sheet information Ownership structure pre- and post-ipo Pricing mechanism, greenshoe option, lead underwriter, primary and secondary shares sold Sector of activity
The Data From stock exchanges we obtain: Date of listing Issue price First closing price Closing prices in first 12 weeks of trading Seasoned equity offerings
The Disclosure Index The DI is modeled on similar metrics for corporate governance quality: S&P Transparency Disclosure Index (US, EU) AIMR reports (US listed companies) IRRC Governance Index (US listed companies) CIFAR disclosure index intended use of funds, number of risk factors These metrics have been used in previous studies
The Disclosure Index 1. the company's corporate strategy is discussed. 2. the company's core/additional business segments are identified. 3. the nature of competition in the core market is discussed. 4. the main competitors in the core market are identified. 5. the barriers to entry to the core market are discussed. 6. the source(s) of competitive advantage are identified. 7. the intended use of IPO proceedings is clearly stated.
The Disclosure Index 8. an estimate of the company's core market value is provided. 9. the company's market share in its core market is quantified. 10. the expected future value of the core market is estimated. 11. sales for the company's core market are given. 12. the value/share of the company's foreign sales (if any) is given. 13. business evolution since the last annual report is discussed. 14. a quarterly breakdown of the financial accounts for the year before the IPO is provided.
The Disclosure Index Disclosure: Frequency plot -----------+-----------+----------------------------------------------------- 0 3 ** 1 5 *** 2 6 **** 3 7 ***** 4 10 ****** 5 26 ***************** 6 36 *********************** 7 46 ****************************** 8 60 *************************************** 9 77 ************************************************** 10 82 ***************************************************** 11 71 ********************************************** 12 46 ****************************** 13 33 ********************* 14 7 ***** -----------+-----------+----------------------------------------------------- Total 515
The Disclosure Index Panel A: Item composition Item: All C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 C12 C13 C14 Mean 8.85 0.86 0.85 0.96 0.57 0.77 0.36 0.41 0.66 0.60 0.33 0.70 0.31 0.72 0.51 St.Dev. 2.93 0.34 0.35 0.17 0.49 0.42 0.48 0.49 0.47 0.48 0.47 0.46 0.46 0.45 0.50
The Disclosure Index Panel B: Sectoral composition Sector: Biomed Soft./Internet Technology Media Telecom Traditional n.: 42 288 82 57 29 17 Mean 8.26 8.96 9.10 8.57 8.68 8.29 St.Dev. 3.01 2.94 2.78 2.59 3.66 2.97
The Disclosure Index Panel C: Market composition Market: Germany France Italy n.: 315 150 40 Mean 8.57 9.38 9.12 St.Dev. 3.02 2.94 1.75
Dependent variable We measure underpricing as the logarithm of the ratio of the first day s closing price to the issue price. This ratio is closer to a normal distribution than the traditional measure of underpricing.
Independent variables We look at four sets of explanatory variables: Disclosure: DISCLOSURE (index 1 to 14) DISCLOSURE-POS (positive deviation from median) DISCLOSURE-NEG (negative deviation from median)
Independent variables Issue characteristics: IPO-SIZE (log of gross proceedings) DILUTION (primary sale, percent) INSIDERS SALES (secondary sales, percent) RETURN VOLATILITY over 12 week (return SD) UNDERWRITER (above median mkt share) UNDERWRITER 2 (international bank)
Independent variables Issue characteristics: PREVIOUS IPOs over 4 and 12 weeks (by mkt) STABILIZATION ( greenshoe potential, perc.) SECONDARY OFFERINGS dummy PRICE-REVISION dummies (UP/DOWN) (no bookbuilding)
Independent variables Company characteristics: AGE at IPO (log of months) SIZE at IPO (log of assets, in euros) VC dummy (at least 12 months presence) R&D dummy SECTOR (6 sectors, from FT classification)
Independent variables Market characteristics: MARKET dummies (France residual) YEAR dummies (1996-97 residual)
Table 1: Descriptive statistics VARIABLE MEAN MEDIAN MIN MAX OBS Underpricing 0.250 0.121 1.265 1.844 515 Disclosure 8.851 9 0 14 515 Disclosure pos 1.027 0 0 5 515 Disclosure neg 1.176 0 0 9 515 IPO Size 17.188 17.211 14.787 21.716 514 Dilution 0.341 0.300 0 1 512 Insiders sales 0.067 0.041 0 0.604 513 Return volatility 0.608 0.055 0.015 0.264 514 Underwriter share 0.225 0 1 515 Underwriter intern 0.196 0 1 515 Previous IPOs 8.085 6 0 24 515 Stabilization 0.041 0.039 0 0.450 514 Seasoned Offerings 0.181 0 1 515 P revision up 0.129 9.000 1.000 14.000 490 P revision down 0.051 9.000 1.000 14.000 490
Table 1: Descriptive statistics VARIABLE MEAN MEDIAN MIN MAX OBS Age 142 114 4 1,125 515 Venture backed 0.144 0 1 515 R&D dummy 0.691 0 1 486 Assets 33.761 14.239 0.414 540.107 511 Leverage 0.633 0.554 1.707 27.046 477 ROA 0.148 0.079 6.993 20.773 509 Germany 0.630 0 1 515 Italy 0.078 0 1 515 France 0.292 0 1 515 Year 1996 0.029 0 1 515 Year 1997 0.056 0 1 515 Year 1998 0.153 0 1 515 Year 1999 0.324 0 1 515 Year 2000 0.400 0 1 515 Year 2001 0.037 0 1 515 Biomed 0.082 0 1 515 Soft./Internet 0.559 0 1 515 Technology 0.159 0 1 515 Media 0.111 0 1 515 Telecom 0.056 0 1 515 Traditional 0.033 0 1 515
Table 2: Correlations Table Disclosure Dilution Insiders IPO Age Assets Venture Underwriter Return Previous Sales Size Backed Share Volatility IPOs Disclosure 1.000 Dilution 0.032 1.000 Insiders Sales 0.055 0.109 1.000 IPO Size 0.068 0.106 0.051 1.000 Age 0.006 0.005 0.204 0.083 1.000 Assets 0.089 0.015 0.050 0.433 0.217 1.000 Venture Backed 0.015 0.004 0.104 0.002 0.077 0.104 1.000 Underwriter Share 0.024 0.022 0.024 0.071 0.055 0.028 0.071 1.000 Return Volatility 0.132 0.024 0.096 0.079 0.140 0.152 0.014 0.013 1.000 Previous IPOs 0.054 0.006 0.034 0.336 0.111 0.045 0.005 0.146 0.136 1.000
Empirical Strategy We start performing (robust) OLS: UP c = α + β 1 DIS c + ISSUE c β 2 + COMPANY c β 3 + MKT m β 4 + ε c We then add: Additional regressors Control for additional financial characteristics We then control for endogeneity: Simultaneity of disclosure and IPO features Effect of VC and underwriter on disclosure Selection on observables
Table 3: Base Model (i) Coeff. (St.Err.) Disclosure 0.011 (0.006) Dilution 0.002 (0.049) Insiders Sales 0.386 (0.203) IPO Size -0.016 (0.026) Age -0.003 (0.021) Assets 0.012 (0.019) Venture Backed 0.075 (0.037) Underwriter share 0.025 (0.036) Return Volatility 3.790 (0.502) Previous IPOs -0.012 (0.004) Germany 0.205 (0.045) Italy 0.066 (0.069) Intercept 0.023 (0.340) Year / Industry effects Yes Number of obs. 506 F(21, 482 ) 6.47 R-square 0.178
Table3:BaseModel (i) (ii) Coeff. (St.Err.) Coeff. (St.Err.) Disclosure 0.011 (0.006) Disclosure pos 0.004 (0.012) Disclosure neg 0.019 (0.010) Dilution 0.002 (0.049) 0.003 (0.049) Insiders Sales 0.386 (0.203) 0.397 (0.202) IPO Size -0.016 (0.026) 0.015 (0.026) Age -0.003 (0.021) 0.004 (0.022) Assets 0.012 (0.019) 0.011 (0.019) Venture Backed 0.075 (0.037) 0.075 (0.037) Underwriter share 0.025 (0.036) 0.019 (0.036) Return Volatility 3.790 (0.502) 3.818 (0.507) Previous IPOs 0.012 (0.004) 0.012 (0.004) Germany 0.205 (0.045) 0.209 (0.045) Italy 0.066 (0.069) 0.077 (0.069) Intercept 0.023 (0.340) 0.108 (0.325) Year / Industry effects Yes Yes Number of obs. 506 506 F(21, 482) 6.47 6.20 R-square 0.178 0.189
Table 4: Main Model Additional Regressors Coeff. (St.Err.) Coeff. (St.Err.) Disclosure 0.014 (0.007) Disclosure pos 0.005 (0.013) Disclosure neg 0.025 (0.011) Dilution 0.029 (0.056) 0.311 (0.232) Insiders Sales 0.298 (0.233) 0.033 (0.056) IPO Size 0.018 (0.029) 0.016 (0.029) Age 0.003 (0.022) 0.004 (0.022) Assets 0.010 (0.021) 0.009 (0.021) Venture Backed 0.068 (0.040) 0.066 (0.040) Underwriter share 0.034 (0.039) 0.027 (0.038) Return Volatility 3.668 (0.515) 3.708 (0.522) Previous IPOs 0.011 (0.004) 0.012 (0.004) Germany 0.214 (0.047) 0.220 (0.048) Italy 0.105 (0.073) 0.118 (0.073) Stabilization 0.280 (0.642) 0.332 (0.639) P Revision up 0.102 (0.048) 0.096 (0.048) P Revision down 0.130 (0.055) 0.131 (0.055) Seasoned Offerings 0.092 (0.051) 0.090 (0.052) Intercept 0.012 (0.363) 0.163 (0.349) Year / Industry effects Yes Yes Number of obs. 481 481 F(25, 453 ) F (26, 452 ) 6.07 5.90 R-square 0.194 0.198
Table 5: Main Model Additional Financial Characteristics Coeff. (St.Err.) Coeff. (St.Err.) Disclosure 0.015 (0.007) Disclosure pos 0.010 (0.013) Disclosure neg 0.031 (0.011) Dilution 0.019 (0.056) 0.021 (0.056) Insiders Sales 0.221 (0.240) 0.238 (0.238) IPO Size 0.016 (0.030) 0.013 (0.030) Age 0.003 (0.022) -0.006 (0.023) Assets 0.009 (0.021) 0.008 (0.021) ROA 0.000 (0.016) 0.004 (0.016) Leverage 0.019 (0.010) 0.021 (0.010) Venture Backed 0.087 (0.041) 0.086 (0.040) Underwriter share 0.040 (0.040) 0.031 (0.040) Return Volatility 3.585 (0.603) 3.603 (0.613) Previous IPOs 0.009 (0.004) 0.010 (0.004) Germany 0.189 (0.049) 0.199 (0.049) Italy 0.130 (0.082) 0.152 (0.081) Stabilization 0.058 (0.683) 0.108 (0.676) P Revision up 0.138 (0.052) 0.127 (0.052) P Revision down 0.138 (0.056) 0.138 (0.055) Seasoned Offerings 0.101 (0.054) 0.100 (0.054) Intercept 0.012 (0.370) -0.200 (0.355) Year / Industry effects Yes Yes Number of obs. 447 447 F(27, 419 ) F (28, 418 ) 5.04 5.02 R-square 0.195 0.203
Extension/Robustness We consider several alternative models: We control for R&D Activity (dummy) Include US-GAAP and IAS dummies Employ pre-ipo average of financial data Employ financial data at IPO Secondary sales by founders and managers Underpricing as percentage, not logarithm
Endogeneity First, we recognize that disclosure is choice which is simultaneous to other choices of the IPO process. For this we use a 2SLS where the first stage estimates how much is disclosed
Table 6(a): Simultaneous regressions Coeff. (St.Err.) Coeff. (St.Err.) Coeff. (St.Err.) Disclosure 0.079 (0.040) Disclosure pos 0.146 (0.073) Disclosure neg 0.148 (0.088) Dilution 0.002 (0.076) -0.010 (0.074) 0.002 (0.078) Insiders Sales 0.386 (0.227) 0.347 (0.210) 0.408 (0.257) IPO Size 0.007 (0.033) 0.000 (0.030) -0.011 (0.039) Age 0.008 (0.023) 0.015 (0.024) 0.001 (0.025) Assets 0.008 (0.019) 0.010 (0.019) 0.005 (0.020) Venture Backed -0.113 (0.055) 0.101 (0.055) 0.122 (0.060) Underwriter share 0.036 (0.047) 0.049 (0.050) 0.020 (0.047) Return Volatility 4.205 (0.730) 3.985 (0.700) 4.348 (0.826) Previous IPOs 0.007 (0.004) 0.006 (0.004) 0.008 (0.004) Germany 0.061 (0.076) 0.057 (0.079) 0.080 (0.075) Italy 0.001 (0.089) 0.054 (0.105) 0.068 (0.086) Intercept 0.640 (0.703) 0.011 (0.461) -0.196 (0.455) Year / Industry effects Yes Yes Yes Number of obs. 449 449 449 Wald χ2 73.88 71.04 68.18
Table 6(b): First stage regressions Coeff. (St.Err.) Coeff. (St.Err.) Coeff. (St.Err.) Age 0.102 (0.147) 0.104 (0.077) 0.001 (0.098) ROA 0.195 (0.126) 0.132 (0.066) 0.063 (0.080) Leverage 0.150 (0.071) 0.099 (0.037) 0.066 (0.046) R&D 0.649 (0.256) 0.286 (0.130) 0.357 (0.173) Assets 0.162 (0.106) 0.080 (0.055) 0.080 (0.070) Venture Backed 0.440 (0.339) 0.147 (0.178) 0.290 (0.225) Underwriter share 0.113 (0.286) 0.165 (0.150) 0.051 (0.190) Return Volatility 5.341 (4.261) 1.519 (2.235) 3.831 (2.835) Previous IPOs 0.030 (0.025) 0.020 (0.013) -0.009 (0.017) Germany 1.835 (0.328) 0.984 (0.172) 0.851 (0.218) Italy 1.577 (0.521) 1.188 (0.273) 0.392 (0.347) Year 1998-2.044 (0.544) 0.465 (0.285) 1.580 (0.362) Year 1999 2.566 (0.545) 0.690 (0.286) 1.875 (0.362) Year 2000 2.700 (0.550) 0.880 (0.288) 1.824 (0.366) Year 2001 3.492 (0.810) 1.239 (0.425) 2.255 (0.539) Intercept 9.040 (1.919) 0.956 (1.007) -0.196 (0.455) Industry effects Yes Yes Yes Number of obs. 449 449 449 Wald χ2 89.73 83.45 69.02 R-square 0.084 0.124 0.124
Endogeneity Second, we recognize that disclosure is a choice which is affected by some other regressors, most importantly: The presence of a venture capitalist The choice of underwriter For this we use a 2 step treatement regression model
Table 7(a): Treatment Effects Model Disclosure Coeff. (St.Err.) Coeff. (St.Err.) Disclosure 0.012 (0.006) 0.011 (0.006) Dilution 0.005 (0.068) -0.002 (0.060) Insiders Sales 0.375 (0.203) 0.374 (0.230) IPO Size 0.010 (0.030) 0.008 (0.027) Age 0.014 (0.023) 0.005 (0.024) Assets 0.009 (0.025) 0.014 (0.019) Venture Backed 0.186 (0.176) 0.083 (0.043) Underwriter share 0.022 (0.044) 0.185 (0.268) Return Volatility 3.761 (0.719) 3.916 (0.617) Previous IPOs 0.008 (0.004) 0.010 (0.004) Germany 0.216 (0.052) 0.125 (0.070) Italy 0.157 (0.092) 0.053 (0.108) Intercept 0.124 (0.409) 0.020 (0.373) Year / Industry effects Yes Yes Lambda 0.107 (0.142) 0.123 (0.151) Number of obs. 467 449 Wald χ2 324.60 148.30
Table 7(b): Treatment Effects Model Disclosure deviations Coeff. (St.Err.) Coeff. (St.Err.) Disclosure pos 0.007 (0.015) 0.008 (0.012) Disclosure neg 0.027 (0.012) 0.024 (0.013) Dilution 0.009 (0.069) 0.001 (0.038) Insiders Sales 0.368 (0.199) 0.385 (0.238) IPO Size 0.009 (0.027) -0.009 (0.027) Age 0.008 (0.023) 0.007 (0.025) Assets 0.009 (0.019) 0.015 (0.021) Venture Backed 0.095 (0.221) 0.085 (0.043) Underwriter share 0.030 (0.047) 0.191 (0.270) Return Volatility 3.750 (0.602) 3.941 (0.630) Previous IPOs 0.010 (0.005) 0.010 (0.004) Germany 0.214 (0.059) 0.131 (0.076) Italy 0.149 (0.085) 0.068 (0.095) Intercept 0.270 (0.355) 0.142 (0.390) Year / Industry effects Yes Yes Lambda 0.107 (0.129) 0.123 (0.156) Number of obs. 467 449 Wald χ2 188.59 174.86
Endogeneity With propensity score we check that our results are not due to selection in the amount to disclose due to observable firm characteristics.
Conclusions Voluntary information disclosure seems to reduce UP, even accounting for observable and unobservable selection effects. Disclosure also improves on previous explanations in terms of goodness of fit. We are currently working to extend the analysis of endogenous effects