1 Stabilization activity in Italian IPOs by Dimitri Borienko and Stefano Lombardo: Discussion Carlo Drago, Paolo Santella Italian Society of Law and Economics Fourth Annual Conference, University of Bologna 7-8 November 2008 Governance and Finance
2 Stabilization Actvity in Italian IPOs by Boreiko- Lombardo: Discussion C.Drago P.Santella Relevance: Explanations for the stabilization activity done by underwriters “Theoretical financial economics has yet no provided a sound and completely convincing explanation for the stabilization activity done by underwriters...” “US stabilization is done mainly by short covering only and the use of pure stabilization is not observed empirically...” “There are several empirical studies on stabilization of shares prices...” (In the US) The purposed study explores the italian IPOs stabilisation regime
3 Stabilization Actvity in Italian IPOs by Boreiko- Lombardo: Discussion C.Drago P.Santella The Study Design: the Sample Data on stabilizing activities of the underwriters were gathered from the press releases submitted to the regulating authorities. Estimations are made out of 130 IPOs (with respect to a full sample of 147). However, the authors classify the IPOs as stabilized (85) by underwriters and 57 as not stabilised.
4 Stabilization Actvity in Italian IPOs by Boreiko- Lombardo: Discussion C.Drago P.Santella The Study Design: the Sample Data on length of the intervention period: “...Still up to the end of the period under study, only 19 IPOs out of the 85 stabilized disclosed the detailed data on intervention activities on a daily basis”. The question is: -are all the tests run out of 85 stabilised observations? Or are they run out of 19? -If the answer is 85, it would be helpful to have a more detailed illustration of the initial inferential design.
5 Stabilization Actvity in Italian IPOs by Boreiko- Lombardo: Discussion C.Drago P.Santella The Statistical Analysis\Tests Descriptive Statistics on the full sample Regression of stabilization choice and intensity (Logit regression on the Stabilization dichotomic variable, Ordinary Least Squares regression on the stabilization intensity) Descriptive Statistics on the Stabilization activities in detail Price adjustment following the end of stabilization (T-test and Wilcoxon signed rank test)
6 Stabilization Actvity in Italian IPOs by Boreiko- Lombardo: Discussion C.Drago P.Santella The Study conclusion “We are able to identify precisely the listings that were stabilized and run a series of statistical tests designed to identify the determinants of the underwriter's decision to stabilize the offering price and factors influencing stabilization intensity” “...Unlike previous few empirical studies (related to the US) we observe not only short covering transactions but also direct trading in securities in order to support the price”
7 Stabilization Actvity in Italian IPOs by Boreiko- Lombardo: Discussion C.Drago P.Santella Missing data\Sample reduction Is there a statistical pattern in the observed sample reduction? Sample selection based on the available disclosure?...are the missing data randomly distributed? What are the effects on the statistical results?
8 Stabilization Actvity in Italian IPOs by Boreiko- Lombardo: Discussion C.Drago P.Santella The Underlying Data Structure Need for an exploratory data analysis of the relationships Is the linear model a strong assumption on the data structure? (it would be useful to make recourse to non- linear hypothesis on the data): a non-parametric or an explorative data analysis would add value to the analysis. Outliers, Influence and Leverage effect. Are outliers signalling a new portion of the theory? Are the assumptions respected in a reasonable way? Is it necessary to use a GLS estimator? (more details on Heteroscedasticity would be advisable)
9 Stabilization Actvity in Italian IPOs by Boreiko- Lombardo: Discussion C.Drago P.Santella Robustness Are empirical results robust to different econometric specifications? Are empirical results robusts to different econometric methodologies? Different statistical methodologies can be used to adress different economic points of interest in the data (for example the multiple correspondence analysis can be used to address the timing over the economic cycle of the IPOs)
10 Stabilization Actvity in Italian IPOs by Boreiko- Lombardo: Discussion C.Drago P.Santella Other ideas Making the database accessible (if possible) Making the algorithms used in the work accessible (for any software or computational method used in the analysis) In statistics there is a general enphasis on the replicability of the econometric methodologies (it would be useful for future applications on similar data matrices) Giving documentation on the software used
11 Stabilization Actvity in Italian IPOs by Boreiko- Lombardo: Discussion C.Drago P.Santella Bibliography (I) Allison (2001) “Missing Data” Thousand Oaks Sage publication Barnett Lewis (1994) “Outliers in Statistical Data” Wiley Benzècri (1976) “L'Analyse de Donnèes” Paris Dunnod Blasius Greenacre (2006) “Multiple Correspondence Analysis and related methods” CRC Press Boreiko Lombardo (2008) “Stabilization Activity in Italian IPOs” Working Paper presented at SIDE-ISLE fourth annual conference 7 november 2008 Cleveland (1993) “Visualizing Data” Hobart Press
12 Stabilization Actvity in Italian IPOs by Boreiko- Lombardo: Discussion C.Drago P.Santella Bibliography (II) Greene (2008) “Econometric Analyis” Prentice Hall Koenker (2006) “Reproducibility in Econometric Research” available online: Little Rubin (1987) “Statistical Analysis with Missing Data” New York, Wiley Mardia Kent Bibby (1980) “Multivariate Analysis” Academic Press Peracchi (2001) “Econometrics” Wiley and Sons Tukey (1977) “Exploratory Data Analysis” Addison Wesley