Offline Oversubscription, Issue Size, and Market Momentum: The Driving Forces for ChiNext IPOs’ Initial Underpricing

Abstract

This paper reexamines the driving forces for the first day initial return for ChiNext IPOs. We start from screening 29 potential explanatory variables, 4 policy break dummies, and 2 intraday trading suspension dummies, using an OLS model with dimension reduction techniques to identify significant variables. We then apply a 2SLS procedure to remove endogeneity without losing any important information. With the variables identified from the 2SLS model, we further apply a GARCH-M model with an ARMA(1,1) adjustment in the residuals to correct possible autocorrelation in the regression residuals and cross-correlation between the initial return and its conditional return variance. We find that the model fits the data well. From a number of potential factors in pricing Chinese IPOs, we identify three factors that drive the initial underpricing of ChiNext IPOs: the pre-issue share allocation multiplier from institutional investors (offline oversubscription), issue size (size effect), and the listing day stock market condition (market momentum). We estimate the contribution to the initial underpricing from each of the significant variables.

Notes

See, for example, Loughran, Ritter, and Rydqvist (1994); Mok and Hui (1998); Su and Fleisher (1999); Ritter and Welch (2002); Gu (2003); Groenewold, Wu, Tang, and Fan (2004); Loughran and Ritter (2004); Guo and Brooks (2008); Li, Fowler, and Naughton (2008); Lowry, Officer, and Schwert (2010); Zhou and Zhou (2010); Guo, Brooks, and Fung (2011); Hussein and Zhou (2014); among others.

The turnover ratio is defined as the ratio of trading volume of an issue on a day to the total number of floating shares of the issue.

The business hours at the SZSE are 9:15 A.M.–9:25 A.M., pre-open session; 9:30 A.M.–11:30 A.M., morning session; 13:00 P.M.–14:57 P.M., afternoon session; and 14:57 P.M.–15:00 P.M., closing call auction.

All data are hand-collected from http://www.szse.cn/main/chinext/ and http://finance.sina.com.cn/stock/. The firm-level financials are collected from the related sections in the issuers’ prospectuses.

Some of the variables can be correlated. That problem will be taken care of when we use an OLS model with variable reduction techniques to identify the significant variables and, at the same time, to remove any of the insignificant (redundant) variables.

We perform the Box and Jenkins (1976) procedure on the first day initial return for 352 ChiNext IPOs. We conduct different tests and combinations of ARMA(p,q) procedures. We find that an ARMA(1,1) process offers the best model fitness. The detailed analysis is available from the authors upon request.

Unfortunately, the SPSS software does not produce standardized Beta for each of the significant variables or the individual variable’s contribution to the overall adjusted R².

Even though not reported, we find that the distribution of the regression residuals behaves even better under the GARCH-M model with an ARMA(1,1) adjustment, evidenced by the Q-statistics on the residuals at the residual level and the variance level, respectively. The detailed results are available upon request.