טכניון מכון טכנולוגי לישראל
הטכניון מכון טכנולוגי לישראל - בית הספר ללימודי מוסמכים  
M.Sc Thesis
M.Sc StudentRonen Rave
SubjectTesting the Ability of Predicting Ipo's Short-Term Returns
DepartmentDepartment of Industrial Engineering and Management
Supervisor Mr. Yossi Yagil
Full Thesis text - in Hebrew Full thesis text - Hebrew Version


Abstract

A well known IPO behavior pattern is short-term over-performance and long-term under-performance. A less known aspect of the short-term behavior is the “hot issue market”- a period of about one to two years, in which IPOs exhibit extremely high first day over-performance. During the “hot issue market” of 1999-2000, IPOs exhibited an average first day return of 65%. In view of this phenomenon it is interesting to test whether the IPO market is efficient in the weak form; that is, whether IPOs short-term returns can be predicted? The two main studies of “hot issue markets” are diverse in their results. Ibbotson and Jaffe (1975) found clear evidence for the ability to predict IPOs short-term returns, while Ritter (1984) found a prediction ability for specific industries but not for the entire market.

This study investigates the ability to predict IPOs short-term returns for the “hot issue market” of the 90’s by applying a wider set of tests than those examined in the literature. I also investigate the prediction relationship of IPOs short-term returns in relation to market variables. Finally, I will test for “hot issue market” segmentation; i.e., internet companies versus the market as a whole.

The sample contains 2,775 IPOs in 102 months from 1995 to 2003. Six time series of excess returns were computed. For each time series I have conducted five tests: (1) Stationary test; (2) Sequence & Reversal test; (3) Run test; (4) Box-Ljung test and (5) long memory test. The findings indicates that the four time series computed with respect to the closing prices were stationary, exhibited random walk and did not have long term memory. Hence, market efficiency in its weak form exists.  On the other hand, the two time series, computed with respect to the offer price, were stationary but did not exhibit random walk and had long term memory, implying that market efficiency does not apply to those series, and that they might be predictable. The findings also indicate that the IPOs over-performance in the first trading day disappears during the next day, implying that small investors for whom buying at the offer price is not feasible can not benefit from the first day abnormal return. Tests involving the prediction of short-term returns indicate no prediction ability. Tests for market segmentation demonstrate a “hot issue market” in the internet industry, but a weaker “hot issue market” for the market as whole.