Accounting and various aspects of finance
Javad Shekakhah; Iraj Asghari
Abstract
This article deals with modeling the long-term performance of IPOs in the Tehran Stock Exchange and OTC. Due to the difficulty of determining the definition of the long-term period, modeling was initially conducted for 12 periods. These periods ranged from 3 to 36 months. The purpose of this modeling ...
Read More
This article deals with modeling the long-term performance of IPOs in the Tehran Stock Exchange and OTC. Due to the difficulty of determining the definition of the long-term period, modeling was initially conducted for 12 periods. These periods ranged from 3 to 36 months. The purpose of this modeling was to analyze and compare the results and identify the most suitable periods for explaining the long-term performance of IPOs. Modeling has been conducted at the portfolio level using a Stepwise approach. For this purpose, the monthly time series was formed, and data from 236 IPOs in the Tehran Stock Exchange and OTC markets from 2009 to 2022 have been analyzed. The results showed that the return of the portfolios formed from initial offerings could be explained at a satisfactory level. While the primary factor in explaining the long-term performance of IPOs is market return, the profitability, and its distribution also play a significant role. Finally, the most suitable periods for use as the definition of the long-term period are 12, 21, and 27 months.IntroductionThe long-term performance of Initial Public Offerings (IPOs) has always been disputed by researchers. The inherent challenges of conducting long-term research and the complexities associated with Initial Public Offerings have led researchers to use different methods resulting in inconsistent findings.A prevalent approach in studying long-term IPOs is the use of “factor models” to identify the factors influencing IPO portfolio performance. However, the literature has presented and utilized several factor models. Examples of these models include Fama and French (1993), Carhart (2004), Fama and French (2015), and Ho et al. (2015). Despite some similarities, each of these models employs different factors and variables to explain IPO performance. In recent years, many researchers have criticized the use of these common models in their respective countries, citing reasons such as ineffectiveness of these models. These researchers argue that neglecting the socio-economic context of societies can lead to misinterpretation of return and yield inappropriate results for decision-makers. Consequently, each society should develop and employ its own models. Considering these issues, this research aims to provide models that explain the long-term performance of Iranian IPOs. Specifically, by testing various factors and variables, this study identifies the most effective models for explaining the long-term performance of IPOs in Iran.MethodologyIn this research, a stepwise approach was employed. Monthly data of 236 IPOs between 2009 and 2022 were utilized to construct relevant time series, and the returns of the IPO portfolios were analyzed with respect to potential factors that explain the return. To determine the initial set of variables, a systematic review approach was adopted. Due to the high correlation and multiple proxies for the liquidity factor, the liquidity variables were first reduced to three factors using principal component analysis. In total, 19 different factors and variables were included in the analysis.Given the lack of consensus among researchers regarding the definition of the long-term period, the modeling process in this research considered 12 different periods ranging from 3 to 36 months with a three-month increment. The selection of appropriate models was based on the criteria of accuracy and quality forecast, specifically Theil’s (1975) criterion. Three models that nest met these criteria were chosen, and the corresponding portfolio periods were identified as the defining terms for the long-term period. The validation of the selected models was performed by comparing their adjusted R2 values with those of common models found in the literature. Additionally, out-of-sample testing was conducted using 10% of the data to assess the model’s performance.Results and DiscussionThe research findings indicate that the models developed in this study exhibit a strong explanatory power, accounting for approximately 80% of the variations in the returns of IPO portfolios. Among the different portfolio periods considered, the models constructed using 12, 21, and 27-month portfolios demonstrated superior accuracy and forecast quality according to Theil’s (1975) criteria. As a result, these specific periods were identified as the most suitable definitions for the long-term period in this context. The significant variables identified in the models include market return, profitability, size, and dividend. Although the models generally incorporate a set of relatively common variables, the specific model associated with each defined period can be employed to achieve better results, taking into account the specific characteristics of the long-term period under consideration. Furthermore, it is worth noting that the intercept of the designed models, as well as the intercepts of the common models found in the literature, were found to lack statistical significance.ConclusionBased on the analysis conducted in the research, it can be concluded that utilizing native models specifically designed for IPOs provides a suitable explanation for their long-term performance. The primary factor in explaining the long-term performance of IPOs is found to be the market return. This suggests that the performance of initial offerings is primarily influenced by the overall market conditions, while other variables, such as profitability help modulate this effect. Additionally, the non-significance intercept in the models indicates that there is no evidence of long-term under or over-performance of IPOs in Tehran's financial markets. The superiority of the designed models compared to other common models is evident primarily in the 12-month period. While the performance of the models in other periods depends on the specific model employed.
