A New Modus Operandi for Determining Post - IPO Pricing : Analysis of Indian IPOs using Artificial Neural Networks

Abstract

The objective of this study was to identify different factors useful in determining post-IPO pricing and test their relative significance by comparing stock performance across 3-, 6-, and 12-months post listing. To do so, the study analyzed data from 299 non-financial companies that had their IPOs listed on the Bombay Stock Exchange from 2005–2018 in India. The data collected were used to train a neural network, called the multilayer perceptron model. The study grouped all factors into four categories viz-a-viz macroeconomic, issue-specific, technical, and fundamental. Analysis of the results generated from 20 iterative constructions of the neural network revealed that the highest relative relevance in prediction was attributed to technical factors. It was also observed that the importance of fundamental factors increased with the investment horizon. The results are country-specific and found that the importance of “underpricing” and “listing gains” as factors reduced within a year post-listing and thus, provide a helpful addition to the present knowledge of financial gains resulting to investors from IPOs.