ABSTRACT
This study presents a comparative analysis of under-pricing and short-term performance of IPOs issued during the COVID-19 pandemic period and the pre-pandemic tranquil decade (2009 to 2019) in a cross-country setup. We find evidence of higher underpricing of IPOs issued during COVID-19 which, however, gets corrected shortly. Factors, such as underwriter reputation, percentage of the net proceeds to the company, new shareholder participation, industry affiliation of the issuing firm and the severity of the pandemic in respective countries seem to affect these patterns. Our results are robust and remain mostly unaltered through a series of robustness tests.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Supplementary material
Supplemental data for this article can be accessed online at https://doi.org/10.1080/1540496X.2022.2147780.
Notes
1. We extend the study period by another 8 months till July 2021 to conduct an “out of sample” analyses as one of the robustness tests.
5. We use three robustness tests, namely an alternative model, out of sample analysis, and analysis over a winsorized sample. The details are provided in Section 3.6 later.
6. Morgan Stanley Capital International (MSCI) country classification based on criteria such as economic development, size, liquidity requirement, and market accessibility, categorizes countries into four categories: developed, emerging, frontier, and standalone markets. https://www.msci.com/market-classification.
7. We take IPOs from only those countries in our sample for which data for at least 10 IPOs are available, and defined the start date of data collection for COVID period for each country as the date when the country reported 100+ cases for the first time. Also, to avoid the impact of Chinese market crash on IPOs, we removed Chinese IPOs issued during 1st Mar 2015 to 29th Feb 2016.
8. We run all relevant diagnostic tests for the data before running models (1), (2) and (3) to check for the sources of errors. We run the VIF values for the model to check for multicollinearity, Breusch-Pagan (BP) test (Breusch and Pagan Citation1979) and White’s test (White Citation1980) to check for heteroskedasticity of data. We do not find any significant problems with the data to influence our results: The VIFs range between 1.2 and 1.4 and none of the heteroskedasticity test statistics are significant at the 5% level of significance. In few cases, the heteroskedasticity statistic exhibits weak significance (10%) for which we use robust standard error and report updated significance level in the tables, in line with White (Citation1980). Further, we also check for autocorrelation using Durbin-Watson (DW) test introduced by Durbin and Watson (Citation1950), and we do not find any significant evidence of serial-correlation in our data.
9. As quantile regression estimates the coefficients by minimizing the weighted sum of absolute deviation, the parameter estimates tend to be less sensitive to fat-tailed distribution. Also, Buchinsky (Citation1998); John and Nduka (Citation2009); Rodriguez and Yao (Citation2017) explain that quantile regression, unlike least squares regression neither assumes any parametric distribution nor constant variance and, hence, provides a robust estimate in the presence of heteroscedasticity. Unlike the OLS method, quantile regression uses median as the measure of central tendency and is useful when data contains outliers that might influence the “mean” of the observations.
10. We also estimate the other elements of descriptive statistics like median, standard deviation, maximum, minimum and quartile values but we do not report and discuss them here for the sake of brevity.
11. Ideally, from an investor’s perspective, equity performance should be studied over intermediate (1–2 years) or long-term (3–5 years). However, as we are studying the COVID period IPOs and the length of the study period, in this case, is just about a year, we had to restrict our time period to the next 30 trading days post-listing.
12. GICS industry classification: https://www.msci.com/our-solutions/indexes/gics.
13. Impact ratio = (total covid cases of country i)/(total population of country i).
14. The data on the total number of COVID cases and country population was collected as on 30th November 2020 from the World Health Organisation (WHO) COVID 19 database (https://covid19.who.int/), and the World Bank database(https://data.worldbank.org/indicator/SP.POP.TOTL), respectively.
15. Interested readers may please refer to table 3S, panel A, B and C in the supplementary document to access the tables.