Factors Influencing IPO Underpricing in Poland

We review the theory and evidence on IPO activity and underpricing focusing on the War‐ saw Stock Exchange. Although the topic has been under investigation in the past, we believe that the recent decade of low interest rates deserves inquiry. We research the extent of underpricing dur‐ ing this period and further conclude that three factors had a statistically significant influence on ini‐ tial public offering underpricing during this period: the year of IPO, risk‐free rate and WIG close value.


Introduction
Companies that sell shares on the stock exchange for the first time (Initial Public Offering) advertise and offer incentives to new shareholders. To attract investors, shares are usually offered at a price which is lower than that resulting from valuations. The procedure is called "underpricing". The size of the incentives and pre-IPO valuations are confidential and as such difficult to research. To assess underpricing, an indicator of Initial Return (IR) is calculated based on the price at the end of the first day of trading and the price at which shares were sold to new investors.
Data gathered by Jay Ritter (2018) indicate that IPO underpricing in the United States fluctuates substantially, averaging 21.2 in the 1960s, 7.1% in 1970s, 6.9% in the 1980s, 21% in the 1990s, and 21.1% in the years 2001-2017 1 . Ritter further indicates that underpricing depends upon the size of the company issuing shares and prior financing sources. Initial return decreases with the size of the company (measured in revenues). Underpricing is also higher for venture capital backed companies than it is for growth and buy-out funds financed companies.
In this study, we investigate underpricing in IPOs on the Warsaw Stock Exchange (WSE) between 2005 and 2016. Since there is sufficient scientific evidence to assume that IPO underpricing existed in this period and that it is likely to exist in the future, these results are used as a background for further analysis of factors influencing the extent of underpricing in this period. We analysed three groups of potential factors and focused on the impact of three: the year of IPO, risk-free rate and WIG close value. We admit that findings of this study are limited to the WSE and one economic cycle.
We believe that the results of this study may be interesting for analysts, investors, consultants and managers involved in IPOs. The researched period was characterised by low interest rates and we believe that this factor makes our research and its results worth considering.

Dataset and methodology
In this paper, we investigate IPOs of 349 companies quoted on the Main Market of the Warsaw Stock Exchange that took place between 2005 and 2016. Only companies which offered their shares to the public for the first time were taken into consideration. Firms previously quoted on the WSE or another market were excluded from the sample. Data for the WIG -Warszawski Indeks Giełdowy (Warsaw Stock Exchange Index) -are published by the Warsaw Stock Exchange. During the researched period, the number of IPOs and WIG closing values were loosely correlated. The number of public offerings dropped after 2008 and remained lower than in prior years even though the stock market recovered and reached higher valuations. The reasons for that may vary but the primary reason may be that better access to debt financing combined with low interest rates seem to decrease the attractiveness of public equity financing.
Our research results indicate that average underpricing on the WSE in the years 2005-2016 was 12.35%. Further analysis indicates that underpricing changes over time, with the size of the offer, and due to other factors.
To assess underpricing, we used two methods. Using the first method, we compared the first day closing price to the offering price. Using the second method, we subtracted broad market index change.
reached higher valuations. The reasons for that may vary but the primary reason may be that better access to debt financing combined with low interest rates seem to decrease the attractiveness of public equity financing. To assess underpricing, we used two methods. Using the first method, we compared the first day closing price to the offering price. Using the second method, we subtracted broad market index change.
In the second step, the broad market index change is subtracted from the initial return. In this paper, we used Warszawski Indeks Giełdowy (WIG), the main index of the Warsaw Stock Exchange. This method is less popular than the first one but, according to Hunger (2003), Sieradzki (2013: 1-37) and Czapiewski et al. (2014: 571-590), it provides more reliable results, as stock market (index) changes may impact first day closing prices.
, , where: IAR -initial adjusted return on the day (t) that the company (i) debuted, IR WIG i,t -return of broad market index on the day (t) that the company (i) debuted.
Any list of factors influencing underpricing is subjective. The list of factors used in this research was created from most commonly used statistics in publications of the WSE.

Factor Description
Year The year of IPO. Offering Price Initial price of one share. New issue [%] Part of the company that was offered to the public in percentage.

IR [%]
Initial rate of return (underpricing). WIG change [%] Percentage change of WIG on the day that the company was offered.

WIG close
Value of broad market index in base points.

Offering value
Total offering value of the company in PLN.

Risk-free rate
Risk free rate on the day that the company was offered.

Number of IPOs
Number of IPOs in the year that the company was offered.

Source: own elaboration
As a risk-free rate, we have taken the reference rate of the National Bank of Poland. We have chosen ordinary least squares methodology to look for correlations between underpricing and factors. This methodology was previously used in underpricing examination of IPOs on New Connect by Fijałkowska, Muszyński and Pauka (2013: 415-426).
We used Gretl open source statistical package to perform the analysis. IR was the dependent variable. Factors listed in Table 2 were used as regressors.

Underpricing
Average underpricing (initial return) of the 349 companies which debuted on the Warsaw Stock Exchange from 2005 to 2016 was 12.35%. In comparison to 2005-2013(Pomykalski, Domagalski, 2015, the average rate of return on the first day increased by 0.46%. The average adjusted IR (IAR) underpricing of IPOs is 11.84%, which is smaller than IR. Decreasing change of the adjusted IR is expected. It means that IPOs were increasing their value more when there was a positive change of WIG. According to the calculation performed, the average influence of WIG change on the closing price is 0.51%. The median for IR is 4.71% and it is a lot lower than the average which is 12.35%. The same situation is for the corrected IAR -4.15% compared to 11.84%.
The distribution is far from normal (curtosis of 81.52 for IR and 82.23 for IAR). Skewness is positive (7.34 for IR and 7.42% for IAR), which means there are more results above the average than below the average values.
Almost 27% of the offers were overpriced (29% if IAR is used). 238 companies brought a positive initial return while considering IR, which is 68% of all IPOs (249 companies or 71% if IAR is used). The majority of IPOs are underpriced and results indicate the existence of underpricing on the Warsaw Stock Exchange.
The level of underpricing varies. In 27% of the cases, initial return was negative and in one case it amounted to -74.07%. This means that investing in IPOs on the WSE in hope of exploiting underpricing is associated with risk and may bring disappointing results.

WIG log
Year

Source: own elaboration
There are statistically significant correlations between independent variables at a level of 0.01 and 0.05. In order to meet ordinary least squares requirements, we have performed Variance Inflation Factors statistics to examine a collinearity problem. Since all of the values are below 10, collinearity problem does not exist. Two of the independent variables have much higher values than the rest: WIGclose and OfferingValue. In order to have a similar scale, natural logarithms of these variables were used in the model.
Four OLS models were built. The first model consists of all variables listed in Table 7. P-value was the highest for the variable NewIssue which was rejected from the second model. The second model R-squared was higher than the first model R-squared. In the second model, p-value was the highest for the variable Of-feringPrice which was rejected from the third model. The third model R-squared was higher than the second model R-squared. Since rejections from the third model would lower the R-squared, the third model was the final one.
Two of the variables have a statistically significant impact on IR at a level of 0.01: the year and risk-free rate. One variable has a statistically significant impact on IR at a level of 0.05: the number of IPOs. Considering significance at the 0.1 level, one variable has an impact on IR: WIGclose. The factor with the most significance is the year of the IPO (Figure 3). The coefficient of the year variable was -0.06, which means that the higher the year value, the lower the underpricing level. In developing markets, this can be explained by efficient market hypothesis (EMH). Fama (1965: 34-105) described an efficient market as "a market where prices at every point in time represent best estimates of intrinsic value. This implies in turn that, when an intrinsic value changes, the actual price will adjust "instantaneously", where instantaneously means, among other things, that the actual price will initially overshoot the new intrinsic values as often as it will undershoot it". Significance of year variable means that the WSE is more efficient now than it was in the past. This explanation is controversial. Adams, Thornton and Hall (2008: 67-74) in their study of IPO pricing argue against associating EMH with IPOs. means, among other things, that the actual price will initially overshoot the new intrinsic values as often as it will undershoot it". Significance of year variable means that the WSE is more efficient now than it was in the past. This explanation is controversial. Adams, Thornton and Hall (2008: 67-74) in their study of IPO pricing argue against associating EMH with IPOs.
Results of this and other studies conducted in Poland using data starting after the year 2000 (Czapiewski et al., 2014: 571-590;Pomykalski, Domagalski, 2015: 117-132) indicate that the level of underpricing was below 21.1% observed by Ritter in the US markets. Assuming that US markets are more developed, the conclusion that underpricing is higher in less developed countries does not hold. The second most significant factor is the risk-free rate (Figure 4). The coefficient is -9.49.
It means that the higher is the risk-free rate, the lower is the IPO initial rate of return. The sign of the coefficient is also expected. When interest rates are high, investors usually save their money because profits from bank deposits are satisfactory. Conversely, if interest rates are low, Results of this and other studies conducted in Poland using data starting after the year 2000 (Czapiewski et al., 2014: 571-590;Pomykalski, Domagalski, 2015: 117-132) indicate that the level of underpricing was below 21.1% observed by Ritter in the US markets. Assuming that US markets are more developed, the conclusion that underpricing is higher in less developed countries does not hold.
The second most significant factor is the risk-free rate (Figure 4). The coefficient is -9.49. It means that the higher is the risk-free rate, the lower is the IPO initial rate of return. The sign of the coefficient is also expected. When interest rates are high, investors usually save their money because profits from bank deposits are satisfactory. Conversely, if interest rates are low, investors are looking for different investment opportunities because they cannot earn as much as they wish on a bank deposit. For that reason, more people are likely to buy shares when interest rates are low. High interest rates increase demand for shares and consequently first day closing prices. The third most significant factor influencing IPO underpricing is the number of IPOs that took place in a given year (Figure 2). The coefficient is -0.0037 and it is negative, which means that when more IPOs took place in a given year, the underpricing level was lower. One possible explanation is the law of supply and demand. When there are more IPO offers on the market, people are not willing to pay as much as if there were fewer offers.
The least important but also significant factor is the WIG closing level. The WIG (Warszawski Indeks Giełdowy) is a cumulated value of all securities quoted on the Warsaw Stock Exchange. In this case, the WIG close is the level measured at the end of IPO day. The WIG coefficient is positive, which means that the higher the broad market index, the higher the IPO initial return. There is no clear explanation of this phenomenon in the literature. One  The third most significant factor influencing IPO underpricing is the number of IPOs that took place in a given year (Figure 2). The coefficient is -0.0037 and it is negative, which means that when more IPOs took place in a given year, the underpricing level was lower. One possible explanation is the law of supply and demand. When there are more IPO offers on the market, people are not willing to pay as much as if there were fewer offers.
The least important but also significant factor is the WIG closing level. The WIG (Warszawski Indeks Giełdowy) is a cumulated value of all securities quoted on the Warsaw Stock Exchange. In this case, the WIG close is the level measured at the end of IPO day. The WIG coefficient is positive, which means that the higher the broad market index, the higher the IPO initial return. There is no clear explanation of this phenomenon in the literature. One possible reason of this de-pendency is stock exchange attractiveness. If the value of broad market index is at a high level, the stock exchange seems to be more attractive to investors who are more likely to invest in IPOs. That causes higher initial returns of IPOs. Adjusted R-squared of the final model is 7.47%. Since the goal of this paper was not to develop a model describing underpricing but to identify factors influencing initial return, the value is rewarding. The impact of WIG on IAR was stronger in the period 2005-2008. This indicates that lower interest rates and the number of IPOs in the later years impacted underpricing to a larger extent than the broad stock market index. This conclusion is based on a short period of analysis and requires further research.

Limitations of this study
This study is limited to the WSE and one economic cycle. Results may be different in both more mature and less mature markets. We researched the impact of factors from a list of statistics used by the WSE. There may be other factors influencing underpricing as shown by Wołoszyn and Zarzecki (2013: 121-135).
Further research can concentrate on factors such as underwriters' reputation, free float of shares, market segment affiliation and oversubscription.

Conclusions
We have examined 349 companies which went public in 2005-2016. We have calculated the The impact of WIG on IAR was stronger in the period 2005-2008. This indicates that lower interest rates and the number of IPOs in the later years impacted underpricing to a larger extent than the broad stock market index. This conclusion is based on a short period of analysis and requires further research.

Limitations of this study
This study is limited to the WSE and one economic cycle. Results may be different in both more mature and less mature markets. We researched the impact of factors from a list of statistics used by the WSE. There may be other factors influencing underpricing as shown by Wołoszyn and Zarzecki (2013: 121-135).
Further research can concentrate on factors such as underwriters' reputation, free float of shares, market segment affiliation and oversubscription.

Conclusions
We have examined 349 companies which went public in 2005-2016. We have calculated the initial return and adjusted initial return and obtained results of 12.35% and 11.84% respectively. We have confirmed that underpricing existed during the researched period.
We have also examined the influence of selected factors on underpricing and can conclude that during the researched period three of the examined factors had a statistically significant influence on initial public offering underpricing. The year of IPO (negatively), risk-free rate (negatively) and WIG close value (positively) influenced underpricing during the researched period.
Due to a limited scope of our research (a short period and one market), our results should be treated with caution and used in further inquiries into a possible impact of interest rates and stock market indexes on underpricing.