raw underpricing and mean market adjusted underpricing. The result is expected as companies usually do not issue 100% of their equity when conducting an IPO, which is why the underpricing is lower when adjusting for the fraction of equity sold.
A simple t-test reveals that the market adjusted underpricing of 4.59% is statistically significantly different from zero on a 1% level with a t-statistic of 4.63, which exceeds the critical value of 2.64. The mean market adjusted underpricing corrected for offer size of 1.97% is also statistically significant on a 1% level, with a t-statistic of 4.71. These results confirm the existence of significant underpricing of IPOs on Scandinavian stock exchanges in the period 2002 – 2015.
The median of the market adjusted underpricing is 2.88%, indicating that half of the IPOs experienced an underpricing higher than 2.88%. Because the mean is higher than the median, the sample data appears to be skewed to the right. Skewness is calculated to be 0.77, indicating that data is
concentrated to the left and that the distribution has a right tale, relative to a normal distribution. This is in part due to the outlier with 38.00% underpricing, but when excluding this observation the
distribution still have a right tail. An excess kurtosis of 1.20 indicates that the data has a high peak and fat tails, and supports the notion that the data does not seem follow a normal distribution.
Figure 4: Distribution of underpricing adjusted for market return
The same pattern is evident in the distribution of offer size adjusted underpricing. The median of the offer size adjusted underpricing is 1.10%, significantly lower than the market adjusted underpricing.
43
The skewness of the offer size adjusted underpricing is slightly higher than for the market adjusted underpricing, 0.78, while the kurtosis is smaller, 0.48. This means that the data is slightly more
concentrated to the left and the distribution has a heavier right tale. The lower kurtosis indicates that the offer size adjusted underpricing has a lower peak than the market adjusted underpricing, although still higher than a normal distribution.
Figure 5: Distribution of underpricing adjusted for market return and offer size
It may be argued that for the t-test to be reliable, the data must follow a normal distribution. However, the central limit theorem states that the mean of a sample will be approximately normally distributed, as long as there are a sufficiently large number of observations. As our data sample contains 89 observations, the size is deemed to be sufficiently large for the t-test to be reliable for the purpose of testing whether there exist significant underpricing in the period of interest. Below, underpricing in each of the three countries is analyzed. Underpricing has been shown in earlier research to vary greatly between Scandinavian countries; this is also evident in our data.
44
Denmark
Obs = 7 Underpricing_raw Underpricing_adjusted Underpricing_offersize
Mean 0.0991 0.0938 0.0428
Standard Error 0.0433 0.0422 0.0167
Median 0.0750 0.0707 0.0307
Standard deviation 0.1146 0.1117 0.0443
Sample Variance 0.0131 0.0125 0.0020
Highest initial return 0.2560 0.2519 0.1159
Lowest Initial return -0.0814 -0.0805 -0.0159
Table 3: Key figures for underpricing Denmark
Norway
Obs = 43 Underpricing_raw Underpricing_adjusted Underpricing_offersize
Mean 0.0250 0.0191 0.0058
Standard Error 0.0120 0.0116 0.0040
Median 0.0204 0.0093 0.0026
Standard deviation 0.0786 0.0760 0.0262
Sample Variance 0.0062 0.0058 0.0069
Highest initial return 0.2316 0.2316 0.0882
Lowest Initial return -0.1567 -0.1353 -0.0601
Table 4: Key figures for underpricing in Norway
Sweden
Obs = 39 Underpricing_raw Underpricing_adjusted Underpricing_offersize
Mean 0.0600 0.0668 0.0309
Standard Error 0.0168 0.0162 0.0073
Median 0.0596 0.0524 0.0324
Standard deviation 0.1049 0.1012 0.0458
Sample Variance 0.0110 0.0102 0.0021
Highest initial return 0.3800 0.3845 0.1306
Lowest Initial return -0.1526 -0.1539 -0.0627
Table 5: Key figures for underpricing in Sweden
Denmark has the highest average underpricing of the countries in our sample with a market adjusted underpricing of 9.38% compared to 1.91% and 6.68% in Norway and Sweden, respectively. It is worth noting that the Danish sample is also the smallest one with 7 observations as opposed to 43 in Norway and 39 in Sweden. As a result, underpricing found in Denmark may not be representative for the actual average underpricing in the country over our sample period. The finding is also contrary to the findings of previous research, which has found Denmark to have the lowest level of underpricing when
considering the Scandinavian countries.
45
The highest observed underpricing is found in Sweden, with the initial return being 38.45%, compared to 23.16% and 25.19% in Norway and Denmark, respectively. The highest overpricing on the other hand is also found in Sweden, with initial return being -15.39%, while the lowest initial return in Denmark is -8.05% and in Norway -13.53%.
Previous research has shown that IPO underpricing may vary greatly over time.
Figure 6: Average underpricing by year
Our data consist of one year (2008) in which the initial return was negative, that is there was average overpricing in the market. Note that 2008 only had one observation, indicating that it might not be representative for the overall underpricing in the market that year. The single observation may be coincidental to the fact that 2008 was one of the peak years of the financial crisis. The observed initial return is negative, indicating that the issue was overpriced and supporting the notion that financial markets were illiquid and unstable. 2015 was the year that experienced the highest average
underpricing; a result deemed representative as 2015 have 20 observations. The other years vary from a low in 2013 of 0.41% to 8.80% in 2012. In 2009, there are no observations in our sample. The lack of observations is partly due to no IPOs being conducted on Oslo Stock Exchange, which is not surprising as 2009 was the second peak year of the financial crisis. The few IPOs conducted on the Danish and Swedish exchanges were not included in the sample as they were secondary listings, fixed price offerings or the prospectus was not available.
46
The varying underpricing might imply a time trend in the data, which may influence regression results and controlling for years in the regression might be essential. However, as there is large variation of observations each year and no clear time trend, the yearly means might be of little relevance and we will not control for the year of the IPO in our regression model.
Finally, we investigate whether underpricing varies across industries.
Figure 7: Average underpricing by industry
The industries in our sample are classified according to the Global Industry Classification Standard (GICS) (MSCI, 2016). The Consumer Staples category saw the highest underpricing in the period of interest with an average of 10.45%. However, there is a mere 4 observations within this industry, which may imply that the results are not representative for the industry as a whole. This is also a problem in the Materials and Utilities industries which has 3 and 2 observations, respectively. Utilities is the industry with the lowest level of underpricing in our sample (0.57%), but the low number of observations makes the observation quite unreliable.
Information Technology has the second highest underpricing with an average of 8.93%. In the middle of the spectrum we have Consumer Discretionary and Industrials with average underpricing of 5.26%
and 5.69%, respectively. In the lower end we see Financials, Health Care, Materials and Energy, with underpricing ranging from 1.67 – 3.03%. The illustration gives a good indication of the fact that underpricing varies a lot across industries, which potentially reflects the different risk profiles.
47