Alternative Investment Market is a part of the London Stock Exchange market and it is designed for small and emerging companies to raise money from the public. The purpose of this report is to find and analyse the variables affect the non-financial companies IPO initial return in AIM between 2005 and 2010.

Data and Sampling procedure

Our main sources of information are the London Stock Exchange website, Perfect filings and Bloomberg, where we obtained all the necessary company information for analysis. First of all, we selected all the companies which started floating on the AIM between January 2005 and December 2010 from the London Stock Exchange Website, there were 1142 companies in total. After excluding all the financial-related companies and outliers which had exceptionally high initial returns, we then split up the remaining companies into two groups according to their listing year to observe the impact of financial crisis. We assumed the beginning of 2008 is the breaking point of the crisis. Secondly, we applied the stratified random sampling procedure to select companies respectively for each sample. We obtained two samples of 46 companies each, for which Sample 1 consists of companies that were listed before 2008 and Sample 2 consists of companies that listed after 2008.

Descriptive statistics

Regarding to our two samples, on average, companies from Sample 1 have raised £160.21m and companies from Sample 2 have raised £231.33m from the AIM where INTERNATIONAL FERRO METALS and SKIL PORTS & LOGISTICS LTD are the highest-valued IPO deals within their sample which rose £800m and £760m respectively. In Sample 1, the offer price ranges between 2p and 503.7p and has a mean of 102.92p (See Appendix Table 1). On the other hand, the offer size of Sample 2 ranges between 0.6p and 13000p with a mean of 385.83p in Sample 2 (See Appendix Table 2). Lastly, Sample 1 has a standard deviation of 92.141 where that of Sample 2 is 1904.855, this huge difference is mainly due to the 13000p offer price issued by TGE MARINE AG.

Methodology

Formulae for Initial return:

Firstly, we calculated the initial return (ri) for each company and this was computed by finding the difference between the closing price on the 7th day (Pi) after flotation and the offering price (OPi) and then divide it by the OPi. We then computed the average returns and the standard deviations of the initial returns for each sample.

Formulae for Mean and Standard Deviation:

(Note: n is the number of companies in each sample)

In order to test whether the initial return of each sample is statistically different from zero, we performed a t-test (since the population variance is unknown) to check whether the mean initial return of each sample is different from zero, and the Null Hypothesis is expressed as follow:

Furthermore, we applied another t-test to assess whether the initial returns of our 2 samples are statistically different. Since it is not reasonable to assume the population standard deviations of the two samples are equal and at the same time they are unknown, we apply the t statistic and the Hypothesis Test as follow:

After the tests, we proceeded to the next stage where we analysed the factors and variables that would affect the initial returns of IPOs on the AIM. We performed a Linear Regression Analysis to explore the correlations between our chosen independent variables and the initial returns whilst taking into account of the additive and multiplicative dummy variables which would represent the qualitative variables [1] .

Results

Initial returns

We tested each sample individually and we found that the mean of each sample is significantly different from zero (at 5% significant level) and in general the closing prices of the stocks after 7days are higher than the offer prices (See Appendix Table 3-5). This is reflected in the positive means of initial returns in the two samples. Hence, the companies in both samples are on average underpriced and when investors buy an IPO and hold it for a week, they can still obtain a minimum of 10% positive return.

We also took another test to check whether the initial returns of the two samples are significantly different from each other. However, the result is insignificant (at 5% significant level) which reflects the 2008 financial crisis has no real impact on the initial returns of IPOs after 2008 (See Appendix Table 6).

Since the test results show that the IPO’s are underpriced, we identified several factors both quantitative and qualitative that might contribute to such outcome.

Money left on the table

Money left on the table is the difference between the first day closing price and the offer price and multiplied by the number of shares. It is the total amount of money investors are willing to pay over the original money required by the issuers. According to Ritter and Loughran (2004), if the investors are paying a premium over the issue price value on the first day of quoting, then it is expected that the short-term returns on the investment on those companies should be higher. Our results (coefficient 2.10e-09) (See Appendix Table 7) indicated that money left on the table has a positive impact on the return obtained by investor on a week after the issue date, and such low value is explained by Money left on the table usually is a big amount on money in comparison with percentage return registered on the initial return. It has a p-value of 0.000 at 5% significant level, hence it should be included in the model.

Market value

Market value takes into account the market capitalization and the debt outstanding in the company. Market value can be defined as the company size by proxy. Ibboston and Ritter (1994) found that in short run there is a negative impact on the return on the investments of new companies when the companies are larger and it is presumed that there is less uncertainty. According to our model results, Market value has the expected negative effect on initial return, with coefficient of -6.27e10 and it has a p-value of 0.000 (See Appendix Table 7) which is highly significant, so we are keeping the variable in the model.

Market index return

According to Boubaker (2011), market return over a period has a significant effect on the IPO listing. The market index acts as an investor behaviour sentiment indicator, so the return of stock on that day is highly correlated to the return on the index and therefore contributes to the underpricing or overpricing of IPOs. The linear regression gave the market index (AXX Index) return a p-value of 0.009 and a correlation coefficient of 1.57 (See Appendix Table 7) which we were able to prove that it is a significant independent variable that has a positive correlation with the initial return of IPO.

Earnings before IPO

Profit of a company is a very important factor for investor to consider as it represents the profitability of the firm and would ultimately affect investor’s expectation and return. As the study paper (Rhee, 2002) prepared for the OECD Round Table meeting shows there is a 30% difference in returns for IPOs between companies with positive and negative earnings. We took this as an independent variable and obtained the profit figures from every company’s annual report in our samples. The result shows that the coefficient is and p-value of this variable is 0.347 (See Appendix Table 7), hence it is insignificant at the 5% significance level. We found no evidence for effects of earning management to initial returns for IPO.

Number of shares traded after 7 days to number of shares issued

The number of shares traded on the day has an impact on the return on share, Cornelli et al (2006) discovered that the total volume of transaction is positively correlated to investor behaviour which leads to higher prices, and hence higher initial returns. On the basis of our two samples extracted from the AIM, we are unable to establish these findings, since the regression result demonstrated the p-value of this variable is 0.145 which is higher than the 5% significant level (See Appendix Table 7). Therefore, there is no correlation between this factor and the initial return and it should not be incorporated into our model.

Dummy Variables

Regression with dummies and interactive dummies:

We considered 3 models:

Model 1(dummies only):

Model 2(interactive dummies only):

Model 3(both dummies and interactive dummies):

(Note: The full outcome of regression analysis is shown in the Appendix Table 8-10)

Monday effect dummy variable

On the secondary market, there are several anomalies which are not following the Efficient Market Hypothesis (EMH) and capital asset pricing model (CAPM), such as the January effect (Haugen and Lakonishok, 1998; Ligon, 1997), the holiday effect (Ariel, 1990) and Monday effect (Jones and Ligon, 2009). IPOs offered on Monday are claimed to have higher initial return comparing to the other days in the week. We adopt this variable and take it as a dummy in our model where companies listed on Monday are denoted as "1" and companies listed on other days are "0". Research shows that calendar effect is not apparent in our samples with a p-value of 0.447 and a coefficient of 0.0238259. The result turns out to be insignificant at 5% significant level and it is inconsistent with the findings mentioned above.

Sector dummy variable

Ritter J.R. (1984) suggested that the returns are clustered according to sectors, therefore it is possible that sector variable can influence the return of the stock so we decided to use sector as a dummy variable. We used "1" for companies that belong to the energy and commodities industry and "0" for the rest. From the regression results, we were not able to identify a regular pattern that would justify the sector dummy variable having an impact on the initial returns. The sector dummy has a p-value of 0.211 which is not significant.

Country dummy variable

Demaskey and Olson (2006) explained that by disaggregating the return of ADR’s according to the country of origin, there are significant differences in initial and aftermarket performances. We decided to use country origin of the company as a dummy variable and put "1" for UK companies and "0" for others. The p-value is 0.402 which suggests that the country variable is insignificant for the sample used.

Underwriter ranking dummy variable

Lowry et al (2010) believed that the actions (i.e. price revision) of underwriter would affect the pricing error of IPOs. They also mentioned that the higher the reputation of underwriters, the better the after-IPO service. In our model, rather than using Carter and Manaster (1990) underwriter ranking score as Lowry et al did, we chose another updated 2007 underwriter top 50 ranking from Bloomberg, and assumed the ranking of underwriter is constant from 2005 to 2010. We marked "1" for the dummy variable if the underwriter of IPO is in the list. As a result, it has a positive correlation coefficient and the p-value is 0.25 which is not significant.

R2 increases from Model 1 to Model 3 because the more the variables, the higher the explanatory power of the model which implies that R2 is not a reasonable indicator. We then looked at the adjusted R2, and discovered that Model 1 gave the largest adjusted R2 (26.36%), which means putting aside the number of variables, model 1 best explained the dependent variable. Considering the independent variables are significant only in Model 1, we believe it is the best model among the three. Although the adjusted R2 of the original model and Model 1 are similar, we prefer the original model since none of the dummy variables in Model 1 is significant.

Final insights

This study aims at analysing the pricing of the IPOs on AIM and there are several points that we would like to address.

Before conducting any tests, we anticipated that the financial crisis in 2008 would adversely affect the performance of the IPOs, but surprisingly the impact is negligible based on our hypothesis test afterwards.

We tested the IPOs in our samples and proved that they were generally underpriced, and this is consistent with the past studies. We also investigated five quantitative variables and four qualitative variables and ultimately, we restricted and finalised our model with only three explanatory variables which are money left on the table, market value and market index return. Based on our analysis, these three variables have significant effects on the initial returns of IPOs and these are consistent with other empirical findings. However, for the factors that were assumed to be influential for underpricing such as earnings, the listing day, trading volume, sectors, country origin of the company and broker rankings are not significant determinants of the initial returns of the IPOs and this contradicts to the literatures we were referencing to. Finally, we suggest investors to invest in IPOs on the AIM when the following conditions are met: high market index return, companies with small market value and big money left on the table.

Appendix:

Table 1

Table 2

Table 3Table 4

Table 5

Table 6

Table 7

Table 8

Table 9

Table 10