In this paper, I investigate how the IPO underpricing changed at the London Stock Exchange Alternative Investment Market over the period of six years (2004 - 2009).
The list of all IPOs that were launched on the AIM is available from LSE official website - New Issues and IPO Summary Spreadsheet. A total of 1449 IPOs were undertaken at the London Stock Exchange during this period, of which 1089 IPOs were on the Alternative Investment Market.
There were made several common exclusions from the data, consistent with the previous research of the IPO underpricing. IPOs underwritten on best effort offers, REITs (real estate investment trusts), investment trusts and other financials were excluded from the sample. Once all the exclusions were made the sample contained 780 IPOs. Using Microsoft Excel a sample of 150 was randomly generated, including 75 IPOs during the pre-crisis period (2004 - 2006) and 75 IPOs of the crisis period (2007 -2009).
LSE New Issues and IPO Summary Spreadsheet only contained the names of the IPO companies that made it inconvenient for the use in Datastream database. Therefore at first all company names were matched one by one with the Datastream codes. Some companies from the first random sample were not in Datastream and they were replaced by those available in the database in order to reach the target sample of 150. The missing companies were mainly dead companies and a certain survival bias can be observed in the sample. However the main interest of the paper is the short-run performance of the IPOs and the above mentioned bias should not be expected to have a significant effect on the results.
Using the list of Datastream codes the data on the prices on the first day, first week and first month of trades of IPOs was collected from the Datastream database. Company sales and pre-tax income one year prior to the IPO were also obtained from the Datastream.
The age of the company at time of the IPO was collected from company prospectuses and Fame database. Data on the IPO gross proceeds was obtained from the Securities Data Company Platinum (SDC) database and company prospectuses. The information whether the company was classified as a technology or not was acquired from SDC.
Information about the NOMADs that were underwriting issues was available in the LSE New Issues and IPO Summary Spreadsheet; all missing observations were filled using company prospectuses. All further information about the firms (net proceeds from IPO, market value of the company at the time of IPO, number of retained stocks, number of new and secondary stocks in the issue) was collected individually from the company prospectuses available on the Thomson Research Database.
The FTSE AIM All Share Index returns were available from Datastream.
Methodology
A large empirical literature exists documenting the phenomenon of IPO underpricing. This anomaly is widely analysed and a great deal of research papers exists.
Typically, short-run underpricing refers to the initial returns on the first day of trades. The underpricing is calculated as the return on the first trading day using a standard methodology as follows:
rit= (Pi1 - Pi,0)/Pi,0 (1)
where rit is the initial return of IPO;
Pi,0 is the offer price of company i;
and Pi,1 is the closing price of IPO i on the first trading day.
However, initial returns can also be calculated, using the opening price of the IPO on the first trading day or the return on equity after the first week or first month of trades.
What concerns the factors, that can explain changes in underpricing, different researchers have different points of view.
Loughran and Ritter (2002) suggest 3 hypotheses for the underpricing changes over time: the changing risk composition hypothesis, the realignment of incentives hypothesis, and the changing issuer objective function hypothesis.
The first theory is the changing risk composition hypothesis, which predicts, that the underpricing surges as the proportion of the riskier stocks increases. In other words, increased underpricing in the recent years is primarily attributable to the greater riskiness of the companies wishing to go public.
Accordingly, the variables, that reflect riskiness of the firm, are age of the company, size of the company, standard deviation of the first month returns on the stock and the industry in which the company operates.
Age
Empirical evidence by Muscarella and Vetsuypens (1990) and Loughran and Ritter (2002) confirms that the underpricing is higher for the younger companies, which can be explained by the fact that the younger companies are more difficult to valuate than old established companies.
Several types of variables for age can be used. For instance, Lowry and Shu (2002) use a dummy variable for age, which is equal to one in case the age of the company going public is more than 5 years and zero otherwise. However they found this variable insignificant. Instead, Cliff and Denis (2004) used logarithm of the firm age at issuance and Loughran and Ritter (2004) employed natural logarithm of the firm's founding date as of the IPO, and both found age variable significant. In this paper, I will use the measure, suggested by Loughran and Ritter (2004): Ln(1+AGE).
The testable hypothesis for the age would be as follows:
H1: Firm age is negatively related to IPO initial returns.
Industry
The industry classification is usually very broad: technology and internet-related stocks versus all others (Loughran and Ritter, 2002). Such classification became popular when trying to explain the dot com bubble during 1999-2000. However, it is still interesting to examine if there is any difference in pricing technology related companies and the rest. While some authors used the SIC codes (Bradley and Jordan, 2002) to identify whether the company belongs to the technology sector, the other used SDC tech dummy (Lowry and Schwert, 2002). SDC tech dummy is an indicator variable set to one if the IPO firm is categorized as high technology in the SDC New Issues Database and zero otherwise. In this work, I will use the SDC classification for the variable TECH.
The hypothesis to test here is
H2: Tech dummy has a positive relationship with the initial returns.
Size
The size of the issue also matters as the smaller companies tend to be more underpriced, than the large companies. Michaely and Shaw (1994) defined size as logarithm of IPO proceeds; Lowry and Shu (2002) used market capitalisation as the proxy (Market Value at close of 1st day of trading); Loughran and Ritter (2004) use natural logarithm of the pre-issue book value of assets.
Purnanandam and Swaminathan (2004) used the natural log of sales for the fiscal year ending at least three months prior to the IPO date as a control for size. Goergen et al (2007) also used the average pre tax profits (or losses) for the last three years before the listing as a proxy for quality and risk of the firm.
All of these variables are related to the size of the company going public. In this paper several proxies to measure the size of the company were used: IPO gross proceeds (GPR), IPO net proceeds (NPR), market value of the company at time of the IPO (MV), natural logarithm of sales one year prior to IPO (LN(Sales)), pre-tax income one year prior (INCOME). All of these variables are expected to have positive coefficient.
The underlying hypothesis would be
H3: Firm size is negatively related to IPO initial returns.
Standard deviation of returns in the aftermarket
Ritter (1984) used daily standard deviation of returns in the aftermarket as the alternative to capture the riskiness. According to his later paper (Ritter, 1987), standard deviation variable (SD) is related to ex ante uncertainty and it can be suggested that firms with volatile stock prices are firms whose market value was highly uncertain before public trading began. The more volatility is present in the stock during the first month the more uncertain investors are about it and the higher underpricing should be. In order to calculate the daily standard deviation of returns, first month closing prices are used, obtained from the Datastream database.
The working hypothesis would be
H4: Standard deviation of returns in the aftermarket is positively related to initial returns.
Pure primary dummy
The second theory was introduced by Ljungqvist and Wilhelm (2003) and argued that recently the issuers had fewer incentives to bargain for a higher offer price and therefore the higher levels of underpricing became acceptable. One of the possible explanations for that is the fact that fewer IPOs now tend to contain secondary shares and therefore the issuing firms have increasingly acquiesced in leaving money on the table (Loughran and Ritter, 2002). To measure this effect the pure primary dummy (PURE) should be introduced, which is equal to one in case the issue contains only primary shares and is zero if there are also secondary shares in the issue. Apart from that, another variable that is used in the paper is the proportion of the new shares in the issue.
The expected coefficients on the pure primary dummy and the proportion of the new shares in the issue are positive according to realignment of incentives theory. Another possible explanation is that if the issue contains secondary shares, there is less uncertainty about the whole issue and the IPO underpricing should be less severe. In this case a positive sign on the pure primary coefficient should also be expected.
The testable hypothesis is
H5: Pure primary issue is positively related to the initial underpricing.
Underwriter reputation
Another theory suggested by Loughran and Ritter (2002) is the changing issuer objective function hypothesis, which claims that firms readiness to accept higher underpricing has increased. Two main explanations for this hypothesis are presented: the analyst lust and corruption hypothesis. According to the analyst lust hypothesis issuing firms are willing the highly ranked underwriter to place their issue and therefore pay less attention to the underwriters' reputation for excessive underpricing. Corruption hypothesis claims that spinning practice became prevalent and created an incentive to seek underwriters with a reputation for severe underpricing (Loughran and Ritter, 2002).
To measure underwriter reputation Carter and Manaster (1990) used ranking variable based on placement in offering "tombstones" with scale from 1 to 9. Loughran and Ritter (2002) updated measure of Carter and Manster's (1990) underwriter quality. In their paper the prestigious underwriter dummy variable equals one (zero otherwise) if the IPO's lead underwriter has a rank of 8 or above on the 0-9 Carter and Manaster (1990) scale. Coakley et al (2005) defined high-prestige underwriters as those listed in the top-ten of the annual Hambro underwriter rankings.
Espenlaub and Khurshed (2009) used dummy variable for reputable NOMAD that is among top five in ranking based on numbers of issues backed in year prior to IPO.
In this paper, two types of rankings similar to the Hemscott AIM Advisers Rankings Guide were used to measure the reputation of the NOMAD. As this report ranked underwriters using two measures (number of clients and client market capitalisation), two dummies for underwriter quality were included.
In order to make the first ranking, the total number of IPOs that were underwritten by each NOMAD during the year prior to IPO was calculated. Then each NOMAD was given the rank and top 5 underwriters of the ranking were considered as reputable. The dummy variable NOMAD1 was then used, which was set to one in case the NOMAD was among reputable underwriters or zero otherwise. N1 was expected to have a positive coefficient.
The second type of the ranking was based on the total market capitalisation of the IPOs the NOMADs underwrote one year prior to the particular IPO. The rank was assigned to each NOMAD and dummy variables NOMAD2 and NOMAD3 were used, that were equal to one in case the NOMAD was among top 5 or top 10 respectively and zero otherwise. Both dummies were also expected to have positive coefficient.
The hypothesis necessary to test was
H6: Underwriter reputation is positively related to IPO initial returns.
Overhang
Some authors suggest that IPO initial returns can also be explained by the ratio of retained shares to the public float, so called overhang (Bradley and Jordan, 2002). Among the possible explanations is the existence of asymmetric information between the insiders and outside investors. If managers are confident about the future of the company, they will only sell a small fraction of the firm on the IPO and in that way will signal the high quality of the firm. Another explanation is offered by Barry (1989), Habib and Ljungqvist (2001), and Ljungqvist and Wilhelm (2003), who assert that opportunity cost of underpricing to issuers is less if the relative float is small. The scarcity value hypothesis suggests that if the overhang is low, i.e. the number of IPO shares is less than the shares retained by the pre-issue shareholders, the market price rises, which causes high initial returns (Ofek and Richardson, 2003).
Loughran and Ritter (2002) used the following definition of share overhang: the ratio of retained shares to the public float (the number of shares issued). Lowry and Murphy (2007) offer a similar definition. The OVERHANG variable in this paper will be derived in a similar way. The working hypothesis will be as follows:
H7: The overhang size is positively related to the IPO initial returns.
Lagged 15-trading day FTSE AIM All Share Index
Some empirical studies also suggest that the IPO initial returns can be forecasted based on the prior market movements (Loughran and Ritter (2002), Smart and Zutter (2003), Unlu et al (2004).
The first-day IPO return is usually larger when there has been a positive market return in the previous two-three weeks before the IPO and vice versa. Therefore it is necessary to control for market performance.
Loughran and Ritte (2002), and Coakley et al (2005) use similar approaches, in which lagged index variable measures the percentage return on chosen index during the 15 trading days prior to the IPO. Smart and Zutter (2003) use return over 22 trading days preceding the initial public offer.
In this paper I will use 15-trading day lagged FTSE AIM All Share Index (lagged FTSE RETURN).
The working hypothesis would be
H8: The IPO initial returns are positively related to the performance of the FTSE AIM All Share Index.
Crisis dummy
World financial crisis that started in 2007 led to the collapse of large financial institutions and emergency-type government bailouts became a normal thing. Financial turmoil resulted in a burst of the bubbles on the stock markets around the world. The Alternative Investment Market of the London Stock Exchange was not an exception.
In order to evaluate the impact of the world financial crisis on the IPO underpricing crisis dummy should be introduced, that is equal to 1 in case the IPO was launched during the period 2007-2009 and zero otherwise. The hypothesized relationship between the bubble burst and underpricing is negative. It is expected that there is less speculation on the stock market as the investors are more cautious during the crisis and therefore IPO underpricing decreases after the bubble bursts.
The working hypothesis would be
H9: Crisis dummy is negatively related to the initial returns.
In summary, in order to examine how the underpricing changed over time and which factors account for that, I will use the regression analysis. The following multiple regression
INITIAL RETURN = b0 + b1 Ln(1+AGE) + b2 TECH + b3 SIZE +
+ b4 SD + b5 PURE +b6 NOMAD+ b7 OVERHANG+b8 FTSE RETURN b9 CRISIS + e
will be estimated using Ordinary Least Squares procedure.
Purnanandam, Amiyatosh K., and Bhaskaran Swaminathan, 2004, Are IPOs really Underpriced? Review of Financial Studies 17, 811-848.
