Investigation On Lastminute Coms Ipo And Subsequent Performance Finance Essay - Free Essay Example

Sample details Pages: 21 Words: 6155 Downloads: 4 Date added: 2017/06/26 Category Finance Essay Type Cause and effect essay Did you like this example? The short-run underpricing and long-run underperformance of initial public offerings (IPOs) are the best documented anomalies in finance. This paper investigates if these anomalies exist in the aftermarket performance of the IPO of an internet-based company, Lastminute.com plc. Further analyses are then carried out to explain the performance results. Don’t waste time! Our writers will create an original "Investigation On Lastminute Coms Ipo And Subsequent Performance Finance Essay" essay for you Create order Evidence are presented that the companys IPO exhibits positive abnormal returns on the first day of trading. There are several theories of underpricing do not apply in explaining the short-run performance of the company. The long-run underperformance of IPO, which found in previous research, is not supported by the evidence of this study. The main problem of the long-run performance investigation is the difficulty of justifying systematic long-run underperformance in a reliable manner, especially during the bubble period. Numerous studies have documented two well-known anomalies in initial public offerings (IPOs). One is that the IPOs provide large positive abnormal returns in the initial days of trading (Ritter, 1987; Levis 1990). This anomaly has been observed in almost every finance markets of the world. The other puzzle is IPOs appear to be overpriced in the long run. For example, Ritter (1991) provides evidence that US IPOs significantly underperform in the 3 years following the offering. However, the international evidence on the long-run performance of IPOs are less extensive and unanimously conclusive than the one on underpricing. The purpose of this study is to investigate the short-run as well as the long-run performance of an internet-based company, Lastminute.com plc which floated before dot-com bubble burst by using the event study methodology on IPOs. Besides, this clinical study provides further analyses and possible explanations for the observed results of the analysed firm based on the theoretical models in academic literature. I use market adjusted abnormal returns (MAARs) to measure short-run performance and the intercept from Fama-French (1993) three-factor model, i.e. Jensens alpha ( to measure average monthly abnormal return for different periods in the long run. These returns are adjusted by different benchmarks. Moreover, buy-and-hold abnormal returns (BHARs) are computed as a complementary analysis for the results from the Fama-French three factor model. To summarize the empirical findings of this paper, first, the results provide evidence supporting the general robustness of the prior finding with regard to the short-run underpricing of the Lastminute.coms IPO. There is significant and abnormally high first day return for the studied firm regardless the selection of benchmark. The short-run performance appears to be best explained by the high speculation level over the issue. The Rock (1986) model, the signalling hypothesis and the underwriter prestigious hypothesis are not the appropriate elucidations for the abnormal initial return results. Second, unlike previous research, the interesting finding is an inverse relationship between the periods after the firm floated and its poor post-IPO performance. The results from Fama-French three-factor model are not consistent with the initial prediction based on the academic literature. I find that Lastminute.com outperforms the market indices in the long run in spite of the signs of underperformance in the first year of trading. Results also show that the company performs better than other technology companies when the period tested is longer than 1 year. These results, nevertheless, are all statistically insignificant and therefore it concludes that the company of interest does not exhibit abnormal performance in the long run. Furthermore, the earnings management analysis does not indicate that there are accounting manipulations in the firm prior to the flotation. The dot-com bubble period, 1999-2000 period is extraordinary for the simple fact that both valuations and underpricing of firms simultaneously skyrocketed. It is important to note that Ljungvist and Wilhelm Jr. (2002) provide some evidence for the unique firm characteristics and aberrant pricing behaviour during the period. This paper, however, does not intend to examine the effect of the bubble on asset pricing and investor behaviour. I leave this for other researchers. The rest of my paper is arranged as follows. Section 2 provides the background for the studied firm. Section 3 reviews the IPO academic literature on short-run underpricing and long-run underperformance. Section 4 describes the data, methodology and hypothesis used in the investigation. Section 5 set out the short-run results and analysis. Section 6 set out the long-run results and analysis. Section 7 discusses the limitations and possible future research for the paper. Section 8 concludes the paper. 2. Background of the clinical study The company under investigation, Lastminute.com plc is an online travel agency and e-tailer which specialises in selling inventory such as package holidays, flights and consumer products at cut-rate prices to website customers. It was founded by Martha Lane Fox and Brent Hoberman in 1998. The site became an instant hit with internet travelers, garnering enough business and rapid growth to allow the company to offer stock on the London Stock Exchange in March 2000. The purpose of the offering, according to its prospectus, was to raise approximately 113.5 million. Half of the amount would be used to increase sales and marketing activities, and the remaining for product development and expansion such as broadening supplier base, potential acquisition in UK and internationally. In addition, the company acknowledged in the prospectus that it was not profitable and expected to incur future losses. The major risk factors were reliance on third parties, intense competition and e-commerce uncertainties. Lane Fox and Hoberman became the icon of UK internet business entrepreneurs during the British dot-com boom. The company timed its launch to perfection and floated at the peak of the internet bubble. The issues received high interest of public and the share price spiralled as high as 555p from the offer price 380p on the first day of trading. Following the bubble burst, subsequently, Lastminute.coms stock sank and lost nearly half of its value within three weeks. Lastminute.com, however, unlike hundreds of other first-wave internet companies, it survived the bubble burst and stock market clash in 2001. Moreover, it continues to thrive and grow through expansion and shows reducing loss in its financial accounts. Lastminute.com is still considered as one of the few successful internet-based businesses in the relatively unstable world of internet. 3. Literature review 3.1. Empirical evidence and theories of short-run IPO underpricing A number of studies indicate the mispricing of IPOs which tend to yield substantial returns in the days immediately following issue. Stoll and Curley (1970) are the avant-garde to show the systematic abnormal first-day returns of IPOs. Substantial international evidence show that IPO underpricing has become a worldwide phenomenon. For British IPOs, the studies of Dimson (1979), Buckland, Herbert and Yeomans (1981), the Bank of England (1990), Jenkinson and Mayer (1988) and Levis (1993) exhibit average first day returns ranging from 8.6% to 17%. Ritter (1987), Welch (1989), Ibbotson, Sindelar and Ritter (1994) and Rajan and Servaes (1997) provide evidence suggesting that the existence of average initial returns of up to 16% has been a regular feature of the US new issue market. Lee, Taylor and Walter (1994), Jacquillat (1986), Kaneko and Pettway (1994) and Ljungqvist (1997) among others provide evidence of abnormal returns of up to 14% in the developed markets of the world such as Aus tralia, France, Japan and Germany. The literature abounds with a variety of conjectures that purport to explain the observed underpricing in IPOs based on the economic realities in the marketplace. Rock (1986) model provide a fundamental and convincing explanation for IPO performance by applying winners curse hypothesis to the new issue market. The model classifies investors as either informed or uninformed. The former are those who spend to assess the potential performance of the new issue, whereas the latter do not spend resources on the evaluation of the stock. Informed investors tend to crowd out uninformed investors for the underpriced and possibly lucrative issues. Consequently, uninformed investors hold a disproportionately large amount of overpriced IPOs, as informed investors may not subscribe. Uninformed investors would leave the market as they would systematically make losses. In order to keep the uninformed investors in the IPO market and to compensate their expected losses, the investment bankers have to offer the securities at discounts from their expected after-market prices. The studies by Koh and Walter (1989), Keloharju (1993), and Levis (1990) produce results that are consistent with and thus add further weight to the Rock model. Some theories propose the underpricing as a signalling mechanism of the firm quality. Allen and Faulhaber (1989) conclude that IPO underpricing is a credible indication of a firms post issue prospects. It is assumed that the underpricing in the firms initial offering is an immediate loss to the initial owners and companies with favourable position and performance in the after-market will be able to recoup the loss. These good companies underprice their IPO, because by doing so they direct investors to a favourable subsequent dividend results. Nevertheless, the assumption that company directors are willing to accept the initial loss on the IPO and renounce larger potential funding is doubtful in practical realities of market. Ritter (1984) suggested that gross underpricing may be a result of the monopsony power of the investment bankers in underwriting common stocks of small speculative firms. These investment bankers intentionally underprice the securities and only are allocated to favoured large customers who regularly buy a variety of investment services from the investment bank and thus pay unusually high brokerage fees. Tinic (1988), however, found evidence that the monopsony power hypothesis is not adequate for explaining underpricing behaviour. Johnson and Miller (1988) advance another explanation for underpricing. They analyse the prestige of underwriters and the level of underpricing. The finding is that the greater the underwriter prestige, the lower the degree of underpricing and it also works in the opposite way. This suggests the more prestigious banks require less underpricing to attract investor interest because they deal with lower risk issues. Beatty and Welch (1996) challenge the underwriter prestige hypothesis and show the inverse relation recently reversed for small firms. Muscarella and Vetsuypens (1989) find that investment banks themselves that go public are underpriced as well. The speculative bubble hypothesis claims that large excess returns of the IPOs are attributed to the investors who could not get allocations of the oversubscribed new issues from the underwriters at the offering prices. The speculative appetites of these investors then speculate temporarily inflating the price of the new issues in the aftermarket. When speculative demand diminishes, this speculative bubble should burst and negative excess returns are expected on the post-IPO share. Tests on aftermarket returns by Ritter (1984) and Tinic (1988) could find no evidence supportive of this hypothesis. Tinic (1988) and Hughes and Thakor (1992) argue that IPO underpricing used as a form of insurance to reduce legal liability by both issuers and underwriters. This theory implies that a greater degree of underpricing occur to prevent investors experiencing significant losses on IPOs, as a result they are willing to neglect small errors such as omission of data, inadequate nature of data supplied for the disclosure requirement and thus not led to legal actions. However, Drake and Vetsuypens (1993) criticize and reject this hypothesis by showing evidence that underpriced IPOs are just as likely to be sued as overpriced IPOs and that there is no significant difference in underpricing between sued and non-sued firms. 3.2 Empirical evidence and theories of long-run underperformance The long-run underperformance of IPOs is a well-known perplexity in IPO literature and has attracted much attention from either investors or academic researchers in recent years. From an investors viewpoint, the existence of price patterns may present opportunities for superior returns using active trading strategies. A finding of non-zero market performance would also call into question the informational efficiency of the IPO market which proposes after-market stock price should appropriately reflect the shares intrinsic value. Several authors have examined the returns on IPOs during the three years after going public for a number of countries. The international evidence of long-run underperformance is summarized in Table 1. Table 1 International Evidence on Long-run IPO Overpricing Country Reference Sample size Time period Total abnormal return Australia Lee, Taylor Walter (1996) 266 1976-89 -46.5% Austria Aussenegg (1997) 57 1965-93 -27.3% Brazil Aggarwal, Leal Hernandez (1993) 62 1980-90 -47.0% Canada Jog and Srivistava (1994) 216 1972-93 -17.9% Chile Aggarwal, Leal Hernandez (1993) 28 1982-90 -23.7% Finland Keloharju (1993) 79 1984-89 -21.1% Germany Ljungqvist (1997) 145 1970-90 -12.1% Japan Cai Wei (1997) 172 1971-90 -27.0% Korea Kim, Krinsky Lee (1992) 99 1985-88 +2.0% Singapore Hin Mahmood (1993) 45 1976-84 -9.2% Sweden Loughran, Ritter Rydqvist (1994) 162 1980-90 +1.2% United Kingdom Levis (1993) 712 1980-88 -8.1% United States Loughran Ritter (1995) 4,753 1970-90 -20.0% Sources: Ritter (1998) and various studies cited Notes: Total abnormal returns are measured as , where is the average total return (where a 50% return is measured as 0.5) on the IPOs from the market price shortly after trading commences until the earlier of the delisting date or 3 years; is the average of either the market return or matching-firm returns over the same interval. Theoretical explanations for the long-run underperformance of IPOs are less plenteous compared with IPO underpricing. Miller (1977) provides behavioural and expectation based explanation for the underperformance of new issues. He advances the divergence of opinion hypothesis and suggests that investors who are most optimistic about an IPO will be the buyers. If there is great uncertainty about the value of an IPO, the valuations of optimistic investors will be much higher than those of pessimistic investors. Over time, as the variance of opinions decreases, investors amend their initial share valuations downwards and the price will fall. In tune with this theory, Rajan Servaes (1997) shows that investors gain from initial underpricing suffer poor aftermarket performance. The difficulty of measuring the divergence of opinion, however, resulted in a number of criticisms of this theory. Jain and Kini (1994) provide explanation for the poor long-run performance using the agency costs hypothesis. They investigate the relation between the ownership structure and the long-run performance of IPO and conclude that new issues with greater equity retention by the original shareholders yield better long-run performance. Mikkelson, Partch and Shah (1997), however, show that post-IPO operating performance is not related to the ownership structure. Despite their opposing results, both studies demonstrate that long-run return performance is accompanied by poor financial accounting performance after IPO relative to pre-IPO performance. Teoh, Welch, and Wong (1998) document that IPOs underperformance is attributable to the unusually high discretionary accounting accruals in the IPO-year. They suggest that firms manage their earnings to look good when they conduct their IPO. However, firms find it hard to maintain such accounting manipulations for long periods because accruals reversed over the long run because sum of earnings must equal the sum of cash flows in the long term. Consequently, any higher-than-normal accruals in one period must be reversed. When pre-issue earnings levels not maintained in post- issue periods, the market revises its valuation down and cause a decline in post-issue returns. Risk-measurement hypothesis explains that underperformance caused by the failure to adjust returns for time-varying systematic risk. One of the problems is the choice of benchmark. Dimson and Marsh (1986), Ritter (1991), and Fama and French (1996) and others demonstrated that the measurement of the long-run performance of the IPOs is sensitive to the benchmark employed. These imply that imperfect benchmarking affect the possibility of long-run returns. Brav (1997), Barber and Lyon (1997) and Kothari and Warner (1997) point out that statistical inference conducted using traditional testing methods, such as t-tests is mis-specified. This is a problem for long horizon event studies. Brav (2000) attributes the misspecification to the potentially important violations of the underlying statistical assumptions.[1]A statistical approach to solve this problem is calendar time approach which has been advocated by Fama (1998) and other researchers. Meanwhile, Barber and Lyon (1997) and Brav (2000) and others support the characteristic-based matching approach and address this measurement bias by using size/book-to-market value matched portfolio as their benchmarks. 4. Data, methodology and hypothesis development 4.1 Data The main source of data for the study is the Datastream service. I obtain total return index for calculating the daily returns and month returns of the analysed company in this clinical study, the FTSE All-Share Index, and the FTSE techMARK All-Share Index from this database. Thomson ONE Banker is used to get the lists of historical constituents of FTSE All-Share Index and their respective market value. Besides, annual reports and prospectus of the company under consideration are also obtained from here. Information, news and press reports about the analysed company are generally retrieved from Factiva website. I also analyse the information such as the offer price, the underwriter and the amount raised on the issues from the London Stock Exchange statistical fact sheets.[2]Monthly Fama-French factors data in UK by Gregory, Tharyan and Huang (2009) are used to carry out the analysis of performance. 4.2. Selection of benchmarks There are some benchmarks selected to adjust the returns in tests. The main market index used is the FTSE All-Share Index, which should best represent the performance of London Stock Exchange market and it is used in many UK IPO studies. Furthermore, the FTSE techMARK All-share Index is chosen as the second market index to represent the performance of innovative and technology companies which are similar to the studied company.[3] A reference portfolio matched by size is chosen as the additional benchmark and it is selected from the main market. The reference portfolio selected from the FTSE techMark was attempted but abandoned due to the unavailability of data before April 2001. A matched stock will not be chosen because there is likely to be a large error in any one application (the variation of actual returns around expected returns is typically very large) and therefore it is not appropriate in this clinical study (Strong 1992). I obtain the constituent list of FTSE All-Share Index on the day of the IPO occurred and reconstruct the index by excluding those firms which delisted before the trading day exactly one year from the IPO. Moreover, I exclude the company of interest from the index. (reason?) As in Fama and French (1993), the June market value of common equity (shares outstanding multiplied by June closing price) is measured as the firm size in each year. Size rankings based on market value of equity in year t are then used from year t through year t + 1. Companies without the market capitalization data in June of year t are deleted from the analysis.[4]The companies are then ranked based on their June market capitalization and 10 deciles of equal number of companies are created. In other words, tenth of those companies in the index with lowest market capitalisation are categorised into first decile, the second decile is the next tenth of companies with lowest market capitalization and so forth. The decile portfolio to which studied company would have belonged based on its size in June of the year is selected as its size reference portfolio. This size reference portfolio is changed every year using the same method; i.e. a new reference portfolio is reselected at the trading day exactly one year from the IPO based on the June market value of equity of the year. The monthly return of the size reference portfolio is calculated by averaging the monthly returns across all securities in a particular size decile. The calculation of the size-benchmark return is equivalent to a strategy of investing in an equally weighted size decile portfolio with monthly rebalancing. (Barber and Lyon 1997) 4.3. Short-run performance measurement The initial post-IPO abnormal returns will be calculated as in Khurshed and Mudambi (1999). Firstly, the daily returns of analysed companys stock () and of its benchmarks () are calculated based on their daily total return index (RI) as: ; Where and are the total return index of companys stock and of its benchmarks on the tth day of trading, and are the total return index of companys stock and of its benchmarks on the t 1th day of trading respectively. Using these two returns, the market adjusted abnormal return for the IPO on tth day of trading is computed as: The measure of MAAR does not take into account of the systematic risk associated with the issue. It assumes the systematic risk of the IPO under consideration is the same as that of the benchmark, i.e. beta of the IPO average to unity. Therefore, if the beta of the new listed firm is not equal to one, MAAR is a upward-biased estimate of the IPOs initial performance. However, Khurshed and Mudambi (1999) state that the assumption is unlikely to affect the essence of the results. 4.4. Long-run performance measurement The monthly returns of analysed company and its benchmarks are computed in an analogous manner stated, based on monthly total return index. Cross-dependence problem, as mentioned in Section 3 is considered before the long-run test is carried out. The Fama-French three factor model that will be employed, is a Jensen-alpha approach which is immune to the bias because of the use of calendar-time portfolios. 4.4.1. The Fama-French three factor model The long-run event study tests will be carried out with the use of three-factor model developed by Fama and French (1993). The model is applied by regressing the post-event monthly excess returns for the company of interest on a market factor, a size factor, and a book-to-market factor: Where is the monthly return on the stock of IPO firm, is the one month return on UK Treasury Bills, is the return on the value weighted market index.[5]is the return on a value-weighted portfolio of small stocks less the return on a value-weighted portfolio of big stocks, is the return on a value-weighted portfolio of high book-to-market stocks less the return on a value-weighted portfolio of low book-to-market stocks.[6] Parameter estimates of , , , are obtained using regression analysis, represents the error term in the regression. The parameter of interest is the intercept, (Jensens alpha) which signifies the average monthly abnormal return of the firm over the period. In other words, a positive intercept indicates that the firm has performed better than expected after controlling for market, size and book-to-market factors. Inferences about the abnormal performance are on the basis of the estimated and its statistical significance. 4.4.2. Buy-and-hold abnormal returns (BHARs) Despite the succinctness and the popularity in event studies, Fama-French three factor model is not without weakness. First, Ritter and Welch (2002) point out that the Fama-French three factors are contaminated especially in periods of high IPO issuing, which is the period of this clinical studies. Second, the regression approach assumes that a firms market, size and book-to-market characteristics are stable over time. In contrary, matching portfolio approach allows a firms portfolio assignment to be changed once every year. The method of buy-and-hold abnormal returns in Barber and Lyon (1997) will be adopted as a complementary analysis of the abnormal performance of the company of interest. The return is calculated as the difference between the return on a buy-and-hold investment in the firm under consideration and the return on a buy-and-hold investment in the benchmark with an appropriate expected return: The BHAR approach is more corresponding to the regular investor behaviour. It represents the return of investing in the company under analysis compared to the benchmark stated. Nonetheless, when this approach is used in clinical studies, i.e. a single analysed company only, the statistical significance cannot be tested because the underlying distribution is not clear. Ergo, the results will only be interpreted without statistical test. 4.4.3. Earnings Management Analysis The earning management analysis will be computed as in Toniato (2007) to investigate the earnings management of the company of interest prior to and after the year of IPO. The total accruals for a firm in year t ( comprise of non-discretionary and discretionary portions, it can be derived using the formula: The expectations model for total accruals to control for changes in the economic circumstances of the firm is: Where = change in revenue from year t-1 to year t; = gross property plant and equipment in year t for firm; = total assets in year t-1 for firm; and = residual term in year t for firm. The level of accruals in year t is calculated using the ordinary least squares (OLS) estimates from the regression above as: Knowing that , the prediction error from the OLS regression denotes the level of discretionary accruals for the company. It also represents the proxy for the level of earning management of the company in certain period. 4.5. Hypothesis development There are 2 hypotheses to be tested for the performance of company in relation to the IPO anomalies. First, based on the evidence in section 3 that IPOs are on average underpriced, and the dot.com bubble exacerbates the effect, I test whether the studied companys IPO is underpriced and it performs significantly better than the benchmarks on the first day of issue. The null Hypothesis 1 states that the MAAR of the IPO company is equal or below zero on the 1st day of trading. Second, consider that some business news report that the studied company is a relatively successful internet business, but numerous international evidence demonstrate the long-run underperformance of IPOs, I will study whether the company under analysis over perform or underperform the market within the period of first, second and three years of trading. In other words, I test if there are significant abnormal returns generated over or below those captured by the three factors. The null Hypothesis 2 posits that there is no abnormal performance and the intercept (average monthly abnormal return) is zero. In addition, I expect that the BHAR will be directly related to the results in Fama-French model. However, this method does not allow for a suitable statistical test in clinical study. Therefore, it is merely a supplementary result for the analysis of long-run performance. 4.6. Test statistics To test the null hypothesis that the 1st day MAAR of the IPO firm is equal to zero, I employ the Patell Standardised Residual (PSR) Test by Patell (1976). The parameters are estimated from the observations outside the testing period, i.e. estimation period (EP).[7]The abnormal returns are prediction errors rather than true residuals and should be standardised. A Student t-statistic is calculated using the following formula: Where is an estimate of the variance of the residuals during the EP; reflects the increase in variance due to prediction outside the EP; T = the number of observations in the EP; and The test statistics used in the regression analysis in Fama-French three factor model is based on the time-series variability of the portfolio return residuals and this is obtained from the output of regression test. 5. Short-run results The results of short-run analysis, using measure of 1st day MAAR against different benchmarks are presented in Table 2. Table 2 Short-run performance results Benchmark selection FTSE All-Share Index FTSE techMARK All-Share Index Size Refrence Portfolio (%) 28.00 (4.30)* 28.48 (4.38)* 28.17 (4.33)* t-statistics are presented in parentheses. * Significant at 1%, using a one-tailed test The analysed IPO company, Lastminute.com produces return of about 28% adjusted by all benchmarks in the first trading day. These results are significant at the 1% confidence level. The firm outperforms FTSE techMark All-Share Index slightly higher than the size reference portfolio and FTSE All-Share Index. Overall, the numerical agreements among the 3 sets of results are close and thus the use of different benchmarks does not lead to significant differences in the returns. I conclude that Lastminute.com IPO exhibits positive significant abnormal return on the first day of trading and the return is robust to the choice of benchmarks 5.1 Analysis and discussions for short-run results The results verify the firms IPO underpricing and null Hypothesis 1 is rejected. This implies that Lastminute.com generates substantial returns to its new issue buyer on the first day of listing. Furthermore, we can get different perceptions of the results based on the theoretical models. This single firm example, however, exhibits a quite contradictory result to the Rock (1986) model. As stated in its own prospectus, Lastminute.com has limited operational history and it is difficult for investors to evaluate its business and future prospect.[8]Hence, it is less possible for an informed investor relatively to acquire more information than an uninformed investor. In the absence of the information differential between investors, the firm should have shown a less impressive initial over-performance. The signalling theory provides a mixed support in this study. The firm which experienced underpricing in its new issues, had poor financial and share price performance afterwards. It made millions of losses in following years and its share has never regained the first day high in London Stock Exchange market. Nonetheless, Lastminute.com managed to survive the internet bubble burst and global downturn in the tourism market in 2001 due to the terrorist attack. The company is indeed a successful internet business in the post-bubble period. The Lastminute.com new issues underwriter, Morgan Stanley, is an investment bank which has occupied a leading role in high-quality securities underwriting in the years since the Securities Act of 1933 and garners the highest Carter-Manaster rank of nine.[9]It was expected that a less significant initial return results to be yielded when a top prestigious investment banker would not underprice IPO too much. The high underpricing observed is therefore, inconsistent with the underwriter prestigious hypothesis. The results of underpricing, seems like to be more associated with the level of speculation over the issue in internet bubble period. The interesting finding is the underwriter prestigious hypothesis and the Rock (1986) model work reversely in my studied company. Significant pricing exists even the underwriter is considerably prestigious. The firm with less operational history and therefore less informational differential between informed and uninformed investors generates impressively substantial underpricing instead. It could be best explained by the uncertainty and high level of speculations inflate the share price of the IPO firm. 6. Long-run results Table 3 reports the results of the Fama-French (1993) three-factor time-series regressions. It shows the average monthly abnormal returns of Lastminute.com in 1, 2 and 3 years following its issue. Table 3 Fama-French (1993) Three-factor Time-series Regressions Market Index 1: FTSE All-Share Index 12 months observations -0.09 (-0.81) 3.44 (1.35) 3.49 (1.07) 0.92 (0.40) 0.15 24 months observations -0.01 (-0.15) 2.80 (1.64)** 3.19 (1.77)** 0.20 (0.14) 0.33 36 months observations 0.04 (0.82) 1.20 (1.39) 2.67 (2.16)* -0.87 (-0.90) 0.26 Market Index 2: FTSE techMARK All-Share Index 12 months observations -0.12 (-1.12) 2.95 (1.96)** 4.90 (1.57)** 3.00 (1.17) 0.30 24 months observations 0.004 (0.05) 1.17 (1.25) 3.78 (2.11)* 0.15 (0.11) 0.30 36 months observations 0.05 (0.85) 0.59 (0.99) 2.82 (2.26)* -0.86 (-0.87) 0.24 t-statistics are presented in parentheses. * Significant at 5%, using two-tailed test ** Significant at 10%, using two-tailed test Notes: The 12 months testing period is from April 2000 to March 2001; the 24 months testing period is from April 2000 to March 2002 and the 36 months testing period is from April 2000 to March 2003, using 12, 24 and 36 observations respectively. The intercepts reported in Table 2 are measures of abnormal performance. An intercept of 12 months observations, -0.09 is minus 9 basis points per month, or about -1.11% per year. This implies that Lastminute.coms underperform both All-Share Index and techMark Index about -1.11% and -1.41% respectively in the 12-months period. The company continue to show underperformance nearly at 0.27% against main market index but outperform other technology companies about 0.10% in the first 24-months of trading. Although evidence mentioned in section 3.2 show that IPOs are generally underpriced in the long-run period, an interesting finding is that when the longer period is considered, i.e. 36 months after IPO, the company performs better than both indices from 1.52% to 1.66%. Based on the t-statistics, however, all of the average monthly abnormal returns are not statistically significant. The abnormal returns are well adjusted and captured by those 3 factors, and a statistically significant SMB beta is observed for 12 and 24-months period. Hence, it is not appropriate to say that Lastminute.com shows significant underperformance or outperforms over the market in the long run. The adjusted R-squared for all regressions which are ranging from 15 to 30% also infer tolerably high explanatory power of these tests. Table 4 shows the abnormal return results computed by BHAR approach as complementary information to the Lastminute.com long-run performance results, Table 4 Buy-and-hold Abnormal Returns Benchmark Selection FTSE All-Share Index FTSE techMARK All-Share Index Size Reference Portfolio 12-month BHAR (%) -76.74 -47.37 -92.35 24-month BHAR (%) -71.03 -24.82 -31.75 36-month BHAR (%) -37.85 0.04 -12.28 In the absence of factor adjustments, the Lastminute.coms BHARs represent greater abnormal performance. All results underperformed 2 indices and its size reference portfolio except there is over-performance against techMark Index in 36-months period. The general trend observed is the companys performance against all benchmarks are improving as time goes on, this is consistent with the finding from Fama-French (1993) three-factor model. These results, however, cannot be analysed statistically. Besides, this measure ignores the risk of investment. 6.1. Analysis and discussions for long-run results The reverse results showed in the longer period of trading generally object the prediction of long-run underperformance albeit I obtain evidence of underperformance in the first year of trading. Nonetheless, the statistical significance of these results are inadequate to make the rejection of the null Hypothesis 2 that no underperformance or over-performance exist in the company under analysis in the long-run period. The divergence of opinion hypothesis by Miller (1977) does not offer explanation in this clinical study. Given no prior trading history and limited financial information, investors tend to hold different beliefs in the IPO value of firm with limited operational history like Lastminute.com and therefore the divergence of opinions is large. As time goes on, the company does not demonstrate larger underperformance. Earnings management analysis results summarize in Table 5 provide an opposing result to the accounting manipulation explanations by Teoh, Welch and Wong (1998). Lastminute.com exhibits positive discretionary accruals (DAs) in the IPO year but continue to present three years of positive DAs after that. The accounting accruals in the company do not reverse themselves in later periods as expected. Table 5 Earnings management analysis results Yearly Discretionary Accruals 1999 2000 2001 2002 2003 -0.19 (-0.25) -0.15 (-0.10) -1.82 (-3.69) 0.89 -1.29 0.38 0.38 0.27 0.27 Discretionary accruals are computed as the ordinary least square (OLS) regression residuals, t-statistics are presented in parentheses. 7. Limitations and further research This study attempts to investigate and provide explanations for the observed results in the analysed company based on the theoretical models. Nonetheless, these explanations do not carry statistical significance. The signalling test proposed by Allen Faulhaber (1989) was considered but foregone due to the time and resources constraints. The test requires a sample of all companies that have floated in the London Stock Exchange market and the SEO issues for each company. Besides, the inaccessibility of issuer allocation process makes the winners curse test by Rock (1986) becomes impractical. The Fama-French three-factor model (1993) which employed to test the long-run performance of the company of interest has its limitation. The portfolio factor loadings, which are assumed constant, are likely to vary through time. Ritter and Welch (2002) point out that the asset-pricing literature itself has failed to provide an accepted model of risk-adjusted performance against which one can measure post-IPO performance, it still remains unclear how abnormal post-IPO performance is. They also point out that this model can produce very odd results for internet bubble burst period.[10] The BHARs and characteristic-matched portfolio approaches have been attempted in this study. The statistical test, however, cannot be carried out on BHAR results because there is only one analysed company. Thus, future research could be expanded to a large sample of IPO firms and separating them into technology and non-technology companies to investigate if the same performance behaviour applies to other technology companies and make comparisons of different sectors. The aberrant pricing and trading behaviour in the internet bubble has made it clear that even if there is systematic long-run underperformance, it is difficult or impossible to justify it in a reliable manner (Ritter and Welch, 2002). Still, further work is needed to tell us the appropriate way to assess the post-IPO performance of companies around that period, both in the United Kingdom and in other countries. 8. Summary and conclusions This paper empirically investigates and analyses the Lastminute.coms IPO in year 2000 from both short-run and long-run perspectives. It involves the examination of short-run and long-run performance of the company and the analysis of the theoretical models in academic literature to explain the performance behaviour. The results obtained confirm the IPO underpricing in Lastminute.com. Its initial return of the first trading day is ranging from 28% to 28.48% and exceeds all of the benchmarks selected. The high initial prices on the first day of listing may be due to the speculative bubble by investors which inflate the price of the internet company stocks. However, several other theories of IPO underpricing do not imply a prediction that fits into the companys short-run performance. The long-run performance results provide evidence that Lastminute.com does not perform as the international evidence on the long-run performance of IPOs indicate. Even though statistical insignificant results are found that the company performs better than market when longer period is considered, Lastminute.com does not underperform nor outperform the market in the long run. In addition, there is no evidence shows that the company manipulate earnings prior to IPO. The findings provide argument to the market efficiency hypothesis which is unlikely to explain the first-day returns of 28 percent. These results also challenge some theories which are widely accepted to explain IPO underpricing. Besides, share allocation has an impact on IPO underpricing and the lack of micro-level data on share allocation has limited the research field in finance. Long-run performance is the most controversial area of IPO research. The main caveat is that it is hard to exploit systematic long-run underperformance reliably in the bubble period. The results obtained from this study can provide important for the prospective investors in new issues of technology companies. This study, however, also suggests that long-term investors should always be cautious to analyse IPO firms.