**2. Literature review**

with an inexpensive and flexible tool to invest in a large pool of initial public offerings. On the contrary, investing in such a large number of IPOs individually would not be practically feasible due to the high cost of such a strategy. In addition, IPO ETFs enable robust diversification strategies against the highly volatile IPO market. The origins of IPO ETFs go back to April 2006, when the First Trust US Equity Opportunities ETF was launched on the New York Stock Exchange (NYSE). The Renaissance IPO ETF came to the surface about 7 years later in October 2013. The Renaissance International IPO ETF followed 1 year later. The last entry in the IPO ETF market was the First Trust International IPO ETF. This fund began trading in

*Linear and Non-Linear Financial Econometrics - Theory and Practice*

In this chapter, we examine the short- and the long-term performance of IPO ETFs. In particular, we compute the absolute, benchmark-adjusted, and abnormal returns of ETFs. Abnormal returns are obtained with the usage of the market model successively against the S&P 500 Index and the S&P 600 Small Cap Index. These indices also serve as benchmarks when we calculate the benchmark-adjusted returns of ETFs. Moreover, in the short-run, returns are computed for the first trading day as well as for the first 2, 3, 4, 5, 21, and 63 trading days. At the long run, cumulative absolute, benchmark-adjusted, and abnormal returns are calculated over the first 6, 12, 18, and 24 months of trading and for the entire trading history of each ETF up to October 31, 2016. Respective buy-and-hold returns are computed too. Furthermore, risk-adjusted returns are estimated with the usage of a six-factor model, which follows the Fama and French multivariate model. Finally, a market trend analysis is performed. This analysis assesses the pricing behavior of IPO ETFs

during the descending and the upward phases of the overall stock market.

average abnormal returns are negative after the initial day of trading.

The results show that IPO ETFs provide slightly positive average first-day returns given that the average initial return is positive but well below 1%. Going further, the average absolute return of IPO ETFs is positive over the first five trading days, but it is negative over the first 21 and 63 days of trading. Benchmark-adjusted returns are also positive up to 5 days when the S&P 500 Index is taken into consideration, but they are rather negative when the S&P 600 Small Cap Index is assessed. Finally,

With respect to ETF long-term performance, results reveal positive cumulative

In the last step, the market trend analysis reveals that when the stock market

benchmark-adjusted returns of ETFs moves upward in a rate of about 68% of days and when market returns increase, the benchmark-adjusted performance of ETFs declines in a rate of about 65% of days. A similar one to benchmark-adjusted

To the best of our knowledge, this is the first study on IPO ETFs. Given the convenience of trading with ETFs, the low cost of investing in such products, the high liquidity of the ETF market in general and the great interest of investors and

goes down, the absolute return of IPO ETFs goes down too on about 76% of negative trading days. When market goes up, IPO ETFs go up to in a rate of about 63% of positive trading days. The opposite behavior is displayed by the benchmark-

adjusted return of ETFs. This means that when the market goes down, the

return's behavior is the case for abnormal returns.

absolute returns over the various periods considered, whereas the cumulative benchmark-adjusted and abnormal returns are positive only for the first 6 months of trading with the majority of returns becoming negative over the next time periods examined. In the case of buy-and-hold returns, results indicate that ETFs produce significant such returns in the long run, either when the absolute or the benchmark-adjusted returns are assessed. As far as risk-adjusted return is concerned, the regression analysis shows that just one out of the four IPO ETFs examined can produce robust and statistically significant excess return relative to

November 2014.

market performance.

**98**

Given the lack of any research papers on IPO ETFs, we will provide a brief review of the main findings of the literature concerning the short- and long-run performance of IPOs worldwide.

A plethora of papers have examined the performance of IPOs using data from the United States. In early years, several studies, such as [1–7], have accentuated that IPOs are underpriced as can be inferred by the returns on their first trading days, which are significantly positive. In the same concept, [8] estimate that during 1990–1998, US IPOs left over \$27 billion of money on the table, where the money left on the table is defined as the price gain of the first trading day times the number of shares sold. The money left on the table is translated into significant underpricing of IPOs during the nineties. Furthermore, [9] report that in the 1980s, the average initial return on IPOs was 7%, whereas the average first-day IPO return doubled to almost 15% during the period 1990–1998, before jumping to 65% during the internet bubble years of 1999–2000. Finally, [10] shows that, after the bubble of 1999– 2000, the average initial return of IPOs in the US over the first decade of the new century was moving around 10%.

The short-run performance of IPOs in other developed and emerging markets has attracted the interest of researchers. Loughran et al. [11] show that the move by most East Asian countries to reduce regulatory interference in the setting of offering prices resulted in less short-run underpricing in the 1990s than in the 1980s. Ritter [10] shows that in China, the second largest economy of the world, underpricing of IPOs has been severe with initial returns amounting to up to 200%. However, over the recent years, IPO underpricing in China has started to decline as a result of the changing institutional constraints. The great underpricing of Chinese IPOs is also supported by the findings of [12, 13]. In Australia, Lee et al. [14] report strong first-day returns. Significant underpricing of IPOs is reported for Canada by [15] IPOs are underpriced in Japan too as evidenced by [16]. In the UK, Levis [17] has documented a significant underpricing of the companies going public in the British stock market. The same pattern has been revealed by [18] for Italy and [19] for France. More or less, IPO underpricing is a global phenomenon. To testify this assertion, Loughran et al. [20] report comprehensive statistical evidence of strong first-day IPO returns for a sample of 52 developed and emerging capital markets, which range from 3.3% in Russia to 239.8% in Saudi Arabia.

When it comes to the long-run performance of IPOs in the United States, the main conclusion of the literature is that that the stocks of companies going public tend to be overpriced in the long run. Overpricing is depicted in the underperformance of IPOs versus similar non-IPO stocks or relevant market indices. In this respect, Ibbotson

[21] provides evidence that the initial returns and the long-run performance of IPOs were negatively related during the period 1960–1969. Ritter [7] finds that IPO stocks significantly underperform a set of comparable companies over the 3 years after going public. Rajan and Servaes [22] reveal that over a 5-year period after going public, companies' underperformance relative to the market benchmarks ranges from 17% to 47.1%. Carter et al. [23] estimate an average underperformance of US IPOs over a three-year period after the initial offering of 19.92%. Gompers and Lerner [24] examine the performance for up to 5 years after listing of nearly 3661 IPOs in the US during the period 1935–1972 and find some evidence of underperformance when event time buy-and-hold abnormal returns are used but underperformance disappears when cumulative abnormal returns are utilized.

The second type of initial return computed is the benchmark-adjusted return of ETFs, which, following [7], is computed as the difference between the initial absolute return of the *i*th ETF and the corresponding return of the benchmark. The first-

where BAIR*i,t* = 1 refers to percentage benchmark-adjusted return of the *i*th ETF on its first trading day, IR*i,t* = 1 is defined as above and BR*<sup>t</sup>* = 1 concerns the return of

In our estimations of benchmark-adjusted returns, we employ two alternative stock indices to serve as benchmarks. The first one is the S&P 500 Index, which consists of the 500 largest companies in terms of market capitalization listed on the NYSE or NASDAQ. The second benchmark used is the S&P 600 Small Cap Index, which covers the small-cap range of US stocks. According to [39], indices that consist mostly of small cap companies are better benchmarks when assessing the performance of smaller stocks or portfolios. The S&P 600 Small Cap Index is used because the ETFs that have been selected to be studied are rather small-cap ETFs and, consequently, a small-cap index may be a more appropriate benchmark.<sup>1</sup>

In order to calculate the return of the index, which will correspond to ETF's first-

where *Ri* stands for the daily return of the *i*th ETF, *Rm* represents the return of the market index, namely the return of the S&P 500 Index or the S&P 600 Small Cap Index. We estimate market model to obtain the alpha and beta coefficients of each ETF, which we will then use to compute abnormal returns with the following

<sup>1</sup> As we will explain in a following section, each IPO ETF has its own benchmark and, thus, one could wonder why we do not use each ETF's own benchmark to estimate their benchmark-adjusted

performance. We do not do so, for two reasons. The first one is that the majority of ETFs worldwide and IPO ETFs in particular are passively managed and, thus, the tracking error of these funds, that is the difference in returns between ETFs and underlying indices, is expected to be low. (We will see in **Table 1** that the tracking error of the sample's ETFs is indeed low.) Therefore, a new ETF's price will also generally remain in line with the price of the underlying basket of securities and an "underpricing" pattern like that observed in IPOs of ordinary stocks is not expected to be the case. The second reason is that we try to identify whether IPO ETFs can be an alternative investing tool of investors seeking returns, which will be better than the average market returns, with the market returns being usually

represented by indices such as the two used in our analysis.

day return, we use formula (1) for indices too. This means that given that the trading history of the selected benchmarks is much longer than the history of the sample's ETFs, we calculate the return of indices on ETF's first trading day by subtracting the open price of the index on the day which relates to ETF's first trading day from its close price on the same day and we divide by the open price. The third type of initial return estimated is the abnormal return obtained with the usage of the market model. In order to estimate abnormal returns of ETFs, we follow the approach of [40]. More specifically, so as to estimate the abnormal returns of ETFs, we first need to estimate the time series market model expressed in Eq. (3), via which the return of ETFs is successively regressed on the return of the

BAIR*i*,*t*¼*<sup>1</sup>* ¼ IR*i*,*t*¼*1*–BR*t*¼*<sup>1</sup>* (2)

*Ri* ¼ *α<sup>i</sup>* þ *βiRm* þ *ε<sup>i</sup>* (3)

day benchmark-adjusted return of ETFs is shown in the following formula:

*IPO ETFs: An Alternative Way to Enter the Initial Public Offering Business*

the market index on ETF's first trading day.

*DOI: http://dx.doi.org/10.5772/intechopen.90269*

selected market indices:

model:

**101**

Outside the United Sates, in Australia, How et al. [25] compare the long-run performance of companies going public that payed a dividend and similarly matched firms, which did not pay a dividend revealing strong evidence that the paying firms perform significantly better than the nonpaying firms for a period up to 5 years after the dividend initiation date. Moshirian et al. [26] indicate that in China, Hong Kong, Japan, Korea, Malaysia, and Singapore, whilst there is initial underpricing in Asian IPOs, the existence of long-run underperformance depends on the methodology used. In Japan, Kirkulak [27] reports a three-year underperformance of �18.3% for the stocks listed between 1998 and 2001. In Canada, Kooli and Suret [28] find that investors who buy stocks immediately after their listing and hold these shares for a period of 3 years will incur a loss of about 20%. When a five-year buy-and-hold strategy is considered, underperformance amounts to �26.5%. In the United Kingdom, a number of studies such as those of [17, 29–31] have documented the existence of IPOs' long-run overpricing. Other studies on European IPOs, such as those of [32–34] for Germany, [35] for Austria, [36] for Spain, [18] for Italy, and [37] for France, also reveal significant long-run overpricing of IPOs. Overpricing is evidenced by their poor long-term performance compared to the performance of relevant market indices or reference stock portfolios. Based on these findings, IPOs would not be suitable for long-term buy-and-hold trading strategies.
