Investor attention and the performance of initial public offerings in Europe
MSc Finance Thesis
Name: Peter Nanning Student number: 2418355 Email: p.nanning@student.rug.nl
Study Programme: MSc Finance Supervisor: Prof. dr. W.G. Bessler
2
Abstract
This paper investigates the link between individual investor attention and the performance of initial public offerings (IPO) in Europe. I use Google Search Volume data, a novel, direct proxy for individual investor attention. I find that search volume can predict first-day returns, but the predictive power is fragile. First-day return and total deal size of the IPO both have a significant positive effect on long-term buy-and-hold abnormal returns.
3
1. Introduction
Traditional asset pricing models assume that all information is directly incorporated into asset prices, and that investors have an unlimited information-processing capacity with respect to new information (Fama, 1970). In reality, people have limited attention; it is a scarce cognitive resource (Kahneman, 1973). Investors can only focus on a limited set of assets, and not the on the whole universe of assets.
Researchers have investigated whether attention has an impact on e.g. trading behaviour and stock returns. The problem with this kind of research is measuring investors’ attention. Until recently, only indirect measures of investor attention were employed. For example, Barber and Odean (2008) measure attention grabbing stocks by abnormal return, being in the news and trading volume. They show that individual investors are net-buyers of attention grabbing stocks. Now, researchers see the advantage of using Google Trends1 as a direct measure of individual investor attention. This measure is called the Google Search Volume Index (SVI).
I will research if there is a relationship between limited attention and asset prices in the European stock market. Specifically, I want to investigate whether there is a relationship and causality between individual investors’ attention and Initial Public Offering (IPO) returns with using this direct measure of attention. IPOs usually have initially high returns, and then a long-term reversal. This is called underpricing and underperforming respectively, which are persistent features of the IPO market (Ritter and Welch, 2002). Existing rational theories of IPO underpricing argue that underwriters deliberately underprice IPOs that appear hot and efficiently price cold IPOs (Lin, How, Verhoeven, 2017). Other rational theories are for example asymmetric information, signalling, price support and the role intermediaries. See table 1 and 2 for rational reasons of underpricing and underperforming.
A growing body of the behavioural finance literature shows that IPOs taken public during periods of overoptimistic investor sentiment tend to exhibit low or negative long-run returns (Saade, 2015). Ljungvist, Nanda and Singh (2006) provide a theoretical model which incorporates sentiment individual investors. In this model, issuers deliberately underprice the IPO. They do this so that rational investors such as institutional investors, who sell the shares to sentiment investors, break-even. The model thus predicts high underpricing when individual investor sentiment is high. Because search volume can be measured before an IPO, one could examine if increased attention before an IPO leads to increased underpricing (Da, Engelberg and Gao, 2011). There are many more different theories of underpricing and underperformance. Furthermore, underpricing has changed over time. Ritter (2017) provides a table on his website that shows an overview of underpricing in the United States (U.S.) and shows that for the period of 1980 – 1989 the average first-day return in the United States is 7.2%, while the decade
4 thereafter, 1990 – 1999, the average first-day return increased to 21.0%. During the internet bubble period (1999-2000) initial returns even were 64.6% on average. Now, underpricing has become lower. From 2001 to 2016 first-day returns in the U.S. were about 14%. Moreover, Ritter (2015b, 2018) reports that first-day returns for European countries also saw these trends. But as Europe is a heterogeneous area, the magnitude of underpricing differs per country. In Germany, Italy and the United Kingdom for example, first-day returns were about 5% in 2011, while French IPOs saw initial returns of around 13% in 2011.
Search volume index has been used to investigate a variety of other topics in finance so far. Global index returns and attention have been researched (Chen, 2017), home equity bias and attention (Mondria, Wu and Zhang, 2010), as well as forecasting abnormal stock returns using Google search volume data (Joseph, Babajide and Zhang, 2011).
To my knowledge, SVI has not been used yet to study the European stock market or the European IPO market Additionally, this paper will use more recent data than e.g. Da, Engelberg and Gao (2011) did. The authors used data from 2004 – 2007, whereas I use data from January 2010 until December 2017. As Google is by far the largest and most popular search engine today, with a 65.2% market share worldwide2, utilizing SVI is a sound strategy to research limited attention.
I find that search volume has predictive power when it is solely regressed on first-day returns as the dependent variable. It exerts a highly significant and positive sign. After controlling for IPO-specific control variables search volume remains a significant predictor. When also controlling for aggregate market sentiment however, search volume loses its statistical significance. Search volume does not have predictive power when analysing the long-term performance of IPOs. These findings are discussed in section 5.
This paper is structured as follows. Section 2 addresses the literature on investor attention and its impact on stock and IPO returns. Section 3 contains the hypothesis development. Section 4 discusses data and methods used. Section 5 discusses the results of the empirical model. The last section summarizes, concludes, discuses limitations of this paper and provides implications for further research.
2 Source: https://searchengineland.com/google-worlds-most-popular-search-engine-148089 and
5
2. Literature on investor attention and IPO returns
Already more than three decades ago, Merton (1987) developed a model of capital market equilibrium in a market with incomplete information. He established that “investor recognition” affects expected return negatively for better-known firms. This is because the familiarity causes the rate of investor recognition to increase, the financing and investing activities to increase and the expected return to go down. Furthermore, the value of the security itself will increase due to investor recognition. Hence, according to Merton (1987) attention grabbing behaviour is interconnected to investor recognition. These findings are consistent with Fang and Peress (2009), who investigate the “no media premium”, which posits that stocks that have received little to no media coverage earn higher returns than stocks that have high media coverage. The authors find significant evidence for this premium. Investor recognition and attention has been researched extensively in the finance literature. This section provides a discussion of this literature. Specifically, I look at how investor attention affects stock and IPO returns.
2.1 Investor attention and stock returns
Recent studies in the realm of behavioural finance explain investor recognition by studying how limited attention can affect asset prices. For example, Barber and Odean (2008) study the hypothesis that individual investors are net buyers of attention grabbing stocks. This attention-driven buying stems from the search problem that individual investors have. There are thousands of potential stocks to buy. These investors buy stocks that grab their attention. They confirm this hypothesis by measuring attention with indirect proxies: extreme returns, abnormal trading volume and stocks that are in the news. Dellavigna and Pollet (2009) similarly analyse attention by the news announcements effect: the response of investors to earnings surprises. The authors find that Friday announcements have a lower immediate and higher delayed response. This is explained by limited attention: investors are distracted from work activities and these can cause underreaction to the earnings announcements. This is an explanation for the phenomenon why some managers release worse earnings announcements on Friday. Other indirect proxies of investor attention are: advertising expense (Chemmanur and Yan, 2009; Lou, 2014) and price limits (Seasholes and Wu, 2007).
6 Search Volume (ASVI) as a proxy for retail investor attention3. The authors also use a sample of Russel 30004 stocks and find that SVI is correlated with existing proxies of investor attention, and provide direct evidence that SVI captures individual investors’ attention. Additionally, the authors find that an increase in SVI predicts higher stock prices in the next 2 weeks, and an eventual price reversal within the year. The findings by Da, Engelberg and Gao (2011) confirm the hypothesis by Barber and Odean (2008), that individual investors are net buyers of attention grabbing stocks. In the universe of assets, investors will more likely picks stocks that grab their attention.
With this direct measure of attention, Chen (2017) also finds evidence that investor recognition decreases stock returns. The author investigates whether attention has an effect on global index returns and finds that an increase in attention lead to significantly lower global stock returns. This is attributed to and consistent with the no-media premium effect, first posited by Fang and Persess (2009). Furthermore, the author attributes this no-media premium to local investors, not to foreign investors. Local investors play a dominant role in affecting stock returns of local equity indices. Vozlyublennaia (2014) links investor attention and the performance of different asset classes. The author shows that there is a significant short-term change in index returns after an increase in attention. This is explained by the fact that past returns determine the impact of attention on future returns of different asset classes. This result is consistent with the hypothesis the author developed. It says that new information could be perceived by investors as indicating a higher or lower return in the future and, therefore, could lead to either an increase or a decrease in security returns.
2.2 Investor attention and IPO performance
The performance of IPOs has been investigated extensively over the years. See table 1 and 2 for an overview of the literature of rational reasons for IPO underpricing and long-term underperformance. The main rational reasons for underpricing that the literature discusses are: asymmetric information, how shares are allocated, role of the underwriter/intermediary, price support and the bookbuilding process. For long-term underperformance, rational reasons are lock-up periods, venture capitalist backing and flipping.
Investor attention and sentiment has been acknowledged to be an important determinant for IPO underpricing and underperformance (Ritter and Welch, 2002). However, it is difficult to measure attention and associated sentiment before an IPO because indirect measures such as trading volume and abnormal returns are unavailable prior to the offering. Cornelli, Goldreich and Ljungqvist (2006) however were able to obtain pre-market valuations by investors of IPOs
3 Individual investors and retail investors are used interchangeable in this paper
7 in European countries. They used a dataset of so called grey-markets. In these grey-markets (i.e. pre-markets) investors are able to speculate on the future stock price of companies. Because the investors are retail investors, the researchers could identify the sentiment of these investors. The results show that high pre-market valuations by individual investors lead to 40.5% higher aftermarket prices. Furthermore, these IPOs show reversals and underperforming within a year. The authors attribute this to the fact that retail investors are overoptimistic and pay prices that exceed a firm’s fundamental value. When this over optimism by retail investors occurs in the pre-market, long-term returns of these IPOs are negative because the share price reverts to its fundamental value.
Attention has also been researched in conjunction with IPOs. Chan (2013) investigates if retail sentiment affect IPO returns between 1994 and 2004. The author finds there exists a positive sentiment – volatility relationship on first-day trading, that was strongest during the internet bubble period (1999 – 2000). Furthermore, the author examines the IPOs’ long-run return and shows that higher individual investor demand at the time of offering results in poorer long-run performance of IPO shares. There are of course many reasons for how IPOs perform. Analyst behaviour for example is also a driver of IPO performance. When analysts are affiliated with the lead-underwriter of an IPO, the stock recommendations are often too optimistic, leading to significant long-run underperformance (Bessler and Stanzel, 2009).
Another behavioural explanation of IPO underperformance is overconfidence by individual investors. Overconfidence lead to high short-run returns because these investors think their private information is superior and then overreact. When in the long run public information is disclosed these investors review their previous beliefs and sell, leading to a reversal in the firm’s share price (Daniel et al. 1998).
In their paper on attention Da, Engelberg and Gao (2011) additionally investigate a sample of 185 IPOs taken place in the United States between 2007 and 2010. They use Search Volume Index as a direct proxy for attention because it can measure attention before an IPO takes place. They find that search volume and thus investor attention is a strong predictor of first-day returns and long-run underperfomance.
8 direction (meaning buying or selling) with both high pre-market sentiment and high aftermarket sentiment, but in the opposite direction with high pre-market sentiment and low aftermarket sentiment. Having in mind the importance of who gets shares allocated, and how much is traded by retail and institutional investors, Saade (2015) researches institutional and individual investors sentiment and the returns of new technology and growth firms IPOs. The author focuses on new technology and growth IPOs because these IPOs tend to underperform more heavily than other firms that go public. This is in line with Bessler and Bittelmeyer (2008). They find that technology firms with more experience, which they measure with the number and quality of patents a firm has, underperform less than firms that have low quality or no patents. Saade (2015) measures sentiment directly by using surveys5. The author finds that individual investor sentiment negatively affects IPO shares’ short and long-run performance. On the other hand, institutional investor sentiment does not affect IPO shares’ short-run performance but positively affects their long-run performance. This is explained by the fact that individual investors are considered irrational while institutional investors are rational. Barber and Odean (2008) give two reasons for this: (i) institutions face a significant search problem when selling and they often sell short. Institutions own far more stocks than most individual investors, and therefore the set of assets to choose from is a lot larger. (ii) Institutions allocate more time to pick stocks to buy or sell and use computers and professional tools to employ selection criteria, to narrow down their search. Although individual investors may also use computers and selection criteria, they are less likely to do so.
Lin et al. (2017) research IPO underpricing within a framework of speculative bidding. They hypothesise that information acquisition by investors (i.e. investor attention) increases the sensitivity of market reactions to price amendments. This is called the attention amplification effect. The authors find strong support for this hypothesis: As investor attention increases, the sensitivity of market reactions to price amendments, as measured by the strength of the association between initial price amendments and final price revisions, increases.
Lastly, Ritter (2003) discusses in a survey the main differences between U.S. and European IPO markets. The markets differ in terms of volume, initial returns6, magnitudes of underpricing and more. Also, listing standards, the process of going public, the regulatory bodies of the different countries in Europe, quality of institutions and institutional investors differ. There are of course many more differences. Chen (2017) for example finds that investor attention plays are more prominent role in the United States then elsewhere. Therefore, it is interesting to investigate whether the relationship between investor attention and IPO returns
5 These are the American Association of Individual Investors (AAII) for individual investors and Investor
Intelligence (II) for institutional investors.
10 Table 1. Literature on rational reasons of IPO underpricing
Author(s) Time period Countries Theory/reason Results Loughran, T. & Ritter, J.R.; 1990-1998 United Sates Asymmetric information
Money left on the table by underwriters.
Underwriters deliberately underprice (more than necessary) and then allocate these shares to its clients
Baron, D. Theory/model Asymmetric
information
IPO issuers delegate the pricing decision to
underwriters that have better information about the market. Issuer cannot monitor properly leading to lower than expected offer prices
Aggarwal, R., et al. 1997-1998 United States Allocation of shares
Positive relation between institutional allocation and first-day returns: with strong pre-market demand institutions are given more shares Carter, R. &
Manaster, S.
1979-1983 United States
Role of underwriter Prestigious underwriters are associated with lower risk IPOs. These underwriters choose lower risk issuers to maintain their high reputation.
Benveniste, L. & Spindt, P.
Theory/models p
Book-building Institutions and investors are compensated for premarket valuation and information of the IPO stock. Cornelli, F., & Goldreich, D. 1995-1997 United States
Book-building The authors show empirically that investment banks allocate more shares to bidders that reveal more information in the book-building process. Ruud, J.S. 1982-1983 United
States
Price support The shape of the distribution of initial IPO returns show that IPO stock prices are allowed to rise but not to fall; the author shows that this indicates influence of price support by the underwriter. Habib, M. A., Ljungqvist, A.P. 1991-1995 United States Marketing role of underwriter/Asym metric information
Marketing of an IPO is costly; the authors show that the higher the promotion costs are, the higher the underpricing of the IPO is to compensate Ritter, J.R., Welch, I. Review of literature United States
Flipping Flipping is when investors buy and immediately sell shares at the IPO date. Underwriters do this to create liquidity and to make a quick profit. Leads to upward demand and thus price increase
Cliff, M.T., Denis, D.J.
1993-2000 United States
Analyst coverage The authors find a strong correlation between the (perceived) quality and frequency (i.e. number of) of stock recommendations by analysts for the IPO. Authors interpret findings as underpricing being compensation for post-IPO analyst coverage Rock, K.
(1986)
Theory/model Asymmetric
information
Information asymmetry between informed and uninformed investors. Informed investors only buy shares below fair value, while uninformed
investors tend to overpay (winner’s curse)
11 Table 2. Rational reasons long-term underperformance
Author(s) Time-period Countries Theory/reason Results Aggarwal, R.
et al.
1994-1999 United States Lock-up period Managers strategically underprice IPOs. Demand for the IPO increases which generates higher prices after the lock-up period.
Field, L.C., and Hanka, G.
1988-1997 United States Lock-up period When lockup periods expire, trading volume increases significantly: insiders (e.g. managers) can now sell their shares, driving the price down Megginson,
W.L., Weiss, K.A.
1983-1989 United States Role of venture capitalist (VC) (certification role)
Backing by a VC reduces underpricing, attracts more prestigious underwriters and institutional investors. This leads to less underperforming in the long run in comparison with firms that do not have VC backing.
Brav, A., Gompers, P.A.
1972-1992 United States Role of venture capitalist
Empirical results show that venture-backed IPOs perform significantly better than non-venture-backed IPOs using the three-factor Fama-French model as benchmark for robustness checks. Carter, B.A.,
Dark, F.H., Singh, A.K.
1979-1991 United States Role of underwriter
The authors find that when an IPO has a prestigious underwriter, the underperformance relative to the market over three years is less severe than for IPOs that had a less reputable underwriter. The authors use different measures for underwriter reputation.
Field, L.C. 1979-1989 United States Institutional shareholdings
The paper shows that IPOs that have high(er) institutional shareholdings have better long-run performance than IPOs that have little or no institutional shareholdings.
Bessler, W.G., Stanzel, M.
Germany Analyst coverage (conflicts-of-interest hypothesis)
The authors find that when analysts are affiliated with the lead underwriter of the IPO, the buy recommendations of those analysts underperform significantly in the long-run. Buy
12
3. Hypothesis formulation
In this paper, I will investigate the link between individual investor attention and European IPO stock returns. IPOs usually follow temporarily high returns and then have a reversal in the long-term, which is well documented in the literature (Ritter and Welch, 2002). Ljungqvist et al. (2006) suggest that part of these effects can be attributed to the presence of individual investors’ attention induced buying behaviour. As SVI can be measured before the actual IPO takes place, one can measure how much attention by individual investors an IPO has had and thereby research if there is a relationship. To do this, Abnormal Search Volume Index is used. It is defined as: the natural logarithm of SVI during the IPO week minus the natural logarithm of median SVI of the previous 8 weeks:
𝐴𝑆𝑉𝐼𝑡 = ln(𝑆𝑉𝐼𝑖,𝑡) − ln[(𝑀𝑒𝑑𝑖𝑎𝑛(𝑆𝑉𝐼𝑖,𝑡−1, . . , 𝑆𝑉𝐼𝑖,𝑡−8)] (1) Where ln(𝑆𝑉𝐼𝑖,𝑡) is the natural logarithm of search volume index of IPO i during the IPO week t. ln[(𝑀𝑒𝑑𝑖𝑎𝑛(𝑆𝑉𝐼𝑖,𝑡−1, . . , 𝑆𝑉𝐼𝑖,𝑡−8)] is the natural logarithm of the median search volume value in the previous eight weeks leading up to the IPO week t of IPO i.
Da, Engelberg and Gao (2011) propose two reasons why individual investor attention and individual investor sentiment are related. First, attention is a necessary condition to generate sentiment. Individual investors first have to allocate attention to the IPO before they can become overly optimistic about an upcoming IPO. A second reason the authors provide is that individual investors are most likely to suffer from behavioural biases such as overconfidence (Daniel et al. 1998) and the disposition effect, the tendency of individual investors to hold on to stocks that have lost value and sell well performing stocks (Sherfrin and Statman, 1985).
Barber and Odean (2008) posit the attention induced price pressure hypothesis, which states that attention grabbing stocks (e.g. stocks that are in the news, that have abnormal returns) are most likely to be bought by retail investors, leading to an increase in price pressure due to the buying behaviour of individual investors. This can also be applied to IPOs: IPOs that receive more retail investor attention before the IPO date experience greater buying pressure from these investors, leading to a high initial return. Additionally, it is typically difficult to short-sell IPOs, amplifying this upward price pressure effect (Da, Engelberg and Gao, 2011). Therefore, it is expected that higher abnormal search volumes before the offering will lead to higher first-day returns. The following hypothesis is developed:
Hypothesis 1. An increase in individual investor attention (i.e. abnormal search volume)
before an IPO will lead to high expectations regarding the IPO. This in turn leads to high demand of the IPOs’ shares and therefore a higher first-day return. For the control variables, see the next chapter.
13
Hypothesis 2. Increased abnormal search volume prior to the IPO date will lead to high
expectations and higher demand, pushing up prices and initial returns. When this price pressure by individual investor fades away, the IPO will decrease to its fundamental value, and thus underperform.
3.1 Where does increased attention before IPOs come from?
The next section discusses the data and shows descriptive statistics. Table 4 shows that ASVI is on average 0.77, meaning that there is increased search volume prior to the IPOs. See also figure 1 that confirms this. Before the IPO week, search volume increases to reach its maximum level during the IPO week and then rapidly declines to its pre-IPO average. This increase in search volume is most likely coming from individual investors because institutional investors have access to more sophisticated tools such as Bloomberg terminals and other relevant data sources. Demers and Lewellen (2003) investigate media exposure at the time of IPO issues. They use the number of media cites a company received as an indirect proxy for media attention. They find that media attention for a company increases significantly in the month of the IPO. This could explain the increase in search volume and thus attention by individual investors. This is in line with Lou (2014), who shows that companies adjust advertising to attract investor attention. This leads to increased buying behaviour of these individual investors and consequently will lead to abnormal returns.
Figure 1. Average search volume during IPO week
14
4. Data and Methodology
In this paper I will focus on IPOs that have taken place in the period of 1 January 2010 – 31 December 2017. To identify IPOs, data is obtained from Zephyr7 which contains worldwide mergers and acquisition, IPO and venture capital deals data. One can download data going 10 years back, i.e. starting in 2007. Offer price, deal size (shares issued multiplied with offer price), the price range from the book building process, and number of shares issued are available. The final sample consists of 135 IPOs, completed between January 1st, 2010 and December 31st, 2017. The sample excludes secondary offerings, mutual funds, closed end funds, real estate investment funds, and IPOs with with an offer price below EUR 5) following Da, Engelberg and Gao (2011), Vakrman and Kristoufek (2015) and Saade (2015). This is to have a more restrictive sample, and individual investors are less likely to pay attention to these types of IPOs. Only IPOs that issued common shares are considered. Countries included in the sample are: Austria, Belgium, France, Germany, Italy, the Netherlands and the United Kingdom. This is to maintain an equal distribution of European countries. A country like Poland for example has many IPOs but is difficult to compare with Germany in terms of economic welfare and characteristics, which is always difficult when analysing Europe, as it is a heterogeneous economic area. For an overview of the search strategy conducted with the Zephyr database, see table 3. There, all the search steps are identified.
Closing prices for the first day of trading for companies in the sample are downloaded from Datastream, together with daily and weekly stock prices, total assets of the companies before the issue and trading volume. The first available closing price in Datastream is used to compute the first-day return variable, which is defined as the first available closing price divided by the offering price minus one. I use closing prices to calculate initial returns as is most common and straightforward. One could also use the first price of the day or the mid-day price. However, as the closing price is the most convenient to download, and as it captures the most trading done at the IPO date, the final closing price is used. For definitions for all variables, see table 5.
Google Search Volume data can be downloaded straightforwardly from
https://trends.google.com/. It provides an index of search activity by query or query category and is available in near real time (Stephens-Davidowitz and Varian, 2014). The data starts in January 2004 and thus is available until today. Weekly search volumes for the individual IPO stocks that are in the sample are obtained. I will us the company’s name for the search query, as a stock ticker is not widely available and/or known to individual investors prior to the IPO completion date. However, if a company name is rarely searched, Google Trends yields a zero value for that search query as it provides an index of search activity. This is because the index
15 measures the fraction of queries that include the term in question relative to the total number of queries at that time (Stephens-Davidowitz and Varian, 2014). Thus, the SVI ranges from 0 – 100. I exclude IPOs that do not yield enough search volume. Therefore, some IPOs that have little or no search volume had to be removed from the sample.
Table 4 shows descriptive statistics. What immediately stands out is the low average of first-day returns: 2.2%. The maximum return on the first day is 48.07% and the lowest is – 35.5%. Appendix D shows the average initial return per year for the sample. We see that the years after the financial crisis stand out, first-day returns are low for 2010 and 2011. For the years 2014 – 2016, initial returns are also low, and these years also have the most IPOs in the sample, driving the average first-day return of 2.2% for the total sample. Moreover, 2013 shows the highest first-day return: 7.6%. Ritter (2015)8 reports average first-day returns for the countries in the sample of 13.13% for a period from 1980 to 2015 (see appendix B). In other charts available on his website, Ritter (2015c, 2018) shows that the magnitude of underpricing differs significantly in Europe. In Germany, Italy and the United Kingdom for example, first-day returns were 4 to 5% in 2011, while French IPOs saw initial returns of around 13% in 2011, recovering from average negative first-day returns of -35% in 2009. In the sample of this paper, it is difficult to analyse on a per country basis. Most of the IPOs are French and German (52 and 30 respectively). This is more than half of the sample. French IPOs had a first-day return of -0.9% and German IPOs saw initial returns of 2.6%. These countries drive the results the most as they have low returns and had the most IPOs. For the complete overview, see appendix D, table D.2.
A histogram of first-day returns is presented in appendix A. One sees directly that the number of IPOs that had zero (or a very small) return on the first trading day is 58. This result is in contrast with other papers that investigate individual investor attention and IPO returns. For example, Vakrman and Kristoufek (2015) report initial returns of 16.74%, Da, Engelberg and Gao of 16.98%. Both of these papers utilize an U.S. dataset but both with different time periods (2004 – 2010 and 2004 – 2007 respectively).
Furthermore, as expected, the mean of ASVI is positive. This indicates that on average, the company’s in the sample are searched for actively before the IPO week (see also the previous section for reasons why this increases before an IPO). Moreover, price revision has a small negative mean of -0.023 (= -2.3%). Companies in the sample thus have a lower final offer price than the median of the filing price on average, but the revisions are small. This is largely in line with the findings of Jenkinson, Morrison and Wilhelm (2006). They find that the initial price range for European IPOs is rarely revised. Additionally, charts by Ritter show that when an IPOs have downward revisions to the offer price, the initial returns of these issues are also
8 On his website: https://site.warrington.ufl.edu/ritter/ipo-data/ under: “2015 update of (graph) Average First-day
16 low9 (Ritter, 2015a). Hanley (1993) also shows this with the partial adjustment phenomenon: IPOs that have final offer prices that exceed the filing range tend to have higher initial returns than IPOs that do not exceed their price range. This could partly explain why initial returns in this sample are low.
Table 3. Search strategy to find IPOs in Zephyr database
Step result Search result
Time period 54,743 54,743
Deal types: Initial public offering 38,476 21,773
Countries 25,927 1,488
Excluding REITs and closed end funds 155,734 1279
Deal offer price (EUR): min=5 7,690 351
Deal security types: Common shares 25,643 227
Excluding IPOs with little or no SVI 92 135
TOTAL 135
Notes: The time period is from January 1st 2010 to December 31st 2017. Countries included are: Austria, Belgium,
France, Germany, Italy, the Netherlands and United Kingdom. Real estate investment trusts and closed end funds are excluded from the sample as is common practice. Penny stocks (offer price below EUR 5) are excluded and only firms that issued common shares are in the sample. Lastly, companies that had little or no search volume had to be deleted from the sample. Data for 135 IPOs remain in the sample.
17 Table 4. Descriptive statistics
Variable Observations Mean Std. Dev. Min Max
Offer price revision 135 -0.023 0.10 -0.42 0.50
ln(Total assets) 135 12.39 2.52 6.44 18.92
Total assets 135 4826846 20158859 628 165118000
DSENT 99 -0.02 0.18 -0.82 0.37
Lead underwriter rank 135 0.34 0.38 0.00 1.00
ln(Offer size) 135 11.52 1.89 5.42 15.42
Offering size 135 350,500 593,032 225 4,999,975
ASVI 135 0.77 0.88 -4.03 3.51
1-year cumulative raw return 113 0.08 0.54 -0.98 2.45
Half year cum raw return 106 0.02 0.43 -0.88 1.58
Cumulative raw return t+30 117 0.02 0.27 -0.50 1.17
Cumulative raw return t+60 122 0.05 0.27 -0.43 1.47
Buy and hold abnormal 1-year return
123 -0.00 0.45 -1.19 2.18
First-day return 135 0.022 0.11 -0.36 0.48
ASVI*First-day return 135 0.04 0.19 -0.47 1.30
Average trading volume first
10 days 135 692.78 1631.83 0.18 15055.24
4.1 Data limitations
There are some limitations to the SVI data. Stephens-Davidowitz and Varian (2014) discuss the most important ones. First, query selection can pose a problem, i.e. cherry-picking. A researcher can choose the term that gives a desired result. I will use the company’s name as search queries in Google Trends. This may result in unwanted results when the name of the company resembles an everyday product. Companies that have such a name are deleted from the sample to avoid this problem. The variation in searches then is too little.
Second, the validity of Google measures is questioned when there is little to no correlation with existing measures. However, SVI is positively correlated with existing measures of attention, according to Da, Engelberg, and Gao (2011).
18 Fourth, the data is cached every day, meaning that the sample could differ somewhat when downloaded at different points in time. However, when search data for a company is downloaded multiple times the results are largely the same and the correlation between results is very high.
In addition, almost a hundred IPOs had to be removed from the sample because they have little or no search volume resulting in a total of 135 IPOs. This is quite a lot and therefore one should be careful drawing conclusions about the performance of IPOs in Europe based on search volume. Also, variation in first-day returns is unsatisfactory. 58 IPOs have a return of close to 0%, making analyses difficult. Lastly, although trading volume, the number of shares traded for a stock on a particular day, is obtained from Datastream, it is not possible to dissect how much trading is from retail investors and how much comes from institutional investors. I do not have access to data from for example brokerage accounts. However, the increased buying behaviour of high attention IPOs most likely comes from individual investors due to marketing and advertising efforts of issuing firms and underwriters. See section 3.1 for a more elaborate explanation.
4.2. Empirical model
Da, Engelberg and Gao (2011) propose to use Abnormal Search Volume Index, which is the logarithm of SVI during week t minus the logarithm of the median value of SVI during the prior 8 weeks. I will also use this measure, as it has become common practice in the investor attention literature when using Google Trends data. See section 3 and the previous section for more information about (A)SVI. To investigate the first hypothesis, an ordinary least squares linear regression model is used that takes the following form:
𝑅𝐼𝑃𝑂𝑖 = 𝛽0+ 𝛽1𝐴𝑆𝑉𝐼𝑖+ 𝛽2𝑇𝐴𝑖+ 𝛽3𝑂𝑆𝑖𝑧𝑒𝑖+ 𝛽5𝑃𝑅𝐸𝑉𝐼𝑆𝐼𝑂𝑁𝑖+ 𝛽4𝑈𝑅𝐴𝑁𝐾𝑖+ 𝛽6𝐷𝑆𝐸𝑁𝑇𝑖+ 𝜀𝑖,𝑡−1(2)
19 for an IPO is strong. In this way, underwriters keep underpricing constant which lead to high first-day returns. Ritter and Welch (2002) confirm this with updated data. Moreover, influence of the underwriter (𝑈𝑅𝐴𝑁𝐾). When an IPO is backed by a prestigious underwriter, the quality of the IPO is considered to be of high quality (Carter and Manaster, 1990). Data on underwriter rankings is obtained from Migliorati and Vismara (2014). The authors present rankings for 260 European-based underwriters from 1995 to 2010. An updated version is available on Jay Ritter’s website10. Lastly, investor sentiment, 𝐷𝑆𝐸𝑁𝑇, is obtained from Jeffrey Wurgler’s website. It contains an aggregate market sentiment index based on five sentiment proxies that are orthogonalized with respect to macroeconomic indicators. It is a belief from investors about future cash flows and investment risks (Baker and Wurgler, 2007). Furthermore, Jiang and Li (2013) show that the first-day return is not sufficient to fully grasp investors’ sentiment. It is important to measure investor sentiment both during pre-markets and aftermarket11 periods (Jiang and Li, 2013).
Thus, to investigate whether individual investors’ attention has an impact on long-term IPO returns (underperformance) the following regression model is employed:
𝐵𝐻𝐴𝑅𝐼𝑃𝑂𝑖,𝑡+1𝑦𝑒𝑎𝑟= 𝛽0+ 𝛽1𝐴𝑆𝑉𝐼 + 𝛽2𝑇𝐴𝑖+ 𝛽3𝑂𝑆𝑖𝑧𝑒𝑖+ 𝛽5𝑃𝑅𝐸𝑉𝐼𝑆𝐼𝑂𝑁𝑖+ 𝛽4𝑈𝑅𝐴𝑁𝐾𝑖+ 𝛽6𝐷𝑆𝐸𝑁𝑇𝑖+ 𝜀𝑖,𝑡−1
(3)
Where 𝐵𝐻𝐴𝑅𝐼𝑃𝑂𝑖,𝑡+1𝑦𝑒𝑎𝑟is the buy-and-hold abnormal return of IPO i after 1 year. The control variables are the same as in equation (2). To investigate long-term performance of IPOs, event time methods are used. Cumulative Abnormal Returns (CAR) and Buy-and-Hold Abnormal returns (BHAR) are common methods to capture long-term performance. Barber and Lyon (1997) suggest using BHAR. The authors believe that CARs are a biased predictor of long-term BHARs. Furthermore, BHAR is more appropriate to capture investor experience (Saade, 2015). BHAR is calculated as follows:
𝐵𝐻𝐴𝑅𝑖,𝑡 = 1 𝑁{(∏(1 + 𝑅𝑖,𝑡) 𝑁 𝑡=1 ) − (∏(1 + 𝑅𝑚,𝑡) 𝑁 𝑡=1 )} (4)
Where 𝐵𝐻𝐴𝑅𝑖,𝑡 is the buy-and-hold abnormal return for IPO i at time t, 𝑅𝑖,𝑡is return of IPO i at month t, and 𝑅𝑚,𝑡 the return of the benchmark. N is the number of IPOs in the portfolio. The benchmark used in this study is the return on equity indices of the different countries where IPOs have taken place. For the definitions for all variables used in this paper, see table 5 below.
10https://site.warrington.ufl.edu/ritter/ipo-data/
20 Table 5. Variable definitions
Variable Definition
SVI Search volume index: index of the volume of Google queries. Query’s in this paper are the company names
ASVI Abnormal search volume index. Defined as: the natural logarithm of SVI during the IPO week minus the logarithm of median SVI in the previous 8 weeks
First-day return Defined as: the first available closing price in Datasteam of IPO i divided by the final offer price minus 1.
Buy-and-hold raw return 1 month
The buy-and-hold raw IPO return of IPO i after one month
Buy-and-hold raw return 2 months
The buy-and-hold raw IPO return of IPO i after two months
Buy-and-hold raw return 1 year
The buy-and-hold raw IPO return of IPO i after one year
BHAR return 1 year
The buy-and-hold abnormal return after 1 year, adjusted with the respective market equity indices.
DSENT Investor sentiment index by Baker and Wurgler (2006). Based on five sentiment proxies that have been orthogonalized w.r.t. six
macroeconomic indicators. Updated version from Jeffrey Wurgler’s website12.
Underwriter rank Ranking of the lead underwriter, from Migliorati and Vismara (2014). Updated version from Jay Ritter’s website.
Total assets Natural logarithm of total assets of the company before the IPO has taken place. Downloaded from Datastream
Offering size Total deal value: offer price multiplied with number of shares offered. The natural logarithm is taken. Obtained from Zephyr. Offer price revision The percentage change from the median of the price range to the final
offer price: the offer price minus the median of the price range, divided by the median of the price range
Average trading volume, first 10 days
The average of trading volume for the first 10 days after the IPO completion date. Value in thousands.
21
5. Empirical results
This section will analyse the effect of individual investor attention on first-day and long-run IPO returns. Figure 2 displays the initial returns by dividing the sample into three quantiles: high attention (high-ASVI) IPOs, medium attention IPOs and low attention IPOs. The quantiles are based on the highest 10% ASVI values, the medium and lowest 10% values. As expected, the IPOs that received the most attention also have the highest first-day returns. On average, these firms had an initial return of 10.8% and a median return of 5.7%. On the other hand, the lowattention IPOs show a much lower, even slightly negative initial return: an average of -0.01% and a median return of 0%. The results for the high-ASVI IPO returns make sense: a higher pre-IPO surge in investor attention leads to relatively high initial returns in comparison with the rest of the sample. This is in line with the attention induced buying behaviour hypothesis of Barber and Odean (2008) (see also section 3). This buying behaviour can be seen in figure 5. High-ASVI IPOs trade more on average than low-ASVI IPOs. A thing to note here are that the initial returns presented in this paper are less in comparison with the recent literature, but comparable to average first-day returns for most of the countries in the sample. See also section 4 and the description of the data. A t-test is conducted to see if the difference between high-ASVI IPOs return and low-ASVI IPO’s return are statistically significant. The difference is significant at the 5% level. See appendix C for the results.
Table 3 shows the outcome of the cross-sectional ordinary least squares (OLS) regression model. The dependent variable is first-day return. The first column shows first-day returns solely regressed on ASVI. It shows a highly significant positive relationship: an increase in ASVI prior to the IPO date results in a higher initial return. The effect is fairly large, a standard deviation increase in ASVI leads to an increase in initial return by a magnitude of 2.7% of its standard deviation. Effectively this means an increase of 2.37% (0.879*0.027 = 0.023733) to initial returns. This is significant at the 1% level. So ASVI’s predictive power is large both statistically and economically when regressed on a stand-alone basis. R-squared is 0.051. This outcome confirms the attention-induced price pressure hypothesis of Barber and Odean (2008). Recall that this hypothesis states that high-attention stocks or IPOs are mostly bought by individual investors, leading to an upward, attention-induced price pressure, resulting in high first-day returns. Thus, on a stand-alone basis, ASVI is a strong predictor of first-day returns. This result is in support of hypothesis 1.
22 period from 2009 to December 2015. Therefore, IPOs that have taken place after December 2015 had to be left out in those regressions. Data for 99 IPOs are still available.
Regression 7 shows the model with all variables included. ASVI loses its predictive power, the regression coefficient decreases from 0.037 to 0.018 but it is not significant anymore. The last regression, column 9, employs White Heteroskedasticity Consistent Standard Errors as a robustness check for heteroskedasticity (White, 1980). The regression coefficients of all variables stay the same and are insignificant. However, when controlling only for IPO-specific characteristics and thus excluding DSENT, ASVI is significant at the 5% level. The outcome of this model is shown in column 1 of table 8. The coefficient remains consistent and positive: 0.026. Meaning that a one standard deviation increase in ASVI leads to a 2.32% increase in first-day return (0.026408*0.879= 0.02321). This also holds when doing the analyses with robust standard errors (White, 1980). This is shown in column 2 of table 8. When DSENT is included, some observations are dropped due to the limited availability of the DSENT variable. The model is thus sensitive to the number of observations. See table 8 for the outcome of these regressions.
Thus, after controlling for firm specific characteristics, aggregate market sentiment and adding robust standard errors to control for heteroscedasticity, ASVI does not exerts a statistically significant relationship with first-day returns in the main model. However, when doing the analyses with IPO-specific control variables only, ASVI is statistically and economically significant. Hereby confirming hypothesis 1, ASVI can partly explain underpricing in this sample of European IPOs. Increased investor attention, as measured by abnormal search volume, leads to a surge in individual investor demand for the shares of the IPOs. This in turn leads to an upward price pressure resulting in higher first-day returns. This is largely in line with Da, Engelberg and Gao (2011) who investigate IPOs and attention in the United States. The economical meaning of ASVI on first-day returns is higher however in their sample. This is consistent with Chen (2017), who finds that the effect of investor attention is more pronounced in the United States than elsewhere. Table 4 showed that the lowest value of first-day return was -35.5%. In unreported regressions, deleting such outliers yields that ASVI remains significant and has coefficients that are largely the same as in the regressions presented in this section.
5.2. Long-term returns
23 abnormal returns. So, at a first glance, it seems that high-attention IPOs are under-priced and underperform in the long-run. A time-varying analyses is displayed in figure 4. One can see that high-ASVI IPOs have the highest initial returns but have a strong reversal 4 months after the IPO. Low-ASVI IPOs on the other hand perform well in the long-term: they have low initial returns, but the buy-and-hold abnormal returns increase over time. The effect of individual investor attention on long-term buy-and-hold abnormal IPO returns is formalized in a cross-sectional linear regression, shown in table 4. The dependent variable is the long-term buy-and-hold abnormal IPO return after 1 year. This is the total return from a buy-and-buy-and-hold strategy after buying the stock on the first day of trading and then holding it for one year, adjusted with the return on equity indices of countries where the particular IPO takes place (Ritter, 1993). See equation 4 for the calculation of these returns. There are 123 IPOs left in the sample. This is because some stocks were delisted, and some IPOs did not have had a full year of trading on the stock market yet (e.g. had an IPO in 2017). The first column regresses the 1-year buy-and-hold abnormal IPO return on ASVI, which yields a statistically insignificant result. This may be attributed to the fact that long-term returns in the sample of this paper have a very high variance. The standard deviation of this variable is 0.43.
Column 2 shows the dependent variable regressed on first-day return. This yields a positive, statistically significant result at the 5% level. A (high) positive first-day return will also increase the buy-and-hold abnormal IPO return after 1 year. This result is contradicting at first sight, as underperformance of IPOs is well established in the literature (Ritter, 1991; Ritter and Welch, 2002) and high first-day returns usually lead to underperforming in the long-run. However, on average, the buy-and-hold abnormal 1-year returns are only slightly negative in the sample used in this paper, which could explain the positive effect of first-day returns on long-term returns.
Column 3 presents a regression of buy-and-hold abnormal 1-year returns on an interaction between ASVI and first-day returns, following Da, Engelberg and Gao (2011). A negative coefficient is expected because intuitively, one would expect that IPOs with high first-day returns, that have received significant attention by individual investors, will perform poorly in the long-run because the attention-induced price pressure of individual investors has worn off and the stock will reverse to its fundamental value. The interaction variable is statistically insignificant. This insignificance could be explained by the fact that the IPOs in this sample do not underperform as heavily as reported in other papers. Of course, it could also be that the model is unable to capture long-run performance and that other factors besides investor attention have an impact on long-term returns.
24 past cannot predict how performance will be in the future. Regression 10 is conducted with robust standard errors. First-day return is still significant at the 5% level, but total deal size loses its predictive power.
25 Table 6. The effect of investor attention on first-day returns.
(1) (2) (3) (4) (5) (6) (7) (8)
Dependent variable: First-day returns
ASVI 0.0273*** 0.018 0.018
(0.0102) (0.012) (0.013)
Offer price revision 0.063 0.022 0.022
(0.088) (0.105) (0.113) ln(Offer size) 0.004 0.002 0.002 (0.005) (0.010) (0.007) ln(Total assets) 0.000 -0.001 -0.001 (0.004) (0.007) (0.006) Lead underwriter rank 0.002 0.013 0.013 (0.024) (0.034) (0.033) DSENT 0.042 0.044 0.044 (0.060) (0.062) (0.043) Constant 0.00126 0.024** -0.023 0.022 0.022* 0.020* -0.016 -0.016 (0.0119) (0.009) (0.057) (0.046) (0.012) (0.011) (0.078) (0.056)
Robust S.E.? Yes
Observations 135 135 135 135 135 99 99 99
R-squared 0.051 0.004 0.005 0.000 0.000 0.005 0.032 0.032
26 Table 7. The effect of investor attention on long-term IPO performance.
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Dependent variable: buy-and-hold abnormal 1-year returns ASVI -0.061 -0.060 -0.060 (0.046) (0.049) (0.060) First-day return 1.042** 2.085** 2.085** (0.454) (0.976) (0.880)
ASVI * First-day return 0.191 -0.440 -0.440
(0.230) (0.582) (0.446)
Offer price revision -0.317 0.063 0.063
(0.385) (0.405) (0.276)
ln(Offer size) 0.022 0.067* 0.067
(0.021) (0.037) (0.044)
ln(Total assets) 0.017 -0.017 -0.017
(0.016) (0.025) (0.025)
Lead underwriter rank 0.116 0.096 0.096
(0.105) (0.130) (0.125)
DSENT 0.045 0.158 0.158
(0.247) (0.235) (0.285)
Constant 0.047 -0.022 -0.007 -0.008 -0.253 -0.208 -0.040 -0.045 -0.599** -0.599*
(0.053) (0.041) (0.042) (0.042) (0.245) (0.198) (0.054) (0.045) (0.294) (0.328)
Robust S.E.? Yes
Observations 123 123 123 123 123 123 123 88 88 88
R-squared 0.015 0.042 0.006 0.006 0.009 0.009 0.010 0.000 0.187 0.187
Notes: Dependent variable: the 1-year buy-and-hold return per IPO. Standard errors in parenthesis. This table presents the regression of buy-and-hold IPO returns on
Abnormal Search Volume and IPO characteristics control variables, both firm specific and macroeconomic. The sample period is from 2010 – 2017. In regressions 8 to 10 the number of observations has become less due to the DSENT variable, which runs from 2009 to December 2015, thereby omitting 10 observations.
27 Table 8. First-day return regressed on ASVI, only IPO-specific control variables
(1) (2)
ASVI 0.026** 0.026**
(0.010) (0.013)
Offer price revision 0.035 0.035
(0.087) (0.099)
ln(Offer size) 0.008 0.008
(0.008) (0.006)
ln(Total assets) -0.004 -0.004
(0.005) (0.004)
Lead underwriter rank -0.002 -0.002
(0.028) (0.026)
Constant -0.037 -0.037
(0.061) (0.046)
Robust S.E.? Yes
Observations 135 135
R-squared 0.061 0.061
28 Figure 2. ASVI quantiles and first-day returns
Notes: This figure shows the average and median first-day returns, grouped by ASVI level. The groups are based on the 10% highest ASVI, medium and lowest 10% ASVI of the sample.
Figure 3. Long-term returns and ASVI quantiles
29 Figure 4. Time-varying analyses BHAR
Notes: displayed in this figure is a time-varying analyses of the buy-and-hold abnormal returns, grouped per ASVI level. The BHAR of the total sample is also included, which is the middle line.
Figure 5. Average trading volume, first 10 days after the IPO
30
6. Summary, conclusion, and implications for further research
This paper investigates how investor attention influences the performance of IPO returns and tries to add an explanation to underpricing and underperformance. These two IPO puzzles have been widely acknowledged and researched in the literature (Ritter and Welch, 2002). This in a European context: only IPOs of countries that are in Europe are considered. Behavioural finance adds to the literature that investor attention can partly explain these two phenomena. Barber and Odean (2008) for example show that attention-induced buying behaviour of individual investors push up prices of stocks (and IPOs) which leads to high initial returns. This can also be seen in the sample of this paper. IPOs that had high attention also have a higher trading volume in the first 10 days after the IPO (figure 5). To measure investor attention, researchers often use indirect proxies such as abnormal trading behaviour, being in the news and advertising expenses (Seasholes and Wu, 2007). However, Da, Engelberg and Gao (2011) utilize Google Search Volume Index. This measures search frequency by individuals. By using SVI, one can measure investor attention directly and before an IPO, which indirect proxies cannot.
I find that an increase in SVI before the IPO leads to higher first-day returns. However, this is without control variables. With all control variables included, SVI loses its predictive power. When controlled for IPO-specific characteristics only, leaving out aggregate market sentiment as a control variable, ASVI remains significant and has a positive relationship with initial returns. Next, I investigate whether SVI can predict long-term performance of IPOs. This is not the case in my sample. This may be because there is large variation in long-term returns of this paper’s sample. First-day return however is a large predictor of long-term returns. This is at odds with the literature. When a firm has received a lot of attention prior to its IPO and had a high initial return, one would expect poor performance in the long run because the attention induced price pressure has worn off. In the sample of IPOs I investigated, the mean long-term underperformance is not as prevalent as elsewhere in the literature. However, although long-term returns in this paper are only slightly negative, the high-attention IPOs still underperform IPOs that received little attention significantly. See figure 2 for example.
31 nothing to find or the methodology I used was not able to detect what this paper wants to investigate. Regarding the former, the literature on investor attention that uses search volume is quite extensive and find significant effects. See Da, Engelberg and Gao (2011), Jiang and Li (2013), and Saade (2015) for example. The latter explanation is probably the most logical. Initial returns are quite low in the sample, as well as the underperformance in the long-term. Furthermore, it is difficult to dissect what trading comes from retail investors and what comes from institutional investors. Although I have established theoretically that attention induced buying probably comes from retail investors due to the marketing role of issuing firms and underwriters, it remains a challenge. I do not have access to sophisticated or expensive databases that contain this information. Lastly, because many IPOs had to be removed from the total sample due to having little or no search volume, one should take caution when analysing the results of the model.
32
References
[dataset] Ritter, J.R. 2017. Initial Public Offerings: Underpricing.
https://site.warrington.ufl.edu/ritter/files/2017/05/IPOs2016Underpricing.pdf
[dataset] Ritter, J.R., 2015a. First-day returns categorized by the revision in the offer price from the file price range. https://site.warrington.ufl.edu/ritter/files/2015/04/Percentage-first-day-return.pdf
[dataset] Ritter, J.R., 2015b. Average first-day returns on (mostly) European IPOs.
https://site.warrington.ufl.edu/ritter/files/2015/12/IPOs-International-Underpricing.pptx [dataset] Ritter, J.R., 2018. Average first-day returns for 54 countries. 2018 Update of Table 1. https://site.warrington.ufl.edu/ritter/files/2018/03/Int.pdf
[dataset] Ritter, J.R. 2015c. Powerpoint slides for Australia, Canada, China, France, Germany, Hong Kong, Italy, Japan, Korea, Singapore, Sweden, the U.K., and the U.S.
https://site.warrington.ufl.edu/ritter/files/2015/06/Average-First-day-Returns-and-Volume-by- Year-for-Hong-Kong-Germany-Italy-Japan-Korea-UK-US-China-Singapore-France-Sweden-Australia-Canada-2012-11-13.ppt
[dataset] Wurgler, J., 2016. Investor sentiment data (annual and monthly).
http://people.stern.nyu.edu/jwurgler/data/Copy%20of%20Investor_Sentiment_Data_2016033 1_POST.xlsx
Aggarwal, R., K., Kirgman, L., Womack, K.L., 2002. Strategic IPO underpricing, information momentum, and lockup expiration selling. Journal of Financial Economics 66, 105-137. Aggarwal, R.K., Prabhala, N.R., Puri, M., 2002. Institutional allocation in initial public offerings: empirical evidence. Journal of Finance 57, 1421–1442.
Baker, M., Wurgler, J., 2006. Investor sentiment and the cross-section of stock returns, Journal of Finance 61, 1645–1680.
Baker, M., Wurgler, J., 2007. Investor sentiment in the stock market. Journal of Economic Perspectives, 21 (2), 129-152.
Barber, B.M., Lyon, J.D., 1997. Detecting long-run abnormal stock returns: The empirical power and specification of test statistics. Journal of Financial Economics 43, 341-372.
Barber, B.M., Odean, T., 2008. All that glitters: the effect of attention and news on the buying behavior of individual and institutional investors. Review of Financial Studies 21, 785–818. Baron, D., 1982. A model of the demand for investment banking and distribution services for new issues. Journal of Finance 37, 955–976.
Benveniste, L.M., Spindt, P.A., 1989. How investment bankers determine the offer price and allocation of new issues. Journal of Financial Economics 24, 343–361.
33 Bessler, W.G., Stanzel, M., 2009. Conflicts of interest and research quality of affiliated
analysts in the German universal banking system: evidence from IPO underwriting. European Financial Management 15, 757–786.
Brav, A., Gompers, P.A., 1997. Myth or reality? The long-run underperformance of initial public offerings: evidence from venture and nonventure capital-backed companies. Journal of Finance 52, 1791 – 1821.
Carter, B.A., Dark, F.H., Singh, A.K., 2002. Underwriter reputation, initial returns and the long-run performance of IPO stocks. Journal of Finance 53, 285 – 311
Carter, R., Manaster, S., 1990. Initial Public Offerings and Underwriter Reputation. Journal of Finance 45, 1045–1067.
Chan, Y.-C., 2014. How does retail sentiment affect IPO returns? Evidence from the internet bubble period. International Review of Economics and Finance 29, 235 – 248.
Chemmanur, T.J., Yan, A., 2009. Advertising, attention and stock returns. Available at SSRN:
https://ssrn.com/abstract=1340605
Chen, T., 2017. Investor attention and global stock returns. Journal of Behavioral Finance 18, 358-372.
Cliff, M.T., David, D.J. 2004. Do initial public offering firms purchase analyst coverage with underpricing?, Journal of Finance 59, 2871 – 290.
Cornelli, F., Goldreich, D., 2002. Bookbuilding and strategic allocation. Journal of Finance 56, 2337 - 2369.
Cornelli, F., Goldreich, D., Ljungqvist, A., 2006. Investor sentiment and pre-IPO markets. Journal of Finance 61, 1187 – 1216.
Da, Z., Engelberg, J., Gao, P., 2011. In search of attention. Journal of Finance 66, 1461 – 1499.
Daniel, K., Hirshleifer, D., Subrahmanyam, A., 1998. Investor psychology and investor security market under-and overreactions. Journal of Finance 53, 1839–1885.
Dellavigna, S., Pollet, J.M., 2009. Investor inattention and Friday earnings announcements. Journal of Finance 64, 709 – 749.
Demers, E., Lewellen, K., 2003. The marketing role of IPOs: Evidence from web traffic. Journal of Financial Economics 68, 413–437.
Fama, E., 1970. Efficient capital markets: a review of theory and empirical work. Journal of Finance 25, 383–417.
Fang, L., Peress, J., 2009. Media coverage and the cross-section of stock returns. Journal of Finance 64, 2023 – 2052.
34 Field, L.C.,1995. Is institutional investment in initial public offerings related to long‐run performance of these firms?, Unpublished manuscript, UCLA.
Habib, M.A., Ljungqvist, A.P., 2001. Underpricing and entrepreneurial wealth losses in IPOs: theory and evidence. The Review of Financial Studies 14, 433 – 458.
Hanley, K.W., 1993. The underpricing of initial public offerings and the partial adjustment phenomenon. Journal of Financial Economics 34, 231–250.
Jenkinson, T., Morrison, A.D., Wilhelm, W.J., 2006. Why are European IPOs so rarely priced outside the indicative price range?, Journal of Financial Economics 80, 185-209.
Jiang, L., Li, G., 2013. Investor sentiment and IPO pricing during pre-market and aftermarket periods: Evidence from Hong Kong. Pacific Basin Finance Journal. 23, 65–82.
Joseph, K., Babajide, M., Zhang, Z., 2011. Forecasting abnormal stock returns and trading volume using investor sentiment: Evidence from online search. International Journal of Forecasting 27, 1116 – 1127.
Kahneman, D., 1973. Attention and effort. Prentice-Hall Inc., Englewood Cliffs, New Jersey. Lin, P., How, J.C.Y., Verhoeven, P., n.d. Some Evidence of “Publicity Multipliers” in Initial Public Offerings. Available at SSRN:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3034879
Ljungqvist, A.P., Nanda, V., Singh R., 2006. Hot markets, investor sentiment and IPO pricing. The Journal of Business 79, 1667 – 1702.
Lou, D., 2014. Attracting investor attention through advertising. Review of Financial Studies. 27, 1797–1829.
Loughran, T., Ritter, J.R., 2002. Why don’t issuers get upset about leaving money on the table in IPOs? Review Financial Studies 15, 413–443.
Megginson, W.L., Weiss, K.A., 1991. Venture capitalist certification in initial public offerings. Journal of Finance 46, 879 – 903.
Merton, R., 1987. A simple model of capital market equilibrium with incomplete information. Journal of Finance 42, 483–510.
Migliorati, K., Vismara, S., 2014, Ranking underwriters of European IPOs, European Financial Management 20, 891–925.
Mondria, J., Wu, T., Zhang, Y., 2010. The determinants of international investment and attention allocation: Using internet search query data. Journal of International Economics 82, 85-95.
Ritter, J.R., 1991. The long‐run performance of initial public offerings. Journal of Finance 46, 3–27.
35 Ritter, J.R., Welch, I., 2002. A review of IPO activity, pricing and allocations. Journal of Finance 57, 1795 – 1828.
Rock, K., 1986. Why new issues are underpriced. Journal of Financial Economics 15, 187 – 212.
Ruud, J.S., 1993. Underwriter price support and the IPO underpricing puzzle. Journal of Financial Economics 34, 135–151.
Saade, S., 2015. Investor sentiment and the underperformance of technology firms initial public offerings. Research in International Business and Finance. 34, 205–232.
Seasholes, M.S., Wu, G., 2007. Predictable behavior, profits, and attention. Journal of Empirical Finance. 14, 590–610.
Shefrin, H., Statman, M., 1985. The disposition to sell winners too early and ride losers too long: theory and evidence. Journal of Finance 40, 777-790.
Stephens-Davidowitz, S., Varian, H., 2014. A hands-on guide to Google data. Google Inc. Vakrman, T., Kristoufek. L., 2015. Underpricing, underperfomance and overreaction in initial public offerings: Evidence from investor attention using online searches. SpringerPlus 84, 1 – 11.
Vozlyublennaia, N., 2014. Investor attention, index performance, and return predictability. Journal of Banking & Finance 41, 17-35.
Welch, I., 1996. Equity offerings following the IPO. Theory and advice. Journal of Corporate Finance 2, 227-259.
36
Appendix A
Figure A.1 Histogram of first-day returns
37
Appendix B
Figure B.1 Average first-day returns of European IPOs
Source: Prof. Jay Ritter, University of Florida. -10% 0% 10% 20% 30% 40% 50% 60% R uss ia A ust ria D enm ar k N orw ay Tu rke y N et her lan ds Spa in Fr anc e Por tuga l Pol and B el g ium Is ra el Ita ly U ni ted K ingdom Fi nl and U ni ted St at es C yprus Irel and G er m any Sw eden Sw itze rland G re ece
38
Appendix C
Table C.1. T-test for difference between high and low ASVI IPOs’ first-day returns
High-ASVI IPOs Low-ASVI IPOs Mean 0.108 0.000 Variance 0.027 0.007 Observations 14 14 Pearson Correlation 0.049
Hypothesized Mean Difference 0.000
39
Appendix D
Table D.1. Average first-day return per year and corresponding no. of IPOs in the sample
Year
First-day
return average Number of IPOs
2010 1.2% 4 2011 -1.7% 11 2012 1.6% 11 2013 7.7% 14 2014 2.1% 29 2015 0.4% 29 2016 1.0% 15 2017 4.5% 22
Table D.2. Average first-day return per country
Country First-day return 1-year BHAR Number of IPOs
