• No results found

Thesis MSc Finance Rijksuniversiteit Groningen

N/A
N/A
Protected

Academic year: 2021

Share "Thesis MSc Finance Rijksuniversiteit Groningen"

Copied!
34
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

IPO underpricing and long-run performance in the precious

metals mining industry: Evidence from Canada.

Thesis

MSc Finance

Rijksuniversiteit Groningen

Author: Anton-Jan Delker

Student number: S3026752

Supervisor: Dr. Peter Smid Word count: 11,432

Field Key Words: Initial public offering (IPO), Underpricing, Long-run performance, Precious metals, Mining, “hot issue” markets, Canada.

(2)

1

1. Introduction

The initial underpricing and long-run performance of initial public offerings (IPOs) is well documented in the literature across many countries. These researches are mainly focused on country indices and large sectors (industries). IPO underpricing is the phenomenon that the closing price of the IPO on the first day is significantly higher than the initial offer price of the IPO. This means that companies bear an indirect cost of a too low initial offer price (Ibbotson and Jaffe, 1975). IPO underpricing is often a short-term phenomenon (Ritter, 1991). However, in the long-run, many researches have resulted in evidence of underperformance of these underpriced IPOs (time period of 1-3 years). Ritter (1991) has concluded that relatively young companies and IPOs in heavy volume years will result in a long-run underperformance compared to the market (average). Ritter (1991) also finds that many companies go public near the peak of industry hypes, which results in overoptimistic investors. This peak is called a “hot issue” market. Ritter (1984) has studied the “hot issue” market in the U.S.A and concluded that there was a clear ‘’industry effect’’ in the IPOs of natural resources firms. He concluded that natural resources firms are more underpriced than other industries (and underperform in the long-run).

(3)

2

1) A thorough study in the precious metals is useful due to the increase in interest in the precious metals mining sector.

2) As noticed earlier, Ritter (1984) finds evidence that natural resources firms often go public in “hot issue” markets. Precious metals prices increased dramatically in the period 2010 to 2011, and therefore it would be interesting to study for any “hot issue” effects during these periods. 3) The results of long-run performance are also in the interest of shareholders. Possible zero- or negative returns in the precious metals mining industry can be consistent with the ‘’Impresario’’ hypothesis (Shiller, 1990): initial returns are the fact of hypes in “hot issue” markets.

4) This study could give a thorough insight to new precious metals mining companies who are considering an IPO in the near future. Further, this research could show in how well precious metals mining companies can time IPOs and can benefit from the “windows of opportunities” (Ritter, 1991).

5) Ritter (1991) further explained that the attraction of new capital (by new issues) is dependent on the long-run performance of IPOs. Long-run underperformance will be a cost when attracting new capital. This argument is also interesting for new and current precious metals mining companies.

There still is much to investigate in the precious metals mining sector, especially in North-America. This sector is often taken as a small part of the whole industry and is not taken alone, as for example in the studies of How (2000) in Australia and Ritter (1984) in the U.S.A. This has led to the following research questions:

1. To what extend are precious metals mining firms IPOs underpriced in Canada during

the period January 2004- January 2013?

2. What are the stock returns of precious metals mining IPOs in Canada during the period

January 2004- January 2013 in the long-run (1-4 years)?

Different researches try to explain the level of underpricing and long-run performance of IPOs by firm- and market characteristics. This study will also investigate the possible influence of firm- and market characteristics on both underpricing and long-run performance. This has led to the following research sub-questions:

(4)

3

2. Which firm- and/or market characteristics explain the long-run performance of precious metals mining firms during the period January 2004- January 2013 in Canada?

A sample of 58 IPOs of precious metals mining firms will be examined in the period January 2004 to January 2013 in Canada. The level of underpricing will be measured by calculating market-adjusted abnormal returns (MAARs). The long-run performance will be measured by calculating Buy-And-Hold abnormal returns (BHARs) in a period of four years after the IPO day. The NYSE ARCA Gold BUGS Index will be used as benchmark. Ordinary least square (OLS) regressions will be used to explain possible underpricing and long-run underperformance.

The structure is as follows: Section two contains an overview of relevant literature (theories related to the precious metals mining sector and empirical results). Based on this, the hypotheses will be stated. Section three includes an explanation of the data collection and methodology. Section four presents the results related to the hypotheses. Finally, in section five there will be a conclusion to the research- and sub questions, and there will be space for recommendations and the reflection.

2. Literature review

IPO underperformance is a phenomenon that occurs across the world, dependent on countries and different sectors. In this literature review I will focus only on the empirical results of the industry- and mining sector across countries which are closely related to the precious metals mining industry. Many researches measure IPO underpricing as the return of the first day (see, e.g., Ritter, 1991; How, 2000; Rock, 1986; Ibbotson and Jaffe, 1975). However, a more recent study uses different time periods (1, 5, 10 and 20 days) to measure IPO underpricing (Van Heerden and Alagidede, 2012). There is significant evidence of IPO underpricing in the mining industry (South-Africa and Australia), the gold industry (Australia) and the market as a whole including the mining industry (USA). The main results of these findings will be discussed in section 2.1.2.

2.1.1 IPOs underpricing theories

(5)

4

mining firms (and are used in natural resources researches). Most of these theories result in the conclusion that issuers (or underwriters) knowingly underprice their IPOs, and that investors expect that the IPOs are underpriced (Dimovski and Brooks, 2008). The theories of Baron and Holmström (1980), Rock (1986) and Muscarella and Vetsuypens (1989) are based on information asymmetry. Baron and Holmström (1980) explain that investment banks have superior information about the issue market (the demand for the stocks of a new IPO). In that situation is IPO underpricing the result of information asymmetry between the issuers of new securities and underwriters. The model suggests that issuers do not have this superior information and therefore relinquish the issue to investment bankers and allows them to determine the issue price. This results in a compensation to the investment banker in the form of an underpriced IPO. However, this argument was contradicted by the research of Muscarella and Vetsuypens (1989). They find IPO underpricing in the absence of asymmetric information between the issuers and underwriters. Rock (1986) suggests that there are two kinds of investors: informed or un-informed. The informed investors take advantage of their position, and enter only in the more underpriced and profitable IPOs compared to the un-informed investors. The un-informed investors are left with less- underpriced and profitable IPOs. This is called the “Winner’s curse” or “Lemons problem”, where both types of investors have different information about the value of the firm. Consequently, IPOs are underpriced on average, in order to compensate un-informed investors for expected losses (less attractive issues) and to induce them to take part of future issues. Otherwise un-informed investors do not want to participate in IPOs, which results in raising insufficient capital by the issuers.

(6)

5

It suggests that companies can recoup their costs by selling the retained shares in the aftermarket and by future issues.

Other relevant arguments are presented in Benveniste and Spindt (1989), Chalk and Peavy (1987). Benveniste and Spindt (1989) suggest that underpricing is needed by firms to raise the optimal amount of capital. Less underpriced IPOs result in less subscriptions. Chalk and Peavy (1987) suggest that underwriters underprice IPOs to preferred clients, with the aim to recoup their cost by higher services charges in the future.

The years 1980 to 1981 were characterized by a dramatic increase in natural resources prices (gold, silver, oil etc.). The years 2010 to 2011 showed similar increases, especially in the precious metals markets. It is therefore interesting to research the so called “hot issue” markets, which are characterized by high demand for IPOs (large subscriptions) and large initial returns on the first day (Ritter, 1984).

2.1.2 Empirical results underpricing

The precious metals mining industry is often part in researches in the countries: U.S.A, Australia, Canada and South-Africa. The average initial returns are presented in table 1. The precious metals mining sector is often taken as part of the whole country or it is taken as part of industrials in Canada and the U.S.A. The industrial sector in these countries show underpricing of 11.5%-21.65%. Ibbotson and Jaffe (1975) find extremely high underpricing compared to previous studies. This is due the effect of “hot issue” markets.

Table 1: Empirical results average initial returns (underpricing) of IPOs

Source Country Number of IPOsTime period Sector Average initial return Chalk and Peavy, 1987 U.S.A 649 1975-1982 Industrials 21.65%*** Dimovski and

Brooks, 2008 Australia 115 1994-2005 Gold Mining 13.3%***

Dimovski and

Brooks, 2004 Australia 53 1994-1999 Gold Mining 11.3%***

How, 2000 Australia 130 1979-1990 Mining 107.18%***

How, 2000 Australia 100 1979-1990 Resources 119.51%***

Ibbotson and Jaffe, 1975 U.S.A 1879 1960-1969 Industrials 110.9%*** Ibbotson et al., 1994 U.S.A 2439 1975-1984 Country 15.26%*** Jog and Riding, 1987 Canada 635 1971-1983 Country 11.5%*** Ritter, 1984 U.S.A 242 1979-1982 Natural resources 56.2%*** Van Heerden and

Alagidede, 2012 South-Africa 26 2006-2010 Mining 71.16%***

(7)

6

Specific results related to the precious metals mining industry can be found in several researches (see, e.g., Ritter, 1984; How, 2000; Dimovski and Brooks 2004, 2008; Van Heerden and Alagidede, 2012; Jog and Riding, 1987). There is underpricing of 71.16% in South-Africa in the mining industry (a large part of this is related to precious metals firms). Ritter (1984) and How (2000) find underpricing of 56.2%/119.51% in the mining/ gold mining sector, this effect is mainly explained due to “hot issue” effects. More recent studies of Dimovski and Brooks (2004, 2008) also show underpricing of 11.3/13.3% in the gold mining sector but less extreme.

Ibbotson and Jaffe (1975) show in their research that there exist many periods of extremely high returns of IPOs, mainly in 1960 and 1966 to 1968. This research is focused on the market as a whole. Ritter (1984) continued with these results and tried to explain why the “hot issue” markets occur with the model of Rock (1986). This research mainly focused on the natural resources markets. Ritter (1984) divided 1028 IPOs in the period of 1979 to 1982 in non- and natural resources firms. For non-natural resources firms, there was a difference in the initial average returns of 5.2% between “hot” and “cold” periods, which is not convincing. However, the difference for natural resources firms was significant with 92.6% (110.9% in “hot” and 18.3% in “cold” periods). Natural resources firms with IPOs-sales of lower than $500,000 presented even a higher difference in the initial average return of 116.5%. Small firms with smaller issues show significant higher initial returns than larger natural resources firms. How (2000) studies the Australian mining sector during the same period. There is an average initial return of 230.73% in this period (median: 81.50%). The underpricing in the gold mining sector is 119.51% and for other metals 76.9%.

When presenting underpricing theories and the “hot issue” market, it is also relevant to understand which characteristics influence IPO underpricing in the precious metals mining sector. Beatty and Ritter (1986) demonstrate that greater ex ante uncertainty by investors about IPO valuations lead to a higher need of underpricing. Several studies followed to explain IPO underpricing by firms- and markets characteristics.

(8)

7

the gold mining sector in Australia in the period 1979 to 1990. She finds that issue size is significant with univariate OLS, however in the multiple regression it is not significant.

The time of IPOs differs by the age of firms. Such studies as Ritter (1984), Ibbotson et al. (1994) and (How 2000) have found that younger firms are more underpriced than older firms. Young firms face more ex ante uncertainty and therefore the markets require underpricing.

The reputation of underwriters could also explain underpricing (see, e.g., Dimovski and Brooks, 2004; Carter and Manaster, 1990; Beatty and Ritter, 1986). A high signal could be send to the market to leave the control of the issue to a high ranked underwriter. Lower uncertainty leads to lower underpricing. Carter and Manaster (1990) create their own ranking system. Beatty and Ritter (1986) categorize underwriters by their market share. How (2000) uses the reputation of the underwriter but also the reputation of the geologists working with the mining firms. Geologists with a high reputation could lead to lower underpricing (How, 2000).

Dimovski and Brooks (2008) include the market sentiment (percent difference between IPO date of the company and the day before) of the gold mine index as independent variable. The more positive/ negative the performance of the gold mine index, the higher/ lower the underpricing of an IPO. How (2000) finds that gold mining IPOs are more underpriced than other mining sectors (including oil, gas and other metals), this is however not significant in explaining the level of underpricing.

(9)

8 2.2 Long-run performance

IPO long-run performance is also a well-known topic in IPO studies (see, e.g., Ritter, 1991; How, 2000). These studies measure the long-run performance in a time period of one, two and three years. The most common methods to measure long-run performance are: 1) Cumulative abnormal returns (CARs) and 2) Buy-and-hold abnormal returns (BHARs). A benchmark or sized-matched firms are most common to calculate these returns. The choice of benchmark can have a large impact on the returns (Ritter, 1991).

2.2.1 Long-run performance theories

Ritter (1991) finds in his study long-run underperformance (especially during hypes and for small firms) and gives possible reasons for these underperformance: risk measurement (different benchmarks are needed), bad luck, hypes and over optimism which is consistent with the hypothesis of Shiller (1990). Shiller (1990) presents the ‘’Impresario’’ hypothesis as mentioned in the introduction. He argues that IPOs are related to hypes in industries. IPOs are underpriced by underwriters to create a greater demand (maximizing revenue). IPOs who are subjected the most to hypes (and so a higher initial return), will underperform the most in the aftermarket. Investors are too overoptimistic during hypes, which will be corrected in the long-run (Ritter, 1991).

Ritter (1991) and Loughran and Ritter (1995) introduce the so called ‘’Windows of opportunities’’. This means that firms face a low cost of raising equity during hypes or in a good market environment. They argue that firms will probably be more overvalued if firms perfectly can enter the market (during hypes in a well market environment), which will result in lower/negative long-run performance.

2.2.2 Empirical results long-run performance

(10)

9

The long-run performance of mining and industrial (which contain mining) firms are presented in table 2. Lee et al (1996), How (2000) and Ritter (1991) calculate the average returns using a three year after market window. Kooli and Suret (2004) use an aftermarket window of five years (BHAR of -31.31%** in year four and -25.19%* in year five). The North-American industrials and mining firms show significant negative underperformance in all three years. Australian empirical results are mixed. Lee et al. (1996) find significant underperformance in the industrial sector in all three years, however in the mining sector only after two- and three years. How (2000) finds even a positive long-run performance in the first year of mining firms and negative but not significant returns in a three-year period. All researches in table 2 use regression analysis to explain the long-run performance of mining and industrial firms, most of these variables are in line with the independent variables explained in section 2.1.2. The most used independent variables will be presented.

Ritter (1991) finds that smaller offers perform the worst in the long-run for industrial firms. This is consistent with evidence from Canadian mining firms (Kooli and Suret, 2004). They find that issues of less than 10 million CAD perform the worst in the long run.

Both Ritter (1991) and Kooli and Suret (2004) find significant evidence of long-run underperformance of IPOs during “hot issue” periods. Ritter (1991) finds also a significant positive relation between the age of firms and long-run performance.

Ritter (1991) argues that IPOs that are higher underpriced (high initial return), will have the worst long-run performance. Lee et al. (1996) contradict this argument and find no long-run significant underperformance of underpriced IPOs in the mining industry as shown in table 2;

Table 2: Empirical results average Buy-And-Hold abnormal returns (BHARs) of IPOs

Source Country Time period Sector 1- year 2-year 3-year

average return Average return Average return

How, 2000 Australia 1979-1990 Mining 20.23% (23.16%)*** (7.60%)

Kooli and Suret,

2004 Canada 1991-1998 Mining (21.14%)*** (17.82%)* (18.92%)*

(11)

10

this is only the case for the industry sector as a whole. The level of underpricing has a positive significant relation with the long-run performance in the first two years. How (2000) finds negative long-run performance of underpriced industrial IPOs in the first two years. There is a positive but insignificant relation between underpricing and long-run performance in a period of three years. She finds no differences in the long-run of gold mining firms and other mining firms.

2.3 Hypotheses

All in all, previous studies resulted in significant positive MAARs (underpricing) on the first trading day (IPO day). More recent studies in the gold mining sector by Dimovski and Brooks (2008) show lower initial returns compared to older results in this sector, but still significant. This has led to the following hypothesis:

H1: The market-adjusted abnormal returns on the IPO day of Canadian precious metals mining firms are positive during the period January 2004- January 2013.

Previous papers resulted in different explanatory variables, which were significant in explaining the level of underpricing in the gold mining- and resources sector, as explained in section 2.1.2. The variables issue size, the firms age, the use of options (share options and underwriter options) and the reputation of the underwriter had a significant negative relation with the level of underpricing. This means: the higher these variables are (or when options are used), the lower the initial return. The variables market sentiment and IPOs during “hot issue” periods had a significant positive relation with the level of underpricing. This means: the higher these variables are (or IPOs during “hot issue” periods), the higher the initial return. This has led to the following hypotheses:

H2: The higher the age of precious metals mining firms, the lower the level of underpricing.

H3: The bigger the issue size of precious metals mining firms IPOs, the lower the level of underpricing.

H4: The use of options (share options and/or underwriter options) by precious metals mining firms lead to a lower level of underpricing.

(12)

11 H6: The higher the market sentiment (percentage change in the benchmark on the day of the IPO), the higher the level of underpricing.

H7: IPOs during “hot issue” periods are more underpriced than IPOs during “cold issue” periods.

H8: There is no difference in the relation between gold and silver mining firms and the level of underpricing.

As presented in table 2 in section 2.2.2, firms with IPOs underperform significant in the long-run. Industrials in the U.S.A and Canada perform the worst in the long long-run. More recent studies in the mining sector in Australia result in less underperformance in the first year and lower negative returns but significant after one year compared to other studies. This has led to the hypothesis:

H9: Canadian precious metals mining IPOs will have negative Buy-And-Hold abnormal returns in the long-run (1-4 years).

Different independent variables were significant in explaining the long-run underperformance of mining and industrial firms. The issue size and the firms age have had a positive relation with the level of long-run performance. This means: the bigger the size and the older the firm, the higher the long-run performance. The level of underpricing in relation to long-run performance has mixed results in previous studies, however the studies of Lee et al. (1996) and How (2000) are most comparable with this research. Therefore, I expect a positive relation between the level of underpricing and long-run performance. IPOs during “hot issue” periods show significant underperformance. This had led to the following hypotheses:

H10: The bigger the issue size of precious metals mining firms IPOs, the higher the level of long-run performance.

H11: The higher the age of precious metals mining firms, the higher the level of long-run performance.

H12: The higher the level of underpricing of precious metals mining firms, the higher the level of long-run performance.

(13)

12 H14: IPOs during “hot issue” periods underperform more in the long-run compared to IPOs during “cold issue” periods.

3. Data and research method

Section 3.1 contains the relevant data and the data collection. Section 3.2 contains the research method used to calculate the possible underpricing on the first day and the long-run performance by using Buy-And-Hold abnormal returns (BHARs). Regression analysis will be used to explain the possible underpricing and long-run performance.

3.1 Data

There were only three IPOs in the U.S.A in the period of January 2004 to January 2013. Therefore, this paper only focuses on IPOs of precious metals mining firms in Canada during this period. The data related to IPOs of precious metals mining firms have been obtained from ZEPHYR. The US SIC code used in ZEPHYR is: 104 (gold- and silver ores). There were 69 (completed-confirmed) IPOs during January 2004 to January 2013. The data in ZEPHYR include the issue price, issue size, year of the IPO, names of the underwriter and the related news releases. Adjusted stock prices are obtained from Data Stream/ Finance Yahoo to calculate underpricing and long-run performance. There are some requirements for both underpricing/ long-run performance that the IPOs must comply:

1. Data include only Canadian precious metals mining companies. Only companies with core business in gold- and silver mining are included.

2. Firm specific information (age, issue size, news releases, use of options and underwriters of the IPO) must be clearly stated on the website of the firm or in ZEPHYR. Table 3: Number of IPOs between January 2004-2013.

Year Gold mines Silver mines Total

2004 3 1 4 2005 1 1 2 2006 2 1 3 2007 7 5 12 2008 3 3 6 2009 2 1 3 2010 16 6 22* 2011 2 0 2 2012 2 2 4 Total 38 20 58

(14)

13

Five of the 69 firms have their core business in other metals/ mining services and are excluded. Two companies are gold royalty/ streaming firms. These firms provide upfront payments to mining firms in exchange for future mined gold/ silver at a predetermined price (World Gold Council, 2016). These firms cannot be qualified as precious metals mining firms. Four firms miss firm specific information. Based on these requirements, 58 precious metals mining IPOs qualify as shown in table 3.

The year 2010 will be characterized as a “hot issue” period, due to a combination of significant number of IPOs-, great demand for IPOs- and soaring metals prices during this year (Ritter, 1984). The benchmark for calculating the abnormal returns (for both underpricing/ long-run) is: NYSE ARCA Gold BUGS Index (NYSE, 2014). This benchmark contains especially large gold producers in North-America, but most of these companies produce also silver. Dividends of the underlying stocks are reinvested in the NYSE ARCA Gold BUGS Index (NYSE, 2014). This index is most appropriate to calculate abnormal returns, because this index contains similar gold-/ silver mining companies with comparable firm characteristics and similar market sentiment over time (and it is the most used benchmark in the precious metals mining industry). A Global index like the S&P 500 is less appropriate due to the difference in market sentiment and firm characteristics. However, the S&P 500 Index Total Return (with reinvested dividends) will be used for robustness checks.

3.2 Research method

Section 3.2.1 contains the research method related to initial performance on day one and section 3.2.2 contains the research method related to long-run performance of the mining firms.

3.2.1 Underpricing

To measure underpricing of precious metals mining companies, the methodology of Aggarwal, Leal and Hernandez (1993), Van Heerden and Alagidede (2012) are used to calculate the 𝐿𝑁𝑀𝐴𝐴𝑅𝑠 of an IPO. The 𝐿𝑁𝑀𝐴𝐴𝑅𝑠 of the 1st, 5th, 10th, 20th day period will be calculated to measure the short-term performance (underpricing).

The initial total return of a stock (for example of the IPO day), is the natural logarithm of the closing price at 𝜏 = 1 (𝑃𝑖, 1 ) on the first trading day (IPO day) divided by the IPO offer price at 𝜏 = 0 (𝑃𝑖, 0) as shown in figure 1 and equation 1:

𝐿𝑁𝑅𝑖, 𝜏 = ln (𝑃𝑖,𝜏

(15)

14

where,

- 𝐿𝑁𝑅𝑖, 𝜏 is the natural logarithm return of stock 𝑖 of day 𝜏. - 𝑃𝑖, 𝜏 is the closing price of stock 𝑖 of day 𝜏.

- 𝑃𝑖, 0 is the IPO offer price of stock 𝑖.

The initial total return of the benchmark (for example of the IPO day), is the natural logarithm of the closing price at 𝜏 = 1 (𝑃𝓂, 1) on the first trading day (IPO day) divided by the closing price of the previous day at 𝜏 = 0 (𝑃𝓂, 0) as shown in figure 1 and equation 2:

𝐿𝑁𝑅𝑚, 𝜏 = ln (𝑃𝓂,𝜏

𝑃𝓂,0) (2)

where,

- 𝐿𝑁𝑅𝑚, 𝜏 is the natural logarithm of the return of benchmark 𝓂 at the end of day 𝜏 - 𝑃𝓂, 𝜏 is the closing price of the benchmark 𝓂 of day 𝜏.

- 𝑃𝓂, 0 is the closing price of the previous day (of the IPO day).

Equation 1 and 2 are used to calculate the 𝐿𝑁𝑀𝐴𝐴𝑅𝑠 of stock 𝑖 over period 𝜏 = 1, 𝜏 = 5, 𝜏 = 10 and 𝜏 = 20. The initial return in excess of the return of the benchmark (equation 3) is calculated as the difference between equation 1 and 2:

𝐿𝑁𝑀𝐴𝐴𝑅𝑖, 𝜏 = ln (𝑃𝑖,𝜏

𝑃𝑖,0) − ln ( 𝑃𝓂,𝜏

𝑃𝓂,0) (3)

The next step is to calculate the average 𝐿𝑁𝑀𝐴𝐴𝑅𝑠, (𝐿𝑁𝑀𝐴𝐴𝑅𝑖, 𝜏̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅) as shown in equation 4:

𝐿𝑁𝑀𝐴𝐴𝑅𝑖, 𝜏

(16)

15

Before the use of the t-statistic in equation 6, first the data will be tested for a normal distribution. This can be tested by using the Jarque-Bera test (equation 5):

𝐽𝐵 = 𝑁 6 (𝑆

2+(𝐾−3)2

4 ) (5)

where,

- 𝑁 is the number of observations (58) - 𝑆 is Skewness

- 𝐾 is Kurtosis

If the data do not follow a normal distribution, the non-parametric Wilcoxon signed-rank test will be used. In case of a normal distribution, the following t-statistic (equation 6) will be used to test the first hypothesis as presented in section 2.3:

𝑡 =

𝐿𝑁𝑀𝐴𝐴𝑅𝑖,𝜏̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅

𝑠/√𝑁 (6)

where,

- ‘s’ is the standard deviation of 𝐿𝑁𝑀𝐴𝐴𝑅𝑖, 𝜏̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ - N is the number of firms (58)

Section 2.1.2 analyzes the variables which were significant in explaining the level of underpricing in the mining/industrial sector in previous researches, see table 4. This results in the following regression model (equation 7):

𝐿𝑁𝑀𝐴𝐴𝑅𝑖, 𝜏 = 𝛼0 + 𝛽1 ∗ 𝐿𝑁𝐴𝐺𝐸 + 𝛽2 ∗ 𝐿𝑁𝑆𝐼𝑍𝐸 + 𝛽3 ∗ 𝑆𝐻𝑂𝑃𝑇𝐼𝑂𝑁 + 𝛽4 ∗ 𝑈𝑊𝑂𝑃𝑇𝐼𝑂𝑁 + 𝛽5 ∗ 𝑃𝐸𝑅𝐼𝑂𝐷 + 𝛽6 ∗ 𝑆𝐸𝑁𝑇𝐼𝑀𝐸𝑁𝑇 + 𝛽7 ∗ 𝑈𝑁𝐷𝐸𝑅𝑊𝑅

+ 𝛽8 ∗ 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 + 𝜀 (7)

where,

- 𝐿𝑁𝑀𝐴𝐴𝑅𝑖, 𝜏, the natural logarithm underlying the level of underpricing at date 𝜏. - 𝛽 is an unknown parameter.

(17)

16

in the model. In case of heteroskedasticity, White’s heteroskedasticity consistent standard errors will be used. A second issue is multicollinearity. This occurs when two or more independent variables are highly correlated (± 0.7-1). This could lead to less imprecise regression coefficients. The variables will be tested before the regression to detect possible multicollinearity (this could lead to skewed or misleading results). This will be tested with a correlation matrix. Correlated independent variables will be excluded from the regression (equation 7).

The descriptive statistics of the variables are presented in the results section 4.1 in table 6, panel A. As shown in table 6, the variables offer size and age are highly positive skewed. Therefore, the natural logarithm of the variables will be used as presented in table 4 to adjust for its skewness. 𝑆𝐸𝑁𝑇𝐼𝑀𝐸𝑁𝑇 is not highly skewed and therefore the natural logarithm is not necessary.

The firms age is measured as one plus the difference between the IPO year and year zero (start date). Indicator variables (dummies) are used to measure the impact of options, industry (gold or silver), underwriter reputation and the “hot issue” period, 2010. The underwriter reputation of the firms is measured by using the “IPO Underwriter Reputation Rankings” of Ritter (2017). This ranking varies from zero to ten. An underwriter with a reputation of seven or higher is qualified as a high-quality underwriter.

3.2.2 Long-run performance

Previous studies often use different methods to measure the long-run performance of IPOs. The most used methods in the mining industry to measure long-run performance are: 1) average abnormal returns- (AARs) and average cumulative abnormal returns (CAARs), and 2) Buy-And-Hold abnormal returns (BHARs). The aim of this study is to measure the long-run

Variable Parameter Measure

Firms age LNAGE The natural logarithm of the firms age (difference founding year and IPO year plus one). Offer size LNSIZE The natural logarithm of the offer size (offer price * offered shares).

Share options SHOPTION A dummy variable (0 or 1), with a value of 1 in case of share options used to subscribers.

Underwriter-options UWOPTION A dummy variable (0 or 1), with a value of 1 in case of options used to underwriters. "Hot issue" market PERIOD A dummy variable (0 or 1), with a value of 1 in case of a ''hot issue'' period. Gold- and Silver

market sentiment SENTIMENT This is the return of the HUI index on the day of the IPO.

(18)

17

performance in a period of one-, two-, three- and four years. The first method is therefore irrelevant (which is appropriate for calculating monthly returns) and only BHARs will be used to measure long-run performance. The sample of 58 gold- and silver mining firms do not include firms with paying dividends during the period January 2004 to January 2013 (dividends are rare in the gold- and silver mining sector during the first years). The dividends of the benchmark are reinvested as mentioned before. The BHARs do not include the return on day one but start with the first closing price of day one as presented in figure 2.

To calculate the 1-4 years BHARs, the natural logarithm of the BHARs will be used (equation 8). Each year consists of 365 days, which result in 𝜏 = 1 for year one, 𝜏 = 2 for a two-year period, 𝜏 = 3 for a three-year period and 𝜏 = 4 for a four-year period as shown in figure 2.

𝐿𝑁𝐵𝐻𝐴𝑅𝑖,𝜏= [ln (𝑃𝑖,𝜏

𝑃𝑖,0) − ln ( 𝑃𝓂,𝜏

𝑃𝓂,0)] (8)

where,

- 𝐿𝑁𝐵𝐻𝐴𝑅𝑖, 𝜏 is the natural logarithm of the BHAR on day (period) 𝜏. - 𝑃𝑖, 𝜏 is the closing price of stock 𝑖 on day 𝜏.

- 𝑃𝑖, 0 is the closing price of stock 𝑖 on the first day after the IPO, 𝜏 = 0.

- 𝑃𝓂, 𝜏 is the closing price of the benchmark 𝑚 on day 𝜏.

- 𝑃𝑚, 0 is closing price of benchmark 𝑚 on the first day after the IPO, 𝜏 = 0.

Before the use of a normal t-statistic, first the data will be tested for a normal distribution (equation 5). If the data do not follow a normal distribution, the non-parametric Wilcoxon signed-rank test will be used. In case of a normal distribution, the following t-statistic (equation 9) will be used to test hypothesis nine as presented in section 2.2.3:

𝑡 =

𝐿𝑁𝐵𝐻𝐴𝑅𝑡̅̅̅̅̅̅̅̅̅̅̅̅̅̅

(19)

18

where,

- ‘s’ is the standard deviation of 𝐿𝑁𝐵𝐻𝐴𝑅𝑡̅̅̅̅̅̅̅̅̅̅̅̅̅ - N is the number of firms (IPOs)

Section 2.2.2 analyzes the variables which were significant in explaining the level of long-run performance in the mining/industrial sector in previous researches, see table 5. This results in the following regression model:

𝐿𝑁𝐵𝐻𝐴𝑅𝑖, 𝜏 = 𝛼0 + 𝛽1 ∗ 𝐿𝑁𝑆𝐼𝑍𝐸 + 𝛽2 ∗ 𝐿𝑁𝐴𝐺𝐸 + 𝛽3 ∗ 𝐿𝑁𝑀𝐴𝐴𝑅 + 𝛽4 ∗

𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 + 𝛽5 ∗ 𝑃𝐸𝑅𝐼𝑂𝐷 + 𝜀 (10)

where,

- 𝐿𝑁𝐵𝐻𝐴𝑅𝑖, 𝜏 is the natural logarithm of the BHAR on day (period) 𝜏. - 𝛽 is an unknown parameter.

The descriptive statistics will be presented with the t-statistics of the 𝐿𝑁𝐵𝐻𝐴𝑅𝑠. The model will be tested for the presence of heteroskedasticity and/or multicollinearity as explained in section 3.2.1. In case of heteroskedasticity, White’s heteroskedasticity consistent standard errors will be used. Multicollinearity will be tested with a correlation matrix. Correlated independent variables will be excluded from the regression (equation 10).

The independent variable MAAR presents the initial return on the first day of the IPO (underpricing percentage), the natural logarithm will be taken to adjust for its highly positive skewness (𝐿𝑁𝑀𝐴𝐴𝑅). Indicator variables (dummies) are used to measure the industry (gold- or silver mine) and the “hot issue” period, 2010.

4. Results

Section four contains the descriptive statistics and the results related to the hypotheses. To determine whether the t-statistics, non-parametric test and regression estimates are significant,

Variable Parameter Measure

Offer size LNSIZE The natural logarithm of the offer size (offer price * offered shares).

Firm age LNAGE The natural logarithm of the firms age (difference founding year and IPO year plus one). Underpricing LNMAAR The natural logarithm of the underpricing return at day one.

(20)

19

a 1% and 5% (denoted by *, **) significance level is used. The 10% significance level is indicated by *, but is not leading in the results.

4.1 Results and descriptive statistics in the short-run

The descriptive statistics related to the short-run variables are presented in table 6, panel A. As mentioned in section three, there were 58 IPOs in the Canadian precious metals sector in the period January 2004 to January 2013. Of the total sample, there were 38 gold mining IPOs and 20 silver mining IPOs. 38% (22 IPOs) of these IPOs were in the “hot issue” period (2010). This is not surprising due to the significant higher metal prices during this period. The 58 IPOs in total raised 3.048 billion CAD during the period January 2004 to January 2013. Of this total amount, almost one-third (932.32 million CAD) was raised by one major gold producer in 2007. Almost 1 billion CAD (33%) was raised during the “hot issue” period 2010. This amount is relatively low compared to the total amount, due to the small- (junior mining firms) and young firms entering the market during this period. However, the proceeds are relatively high (47%) without the IPO of the major gold producer in 2007.

Table 6: Underpricing descriptive- and distribution statistics for 58 Canadian IPOs between January 2004-2013. Panel A: Descriptive statistics variables short-run¹

Variable SIZE AGE PRICE SENTIMENT SHOPTION UWOPTION INDUSTRY UNDERWR PERIOD

Mean 52.55 4.05 1.78 0.20 0.12 0.51 0.64 0.26 0.34 Median 4.50 2.00 0.50 0.01 0.00 1.00 1.00 0.00 0.00 Maximum 932.32 24.00 15.20 5.16 1.00 1.00 1.00 1.00 1.00 Minimum 0.22 0.00 0.10 (3.89) 0.00 0.00 0.00 0.00 0.00 Sum 3,048 Standard deviation 146.32 5.28 3.05 1.77 0.33 0.50 0.48 0.44 0.48 Skewness 4.44 2.31 3.20 0.24 2.33 0.07 0.57 1.10 0.65 Kurtosis 24.84 8.10 13.60 3.10 6.42 1.00 1.33 2.21 1.43 Jarque-Bera 1343.61*** 114.67*** 370.08*** 0.56 80.74*** 9.67*** 9.93*** 13.23*** 10.10***

Panel B: Distribution statistics Market-adjusted abnormal returns²

Statistic LN MAAR (1) MAAR (5) MAAR (10) MAAR (20)

Mean 16.53 17.18 16.98 16.65 Median 11.61 10.97 11.60 12.92 Maximum 90.51 127.13 140.17 104.23 Minimum (51.57) (31.94) (57.45) (55.23) Standard deviation 25.35 29.67 34.66 32.63 Skewness 0.69 1.37 0.96 0.39 Kurtosis 4.06 5.48 4.99 3.45 Jarque-Bera 7.35** 33.05*** 18.49*** 1.91 T-statistic 4.96*** 4.41*** 3.73*** 3.88*** Wilcoxon signed-rank test 4.56*** 4.03*** 3.58*** 3.65***

¹ Issue size/ sum are in CAD Millions. AGE is in years. PRICE is in CAD. SENTIMENT (HUI index sentiment) is in percentage.

SHOPTION, UWOPTION, INDUSTRY, UNDERWR and PERIOD represents dummy variables. Skewness, Kurtosis are in numbers for all variables.

² The natural logarithm of MAAR for period one, five, ten and twenty days are in percentages. Skewness, Kurtosis, Jarque-Bera and the t-statistics are in numbers.

(21)

20

The average issue size is 52.55 million CAD with a minimum and maximum of 0.22- and 932.32 million CAD. The total amount raised is relatively high compared to the study of Dimovski and Brooks (2008), where 1.65 billion (Australian dollar) was raised by 114 gold mining IPOs in Australia during the period 1994 to 2004. This can be explained by the significant higher metal prices after 2004. The average age of the firms is four years with a minimum and maximum of zero- and 24 years. The company with the age of 24 years raised 932.32 million CAD. The average age of firms during the “hot issue” period is 1.8 years. Seven IPOs (12%) offered options (warrants) to their subscribers and 30 IPOs (52%) offered options to their underwriters. The use of options offered to subscribers is lower compared to other studies (see, e.g., How, 2000; Dimovski and Brooks, 2008). The use of options offered to underwriters is higher. There are 15 underwriters with a high-quality reputation (26%) and 43 underwriters with a low-quality reputation (74%). Precious metals mining firms who raised a higher amount with the IPO, have underwriters with a high-quality reputation (this result is presented in table 8 in section 4.1.2). The average IPO price of the sample is 1.78 CAD with a minimum and maximum of 0.10- and 15.20 CAD. 35 IPOs (60%) offered their IPO at a price below 1 CAD. These companies raised an amount below two million CAD. The market sentiment of the benchmark on the day of the IPOs is on average 0.20% with a minimum and maximum of -3.89% and 5.16%.

As mentioned in section three, the explanatory variables issue size and age are highly positive skewed. The natural logarithm will be taken to adjust for its skewness as presented in the regression analysis in section 4.1.2.

4.1.1 Short-run performance

Panel B of table 6 represents the distribution properties of the natural logarithms of the 𝐿𝑁𝑀𝐴𝐴𝑅𝑠 of the IPO day (day one), 5 days-, 10 days- and 20 days’ period. The 𝐿𝑁𝑀𝐴𝐴𝑅𝑠 are: 16.53% on the IPO day, 17.18% in a 5 days’ period, 16.98% in a 10 days’ period and 16.65% in a 20 days’ period1. The natural logarithms underlying the MAARs are not normal distributed (see Jarque-Bera tests), except the 20 days’ period. Therefore, the non-parametric Wilcoxon signed-rank test is also used to test for significance. As shown in table 6 panel B, all 𝐿𝑁𝑀𝐴𝐴𝑅𝑠 are significant at the 1% level (for both the non- and parametric tests). Consequently

1 The 𝐿𝑁𝑀𝐴𝐴𝑅𝑠 by using the S&P 500 Index Total Return result in 16.26%*** (t-statistic 4.99) at the IPO day,

(22)

21

𝐻0 (of the first hypothesis) is rejected, which means that there is significant underpricing in the precious metals mining sector in Canada. These results are significantly lower compared to older (before 1990) empirical results in the mining-, industrial- and natural resources sector in Australia and the U.S.A. However, these results are slightly higher but consistent with the previous study in Canada (market as a whole) and the more recent study in the gold mining sector in Australia (see, e.g., Jog and Riding, 1987; Dimovski and Brooks, 2008).

A large part of the returns in the 5-, 10- and 20 days’ period is a result of the large initial return on the IPO day. It therefore is interesting to take a closer look at the short-run performance during the first 10 days’ period for any significant excess returns after the IPO day. The cumulative 𝐿𝑁𝑀𝐴𝐴𝑅𝑠 and individual 𝐿𝑁𝑀𝐴𝐴𝑅𝑠 of the first 10 days are presented in table 7 and figure 3. As shown in table 7, there are no significant excess returns after the IPO day. Only the return on the IPO day is significant at the 1% significance level. There are four positive individual average 𝐿𝑁𝑀𝐴𝐴𝑅𝑠 at day two, four, seven and ten and five negative individual average 𝐿𝑁𝑀𝐴𝐴𝑅𝑠 at day three, five, six, eight and nine, but all not significant. All cumulative average 𝐿𝑁𝑀𝐴𝐴𝑅𝑠 are significant at the 1% significance level (5% in a six days’ period), which is not surprising due to the large initial return. It can therefore be concluded that investors can make a significant return on the first day of precious metals mining IPOs in Canada, but there are no significant excess returns within a nine days’ period after the IPO day.

Table 7: Average- LNMAARs and cumulative LNMAARs over the first 10 days.

Period Number Average t-statistic Average Cumulative t-statistic

(in days) of IPOs LNMAARs LNMAARs

1 58 16.53 4.96*** 16.53 4.96*** 2 58 8.88 1.11 25.41 3.04*** 3 58 (8.06) (1.01) 17.35 4.79*** 4 58 0.52 0.85 17.87 4.83*** 5 58 (0.69) (0.87) 17.18 4.41*** 6 58 (3.42) (0.83) 13.76 2.56** 7 58 3.89 0.97 17.65 4.33*** 8 58 (0.07) (0.08) 17.58 4.15*** 9 58 (0.96) (0.71) 16.62 3.66*** 10 58 0.36 0.61 16.98 3.73***

(23)

22 Figure 3: Average cumulative- and 𝑳𝑵𝑴𝑨𝑨𝑹𝒔 over ten days.

4.1.2 Regression results short-run

Eight explanatory variables are presented in table 4 in section 3.2.1, which were significant in explaining the level of underpricing in the mining sector. These variables are: 𝐿𝑁𝐴𝐺𝐸 (firms age), 𝐿𝑁𝑆𝐼𝑍𝐸 (issue size), 𝑆𝐻𝑂𝑃𝑇𝐼𝑂𝑁 (the use of share options) or 𝑈𝑊𝑂𝑃𝑇𝐼𝑂𝑁 (underwriter options), 𝑃𝐸𝑅𝐼𝑂𝐷 (“hot issue” period), 𝑆𝐸𝑁𝑇𝐼𝑀𝐸𝑁𝑇 (the market sentiment), 𝑈𝑁𝐷𝐸𝑅𝑊𝑅 (the underwriter reputation) and 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 (gold- or silver mining firm).

Before the multiple OLS regression, the variables are tested for the presence of multicollinearity. The correlations between the variables are presented in a correlation matrix, see table 8. There is no high correlation between variables. The underwriter reputation and the issue size are moderately correlated, but this is not an issue by running the regression model.

The variables 𝐿𝑁𝐴𝐺𝐸, 𝐿𝑁𝑆𝐼𝑍𝐸, 𝑆𝐻𝑂𝑃𝑇𝐼𝑂𝑁, 𝑈𝑊𝑂𝑃𝑇𝐼𝑂𝑁 and 𝑈𝑁𝐷𝐸𝑅𝑊𝑅 are negative

correlated with 𝐿𝑁𝑀𝐴𝐴𝑅 according to expectation. The variables 𝑃𝐸𝑅𝐼𝑂𝐷, 𝑆𝐸𝑁𝑇𝐼𝑀𝐸𝑁𝑇 are positive correlated with 𝐿𝑁𝑀𝐴𝐴𝑅 according to expectation. The variable 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 is

Table 8: Correlation Matrix dependent and explanatory variables short-run.

LNMAAR LNAGE LNSIZE SHOPTION UWOPTION PERIOD SENTIMENT UNDERWR INDUSTRY

LNMAAR 1 LNAGE (0.2758) 1 LNSIZE (0.4422) 0.0193 1 SHOPTION (0.0011) (0.0551) (0.1456) 1 UWOPTION (0.2707) (0.1138) 0.4790 (0.2776) 1 PERIOD 0.1203 (0.1477) 0.1241 (0.1574) 0.1202 1 SENTIMENT 0.1617 (0.0277) (0.1427) (0.1277) (0.2418) (0.1252) 1 UNDERWR (0.1245) 0.0473 0.6428 (0.2188) 0.4130 0.2342 0.0081 1 INDUSTRY (0.0722) (0.2336) 0.2130 0.0589 0.1337 0.0937 (0.2578) (0.0466) 1

The dependent variable is the natural logarithm of M arket-adjusted abnormal return (on IPO day). The explanatory variables LNSIZE and LNAGE are the natural logarithm of Size and Age and issue price. SHOPTION, UWOPTION, PERIOD, UNDERWR and INDUSTRY are dummy variables.

(24)

23

negative correlated with 𝐿𝑁𝑀𝐴𝐴𝑅, which is not according to expectation. Gold mining firms should be positive correlated according to empirical results as presented in section 2.1.2. However, the sign is slightly negative correlated and close to zero.

As mentioned before, the level of underpricing is mainly the result of the IPO day. Therefore, only the IPO day (𝐿𝑁𝑀𝐴𝐴𝑅 day 1) will be used in the regression analysis as dependent variable. The regression results are presented in table 9. The model has been tested for the presence of heteroskedasticity. White’s heteroskedasticity test results in a F-statistic of 0.64 (p-value: 0.88), therefore heteroskedasticity is no issue in the model and OLS is allowed to be used.

The explanatory variable 𝐿𝑁𝐴𝐺𝐸 is negatively significant at the 5% significance level. This means: the higher the firms age, the less underpriced the IPO is. This is consistent with

Table 9: Regression results short-run LNMAAR (IPO day) for 58 Canadian precious metals mining firms between January 2004-2013.

Regression estimates¹ Dependent variable² LNMAAR (1) INTERCEPT 1.359 (0.0000)*** LNAGE (0.0899) (0.0259)** LNSIZE (0.0698) (0.0116)** SHOPTION (0.0510) (0.5995) UWOPTION (0.0893) (0.2222) PERIOD 0.0502 (0.4440) SENTIMENT 0.5547 (0.7626) UNDERWR 0.1618 (0.0904)* INDUSTRY 0.0113 (0.8673) Adjusted R² 0.2458 Number of IPOs 58

¹ The dependent variables LNMAAR (initial return) is the natural logarithm underlying MAAR (initial return). ² The regression estimates LNAGE, LNSIZE are the natural logarithm of Age and Issue Size.

SHOPTION, UWOPTION, PERIOD, UNDERWR and INDUSTRY are dummy variables. SENTIMENT is in percentage.

(25)

24

empirical results, consequently 𝐻0 (of hypothesis two) of no relation is rejected. The economic meaning (log-log regression) of the coefficient is: a one percent increase in age, leads on average to a 0.0899% decrease in the level of underpricing. The variable 𝐿𝑁𝑆𝐼𝑍𝐸 is negatively significant at the 5% level which means: the higher the issue size, the lower the level of underpricing. This is consistent with empirical results, consequently 𝐻0 (of hypothesis three) of no relation is rejected. The economic meaning (log-log regression) of the coefficient is: a one percent increase in the issue size, leads on average to a 0.0698% decrease in the level of underpricing. The use of options (share options and underwriter options) have a negative relation with the level of underpricing which is consistent, but not significant. Therefore, we cannot reject 𝐻0 (of hypothesis four) of no relation. The explanatory variable 𝑈𝑁𝐷𝐸𝑅𝑊𝑅 has a significance level of 10%, but this is not leading in this study. Therefore, we cannot reject 𝐻0 (of hypothesis five) of no relation. The higher the market sentiment, the higher the level of underpricing, which is consistent with previous results. The variable is not significant, consequently we cannot reject 𝐻0 (of hypothesis six) of no relation. IPOs during the “hot issue” period (2010) are more underpriced compared to “cold” periods, which is consistent. The variable is not significant and therefore 𝐻0 (of hypothesis seven) of no relation is not rejected. Gold mining IPOs are slightly higher underpriced compared to silver mining IPOs, but there is no significant difference. This is consistent with previous results and therefore we cannot reject 𝐻0 (of hypothesis eight) of no difference between the variables.

The adjusted R-squared is 0.2458 which means that 24.58% of the dependent variable (underpricing) variation is explained by the model. This seems low, but it is higher compared to previous empirical results. Dimovski and Brooks (2008) have an adjusted R-squared around 10% to 15% by using different regression models (different variables) in the Australian gold mining sector. How (2000) reports an adjusted R-squared of 15.5% in the Australian mining sector. Lee et al. (1996) report an adjusted R-squared of around 11.28% to 13.27% using different regression models in the Australian industrial sector.

(26)

25

explaining underpricing. All in all, the differences compared to the old regression model are negligible.

4.2 Descriptive statistics and Long-run performance

The descriptive statistics in the long-run are presented in table 11. Panel A contains the descriptive statistics related to the long-run explanatory variables; their explanation is equal to the explanation of the explanatory variables in section 4.1. The long-run Buy-And-Hold abnormal returns are shown in panel B.

The average 𝐿𝑁𝐵𝐻𝐴𝑅 after one year is 7.61%, -10.75% after two years, -18.86% after three years and -26.51% after four years2. The 𝐿𝑁𝐵𝐻𝐴𝑅 after one year has a minimum and maximum

of -200.54% and 198.96%, the 𝐿𝑁𝐵𝐻𝐴𝑅 after two years of -238.21% and 273.65%, the 𝐿𝑁𝐵𝐻𝐴𝑅 after three years of 226.90% and 229.87% and the 𝐿𝑁𝐵𝐻𝐴𝑅 after four years of -216.49% and 223.59%. All 𝐿𝑁𝐵𝐻𝐴𝑅𝑠 are normally distributed. The normal t-statistic and the Wilcoxon signed-rank test are used to test for significance.

2 The 𝐿𝑁𝐵𝐻𝐴𝑅𝑠 by using the S&P 500 Index Total Return result in 6.23% (tstatistic: 0.55) after one year,

-9.99% (-0.77) after two years, -23.48% (-1.54) and -35.12%** (-2.5) after four years. The long-run performance is lower compared to the NYSE ARCA Gold BUGS Index. There is no significant difference between the benchmarks, using a paired-sample t-test. The results of the paired-sample t-test can be obtained on request.

Table 10: Regression results short-run LNMAAR (IPO day) for 58 Canadian precious metals mining firms between January 2004-2013.

Regression estimates¹ Dependent variable² LNMAAR (1) INTERCEPT 1.139 (0.0000)*** LNAGE (0.0853) (0.0243)** LNSIZE (0.0552) (0.0004)*** Adjusted R² 0.2403 Number of IPOs 58

¹ The dependent variables LNMAAR (initial return) is the natural logarithm underlying M AAR (initial return). ² The regression estimates LNAGE , LNSIZE are the natural logarithm of Age and Size.

(27)

26

As presented in panel B table 11, only the 𝐿𝑁𝐵𝐻𝐴𝑅 after four years is negatively significant at a 5% significance level. Consequently, 𝐻0 of hypothesis nine is rejected. The overall pattern of decreasing 𝐿𝑁𝐵𝐻𝐴𝑅𝑠 after a couple of years is consistent with previous results, however the underperformance in the long-run is less compared to previous studies (see, e.g., Ritter, 1991; Lee et al., 1996). These studies calculated only 𝐿𝑁𝐵𝐻𝐴𝑅𝑠 over three years, but their findings were largely significant negative over all three years. The four years 𝐿𝑁𝐵𝐻𝐴𝑅 is in line with the results of previous results in the Canadian mining industry (Kooli and Suret, 2004). This study showed -31.31% (5% significance level) after four years. Therefore, it can be concluded that precious metals mining firms do not underperform as poorly as in industrial- and other mining sectors in Canada and Australia. The positive insignificant 𝐿𝑁𝐵𝐻𝐴𝑅 after one year is also surprising compared to previous papers. However, this is consistent with the most recent long-run performance study in the Australian mining industry (How, 2000).

Table 11: Long run descriptive- and distribution statistics for 58 Canadian IPOs between January 2004-2013 Panel A: Descriptive statistics long run¹

Variable Size Age LNMAAR Industry Period

Mean 52.55 4.05 16.53 0.64 0.34 Median 4.50 2.00 11.61 1.00 0.00 Maximum 932.32 24.00 90.51 1.00 1.00 Minimum 0.22 0.00 (51.57) 0.00 0.00 Standard deviation 146.32 5.28 25.35 0.48 0.48 Skewness 4.44 2.31 0.69 0.57 0.65 Kurtosis 24.84 8.10 4.06 1.33 1.43 Jarque-Bera 1343.61*** 114.67*** 7.35** 9.93*** 10.10***

Panel B: Distribution statistics Buy-And-Hold-Returns²

Statistic LN BHAR Y1 BHAR Y2 BHAR Y3 BHAR Y4

Mean 7.61 (10.75) (18.86) (26.51) Median 11.58 (13.75) (6.37) (13.47) Maximum 198.96 273.65 229.87 223.59 Minimum (200.54) 238.21 (226.90) (216.49) Standard deviation 81.70 98.62 111.68 93.64 Skewness (0.52) 0.10 0.02 0.08 Kurtosis 3.37 3.58 2.34 2.81 Jarque-Bera 2.89 0.90 1.07 0.15 T-statistic 0.71 (0.83) (1.29) (2.16)** Wilcoxon signed-rank test 1.05 0.95 1.14 2.08**

¹ Size is in CAD Millions. Age is in years (difference between IPO date and start date of the company). LNMAAR

is in percentages. INDUSTRY and PERIOD represents dummy variables. Skewness, Kurtosis are in numbers for all variables. ² Buy-And-Hold returns after the years one, two, three and four are calculated in percentages.

¹,² All numbers in brackets are negative.

(28)

27 4.2.1 Regression results long-run

Five explanatory variables are presented in table 5 in section 3.2.2, which were significant in explaining the long-run performance (𝐿𝑁𝐵𝐻𝐴𝑅𝑠) in the mining sector. These variables are: 𝐿𝑁𝑆𝐼𝑍𝐸 (issue size), 𝐿𝑁𝐴𝐺𝐸 (firms age), 𝐿𝑁𝑀𝐴𝐴𝑅 (the level of underpricing), 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 (gold- or silver mining firm) and 𝑃𝐸𝑅𝐼𝑂𝐷 (the “hot issue” period). Before the multiple OLS regression, the variables are tested for the presence of multicollinearity. The correlations between the variables are presented in a correlation matrix, see table 12. There is no high correlation between the variables. The positive correlations of the variables 𝐿𝑁𝑆𝐼𝑍𝐸, 𝐿𝑁𝑀𝐴𝐴𝑅 and 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 with 𝐿𝑁𝐵𝐻𝐴𝑅 𝑌4 are according to expectation. The negative correlation of 𝐿𝑁𝐴𝐺𝐸 and positive correlation of 𝑃𝐸𝑅𝐼𝑂𝐷 with 𝐿𝑁𝐵𝐻𝐴𝑅 𝑌4 are not according to expectations. However, there is a weak correlation between the variables.

As shown in section 4.2, only 𝐿𝑁𝐵𝐻𝐴𝑅 after a period of four years is significant and therefore only the natural logarithm 𝐿𝑁𝐵𝐻𝐴𝑅 𝑌4 will be used as dependent variable in the regression analysis. The regression results are presented in table 13. The model has been tested for the presence of heteroskedasticity. White’s heteroskedasticity test results in a F-statistic of 1.20 (p-value: 0.31), therefore heteroskedasticity is no issue in the model and OLS is allowed to be used.

As shown in table 13, the explanatory variable 𝐿𝑁𝑆𝐼𝑍𝐸 has a positive sign which is consistent with previous results. The explanatory variable is not significant and therefore we cannot reject 𝐻0 (of hypothesis ten) of no relation. The independent variable 𝐿𝑁𝐴𝐺𝐸 is positive, which means: the higher the firms age, the higher the level of long-run performance. This is consistent with previous studies. 𝐿𝑁𝐴𝐺𝐸 is not significant. Consequently we cannot reject 𝐻0 (of

Table 12: Correlation Matrix dependent and explanatory variables long-run

LNBHAR Y4 LNSIZE LNAGE LNMAAR INDUSTRY PERIOD

LNBHAR Y4 1 LNSIZE 0.0573 1 LNAGE (0.0966) 0.0193 1 LNMAAR 0.2847 (0.4422) (0.2758) 1 INDUSTRY 0.0831 0.2130 (0.2336) (0.0722) 1 PERIOD 0.0950 0.1241 (0.1477) 0.1203 0.0937 1

(29)

28

hypothesis eleven) of no relation. The explanatory variable 𝐿𝑁𝑀𝐴𝐴𝑅 is positive and significant at a 1% level. This means that higher underpriced IPOs of precious metals mining firms in Canada lead to higher long-run performance. This is in contradiction with previous industrial results (Ritter, 1999). However, this is consistent with previous mining studies (see, e.g., Lee et al. 1996; How 2000). This also is in line with the signaling theory (see, e.g., Allen and Faulhaber 1989; Welch 1989). Therefore, 𝐻0 (of hypothesis twelve) of no relation is rejected. The economic meaning (log-log regression) of the coefficient is: a one percent increase in the level of underpricing, leads on average to a 1.4319% increase in the level of long-run performance.

Gold mining IPOs perform better in the long-run compared to silver mining firms. This difference is not significant which is consistent. Therefore, we cannot reject 𝐻0 (of hypothesis thirteen) of no difference. IPOs during the “hot issue” period 2010 has a positive sign, which means that these IPOs perform better in the long-run. This is in contradiction with the results of Ritter (1991) and Kooli and Suret (2004). The explanatory variable 𝑃𝐸𝑅𝐼𝑂𝐷 is not significant. Therefore, we cannot reject 𝐻0 (of hypothesis fourteen) of no difference.

Table 13: Regression results long-run Buy-And-Hold abnormal returns for 58 Canadian precious metals mining firms between January 2004-2013.

Regression estimates¹ Dependent variable²

LNBHAR Y4 Intercept (2.1822) (0.0670)* LNSIZE 0.0984 (0.1662) LNAGE 0.0301 (0.8565) LNMAAR 1.4319 (0.0147)** INDUSTRY 0.1358 (0.6099) PERIOD 0.0379 (0.8853) Adjusted R² 0.0438 Number of IPOs 58

¹ The dependent variables LNBHAR Y4 is the natural logarithm of BHAR Y4.

² The regression estimates LNSIZE, LNAGE and LNMAAR (market-adjusted abnormal return) are the natural logarithm of Size, Age and MAAR.

¹, ² Values in bold are the OLS parameter estimates (regression estimates) with related p-values in parenthesis. Significances of 1%, 5% and 10% are indicated by ***,** or *. Regression

estimates between brackets are negative.

(30)

29

The adjusted R-squared is 0.0438, which means that 4.38% of the dependent variable variation (𝐿𝑁𝐵𝐻𝐴𝑅 𝑌4) is explained by the model. The adjusted R-squared is low but in line/ higher compared to previous regression models in explaining long-run performance. Lee et al. (1996) have an adjusted R-squared of 5.39% (three years period 𝐿𝑁𝐵𝐻𝐴𝑅) and How (2000) an adjusted R-squared of 0.38% for the regression model with the three years period 𝐿𝑁𝐵𝐻𝐴𝑅 as dependent variable.

Table 14 includes the regression model with only the significant variable 𝐿𝑁𝑀𝐴𝐴𝑅 from table 13, for robustness checks. The coefficient is lower with the same 5% significance level. The adjusted R-squared is slightly higher, which means that the model improves in explaining long-run performance. All in all, the differences compared to the old regression model are negligible.

Table 14: Regression results long-run Buy-And-Hold abnormal returns for 58 Canadian precious metals mining firms between January 2004-2013.

Regression estimates¹ Dependent variable² LNBHAR Y4 Intercept (0.4389) (0.0032)*** LNMAAR 1.0513 (0.0303)** Adjusted R² 0.0646 Number of IPOs 58

¹ The dependent variables LNBHAR Y4 is the natural logarithm of BHAR Y4.

² The regression estimate LNMAAR (market-adjusted abnormal return) is the natural logarithm underlying the MAAR.

¹,² Values in bold are the OLS parameter estimates (regression estimates) with related p-values in parenthesis.

Significances of 1%, 5% and 10% are indicated by ***,** or *. Regression estimates between brackets are negative.

5. Conclusion

(31)

30

involved with the IPO. This may be the result of the fact that most of these firms are young and these firms also raised a low amount during the IPO. High quality underwriters are involved with higher amounts and relatively older firms.

There is significant underpricing of 16.53% on the IPO day, which is consistent with the more recent study in the Australian gold mining sector (after 1990). However, this percentage is significant lower compared to studies during the periods 1979 to 1990. These periods contained more “hot issue” periods compared to this sample (there is one “hot issue” period, 2010). The level of underpricing in the ten days’ period is mainly the result of the large initial return on the first day. There are no significant excess returns. Therefore, underpricing in this sector is a one-day event. The main explanatory variables in explaining the level of underpricing are: the firms age and issue size of the IPOs. Both explanatory variables have a negative sign as expected, which means lower underpricing when these variables increase. The IPOs perform less poorly in the long-run as previous results predict. The Buy-And-Hold abnormal return after the first year is positive, but not significant. The long-run underperformance is increasing with the years, but only significant after a four-year period with -26.51%. Concluded, there is lower underperformance in the Canadian precious metals mining industry within a long-run period of four years. The main explanatory variable in explaining the fourth-year period Buy-And-Hold abnormal return is the level of underpricing on the IPO day. The sign is significant positive, which is consistent with previous results in the mining industry. The higher the level of underpricing, the higher the long-run performance.

(32)

31

There are some limitations in this study which could be the basis and recommendation for further research in the precious metals mining sector. The demand in the precious metals mining sector is increasing in North-America with the current high- and fluctuating metals prices since 2010. This sample of 58 firms includes only IPOs in Canada, because there were only two IPOs in the U.S.A during this period. These are excluded. The metal sector is large in the U.S.A and therefore further research should also take these IPOs into account (if there are more IPOs in the coming years in the U.S.A). It also would be interesting to test for any differences between the two countries in the precious metals mining sector. How (2000) used the explanatory variable geologist reputation in the regression analysis in explaining the level of underpricing and long-run performance in the Australian mining industry. This explanatory variable has not been taken in this sample due to a lack and unavailability of this data. Firms with high geologist reputations (management experience) are characterized by high graded metal deposits with significant low all-in sustaining costs (AISC) compared to precious metals firms with a low geologist reputation involved. Further research could use this variable in the regression analysis, because the future of precious metals mining firms is highly dependent on the geologist reputation of the firms. This variable could have a significant impact on the level of underpricing and long-run performance.

References

Aggarwal, R., Rivoli, P., 1990. Fads in the initial public offering market? Financial Management 19, 45-57.

Aggarwal, R., Leal, R., Hernandez, L., 1993. The aftermarket performance of initial public offerings in Latin America. Financial Management 22, 42-53.

Allen, F., Faulhaber, G.R., 1989. Signalling by underpricing in the IPO market. Journal of Financial Economics 23, 303-323.

Baron, D.P., Holmström, B., 1980. The investment banking contract for new issues under asymmetric information: Delegation and the incentive problem. The Journal of Finance 35, 1115-1138.

Beatty, R.P., Ritter, J.R., 1986. Investment banking, reputation, and the underpricing of initial public offerings. Journal of Financial Economics 15, 213-232.

(33)

32

Carter, R.B., Manaster, S., 1990. Initial public offerings and underwriter reputation. The Journal of Finance 45, 1045-1067.

Chalk, A.J., Peavy, J.W., 1987. Initial public offerings: daily returns, offering types and the price effect. Financial Analysts Journal 43, 65-69.

Dimovski, W., Brooks, R., 2004. Initial public offerings in Australia 1994 to 1999 recent evidence of underpricing and underperformance. Review of Quantitative Finance and Accounting 22, 179-198.

Dimovski, W., Brooks, R., 2008. The underpricing of gold mining initial public offerings. Research in International Business and Finance 22, 1-16.

Grinblatt, M., Hwang, C.Y., 1989. Signalling and the pricing of new issues. The Journal of Finance 44, 393-420.

How, J.C.Y., 2000. Initial and long-run performance of mining IPOs in Australia. Australian Journal of Management 25, 95-118.

Ibbotson, R.G., Jaffe, J.F., 1975. “Hot issue” markets. The Journal of Finance 30, 1027-1042. Ibbotson, R.G., Sindelar, J.L., Ritter, J.R., 1994. The market’s problems with the pricing of

initial public offerings. Journal of Applied Corporate Finance 7, 66-74.

Jog, V.M., Riding, A.L., 1987. Underpricing in Canadian IPOs. Financial Analysts Journal 43, 48-55.

Kooli, M., Suret, J.M., 2004. The aftermarket performance of initial public offerings in Canada. Journal of Multinational Financial Management 14, 47-66.

Lee, P.J., Taylor, S.L., Walter, T.S., 1996. Australian IPO pricing in the short and long run. Journal of Banking and Finance 20, 1189-1210.

Loughran, T., Ritter, J.R., 1995. The new issues puzzle. The Journal of Finance 50, 23-51. Muscarella, C.J., Vetsuypens, M.R., 1989. A simple test of Baron’s model of IPO

underpricing. Journal of Financial Economics 24, 125-135.

NYSE, 2014: The NYSE Arca Gold BUGS Index (HUI). [online] Available at:

https://www.nyse.com/publicdocs/nyse/indices/nyse_arca_gold_bugs_index.pdf [Accessed 18 Apr. 2017].

Ritter, J.R., 1984. The “hot issue” market of 1980. The Journal of business 57, 215-240. Ritter, J.R., 1991. The long-run performance of initial public offerings. The Journal of

(34)

33

Ritter, J.R. UF Warrington Faculty, 2017. IPO data. [online] Available at: https://site.warrington.ufl.edu/ritter/ipo-data/ [Accessed 8 Feb. 2017].

Rock, K., 1986. Why new issues are underpriced. Journal of Financial Economics 15, 187- 212.

Schultz, P., 1993. Unit initial public offerings: A form of staged financing. Journal of Financial Economics 34, 199-229.

Shiller, R.J., 1990. Speculative prices and popular models. Journal of Economic Perspectives 4, 55-65.

The Mining Association of Canada, 2017. Mining facts. [online] Available at: http://www.mining.ca/resources/mining-facts [Accessed 8 Feb. 2017].

The Silver Institute: The Global Source, 2016. Supply and demand. [online] Available at: http://www.silverinstitute.org/site/supply-demand/ [Accessed 8 Feb. 2017]. Van Heerden, G., Alagidede, P., 2012. Short run underpricing of initial public offerings

(IPOs) in the Johannesburg Stock Exchange (JSE). Review of Development Finance 2, 130-138.

Welch, I., 1989. Seasoned offerings, imitation costs, and the underpricing of initial public offerings. The Journal of Finance 44, 421-449.

Referenties

GERELATEERDE DOCUMENTEN

Even with the two worst months of the May-Oct period added to the Nov-Apr period, there is still a significant outcome for the Total variable on a 10% significance level,

We expect that for more female representation on the boards relative to that of males the effect becomes larger and henceforth, a measure of Gender Diversity will be included in

I conduct t-tests of gender on risk rating, annual return rate of investment and de-investment as well as the risk sensitivity of investors on three financial products

In terms of the explanatory power of control variables, only leverage is negatively related to R&D expense to sales ratio at 10% significance level in firm fixed effect

The significantly higher returns can be explained by the fact that a takeover premium is paid over the market value of the target company, which is beneficial for the shareholders

Besides this, when looking at the regression results, statement 1, “I think it is more important to have safe investments and guaranteed returns, than to take risk to have a

Excessive optimism as an indicator for overconfidence in this thesis, is tested by making an estimation of the economic climate which is subtracted from the subcategory of

Lagged NPL is impaired loans over gross loans at time t-1, lagged reserve ratio is the loan loss reserves over impaired loans at time t-1, Slope EU/US is the yield curve