Professional Documents
Culture Documents
PII: S1042-444X(14)00013-9
DOI: http://dx.doi.org/doi:10.1016/j.mulfin.2014.05.002
Reference: MULFIN 439
Please cite this article as: Vithessonthi, C.,What Explains the Initial Return
of Initial Public Offerings after the 1997 Asian Financial Crisis? Evidence
from Thailand, Journal of Multinational Financial Management (2014),
http://dx.doi.org/10.1016/j.mulfin.2014.05.002
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
*Highlights (for review)
Highlights
> A sample of 187 IPOs in Thailand between 2000 and 2012 is examined.
> Stock market development has a positive effect on the initial return of IPOs.
> Bond market development has a negative effect on the initial return.
> When interest-rate differentials are positive, they have a positive effect on the initial
return.
t
ip
cr
us
an
M
e d
pt
ce
Ac
Page 1 of 55
1) Title Page (WITH Author Details)
What Explains the Initial Return of Initial Public Offerings after the 1997
Chaiporn Vithessonthi
t
ip
University of Otago
cr
Tel: +64 3 479 9047; Fax +64 3 479 8171
us
E-mail: chaiporn.vithessonthi@otago.ac.nz
an
M
ed
pt
ce
Ac
Page 2 of 55
*2) Blinded Manuscript (WITHOUT Author Details)
What Explains the Initial Return of Initial Public Offerings after the 1997
ABSTRACT
t
ip
Do financial development, domestic interest rates, and interest-rate differentials
simultaneously affect the underpricing of initial public offerings (IPOs) in emerging market
cr
countries? Using a sample of 187 IPOs in Thailand between 2000 and 2012, I show that
us
financial development, stock market conditions, bond market development, and interest-rate
differentials affect the initial return, and that several of these effects appear to be contingent
an
on (1) whether the IPOs are hot or cold and (2) whether the interest-rate differential is
positive or not. During periods of positive interest-rate differentials, the effect of the interest-
M
rate differential on the initial return is positive and economically significant. Overall, my
ed
findings suggest that investors might undertake carry-trade strategies in the IPO market. In
addition, the initial return of the IPOs in my sample appears to be lower than the initial
pt
returns reported by previous studies using the pre-1997 financial crisis period sample.
ce
1
Page 3 of 55
1. Introduction
Following the 1997 Asian financial crisis, major financial and regulatory reforms in
Thailand (as well as in other crisis-hit countries, such as Indonesia and South Korea) have
been undertaken under the guidance of the International Monetary Fund. These reforms
t
should have led to (1) lower levels of informational asymmetries in capital markets and (2)
ip
higher levels of transparency in financial markets. In addition, domestic (and, in particular,
cr
retail) investors in Thailand have become more knowledgeable and sophisticated after the
us
crisis as less sophisticated (and/or uninformed) domestic investors were largely driven out of
the markets by the crisis. Given these characteristics, the initial public offering (IPO) market
an
in Thailand1 is an interesting context not only to indirectly test whether the level of
information asymmetry2 affects the level of the IPO underpricing but also to test whether
M
financial development is associated with the IPO underpricing. The IPO underpricing is a
situation in which the first-day market closing price of IPOs is higher than its corresponding
ed
offer price.
If asymmetric information accounts for the level of IPO underpricing, we will naturally
pt
observe declining levels of IPO underpricing in emerging countries when their economy
ce
substantially develops and the levels of asymmetric information decrease. Thus, according to
the information asymmetry hypothesis, there should be a negative relation between financial
Ac
development and the initial return of the IPOs. However, financial development should
attract larger amounts of foreign investment in the stock market, suggesting that the relation
between financial development and the initial return should be positive. Thus, two competing
hypotheses offer a different predicted sign of the impact of financial development on the
1
IPOs in Thailand generally follow the bookbuilding mechanisms as those used in the US.
2
One of explanations for the IPO underpricing is concerned with the level of information asymmetry (e.g.,
Rock, 1986).
2
Page 4 of 55
initial return of the IPOs. However, as Ritter (1991) argues that the levels of optimism about
the economy amongst investors is likely to play a strong role in explaining the short-run stock
return of the IPO than the levels of asymmetric information, it is possible that the positive
Consistent with the financial development literature, I argue that financial development
t
positively affects the initial return of the IPO. As foreign investors are more likely to be
ip
selective in choosing countries in which they invest, the level of financial development might
cr
signal whether equity and bond markets are large enough to attract foreign investors. As the
us
participation of foreign investors in the IPO market increases, all else equal, the initial return
is expected to be higher due to an outward shift in the demand curve for the IPOs. Rising
an
interest-rate differentials should lead to higher initial returns as a demand curve for the IPOs
is shifted outwards due to carry-trade activities in the IPO markets. As several researchers
M
(e.g., Baillie, 2011; Hochradl and Wagner, 2010; Moore and Roche, 2012; Sarno et al., 2012)
show the existence of a forward bias in currency markets, there is a question of whether
ed
carry-trade strategies are also carried out in the IPO market in the emerging market countries.
In a recent study, Coudert and Mignon (2013) note that with respect to carry-trade gains, the
pt
Thai baht (THB) is one of the most rewarding currencies, according to the Sharpe ratios. If
ce
low-yield currencies (e.g., the US dollar and the Japanese yen) when high-yield currencies
Ac
(e.g., the THB) are not expected to depreciate.3 Thus, the IPOs in high-yield currencies may
attract carry-trade positions in the IPO market, suggesting the positive effect of the interest-
3
This strategy is, however, riskier than typical carry-trade strategies in fixed-income securities that take a long
position in high-yield bonds in one currency (e.g., the Brazilian real) and a short position in low-yield-bonds in
another currency (e.g., the Japanese yen).
3
Page 5 of 55
In this paper I focus on previously unexamined dimensions of interest-rate differentials
in determining the initial return of the IPOs: the interplay among the development of
financial markets, the domestic interest rates, and the interest-rate differentials. The effects of
the first two on the initial return of the IPOs have been examined in the literature, but the
effects of the three can occur simultaneously but have yet to be studied. I argue that during
t
periods of stock market booms, falling domestic interest rates, and rising interest-rate
ip
differentials, the initial returns on the IPOs can be large. The potential magnitude of the spike
cr
in the initial return on the IPOs created due to the interest-rate differential effect is better
us
illustrated through a hypothetical scenario in which domestic investors decide to participate
the bond market is weak, thereby resulting in higher yields on both government and corporate
M
bonds, so that some domestic investors allocate a higher weight to relatively cheap bonds (i.e.
with higher yields). In such a case, they might be reluctant to rebalance their portfolio to
ed
include new IPOs. However, as it is often the case, interest rates in emerging market
countries are typically higher than those in developed countries, i.e., interest-rate differentials
pt
are usually positive and large in magnitude. Foreign investors (or domestic investors who
ce
have access to international financial markets) are likely to implement carry-trade strategies
When stock market conditions are good (i.e. bull markets), domestic interest rates are
high, and interest-rate differentials are large, it is logical for carry-trade investors to
participate in the IPO markets. The impact of this synchronization is that while higher interest
rates might potentially lower domestic investors’ demand for IPOs, large interest-rate
differentials, coupled with favorable stock market conditions, induce foreign investors’
growing appetite and demand for IPOs through the use of carry-trade strategies. It is hard to
4
Page 6 of 55
imagine that foreign investors who have already invested in the stock markets in emerging
market countries would forgo this seemingly highly profitable opportunity in the IPO in these
stock markets. However, given the greater risk of stocks in general, and in IPOs in particular,
investors are more likely to be very cautious in their participation in the IPO markets. Thus,
the effect of the interest-rate differential should be more pronounced when IPOs are expected
t
to be hot, which would in turn reduce the probability of incurring losses in the IPO markets
ip
when using carry-trade strategies. The effect of this synchronization should be even greater
cr
when domestic interest rates are low, reducing incentives for holding cash and low-yield
us
assets as evidenced by the implications of the US quantitative easing for the growing demand
for stocks and higher-yield assets after the 2008–2009 financial crisis.
an
I examine the magnitude of this synchronization effect on the initial return of the IPOs
by investigating a sample of 187 IPOs listed on the Stock Exchange of Thailand (SET)
M
between January 2000 and December 2012. In particular, I examine whether financial
development (i.e. stock and bond market development),4 stock market conditions, interest
ed
rates, interest-rate differentials, and exchange rates affect the short-run stock returns of the
IPOs. The results of this paper would likely be applicable to other emerging market countries
pt
that share similar characteristics (e.g., stages of financial development, financial and
ce
regulatory reforms after the financial crises in the 1990s, and positive and large average
interest-rate differentials). Overall, in this paper I aim to contribute to the literature of IPOs
Ac
Does financial development, indicated by stock and bond market development,5 affect
4
According to a review article by Yong (2007), previous studies on Asian IPOs have not examined the effect of
financial market conditions on the level of the IPO underpricing.
5
In the financial development literature, for example, as in Levine and Zervos (1998), Chinn and Ito (2006),
Zagorchev et al. (2011), and Boyd and Jalal (2012), measures for stock market conditions and bond market
development, as used in this paper, are used as indicators of financial market development.
5
Page 7 of 55
Do interest-rate differentials affect the initial return of the IPOs in Thailand?
Does the interplay of stock market development, domestic interest rates, and interest-
Consistent with Allen et al. (1999) and Kim et al. (2004), I find that IPOs in Thailand
between 2000 and 2012 are underpriced in the short run. That is, the average one-day initial
t
return for the full sample is 18% and statistically significant. Furthermore, the average one-
ip
day market-adjusted initial return for the full sample is 18.02% and statistically significant. I
cr
find that the initial return of the Thai IPO during the post-1997 financial crisis period is on
us
average lower than that of the pre-1997 financial crisis period reported by other scholars. I
interpret the findings of lower initial returns of Thai IPOs in recent years as evidence that
an
supports the notion that the level of asymmetric information in Thailand has substantially
decreased after the financial crisis, due to financial and regulatory reforms. This finding
M
provides further support to the asymmetric information hypothesis of Rock (1986).
I address the first research question by using (1) stock market development, (2) stock
ed
market conditions, and (3) bond market development as proxies for financial development. I
find that the stock market development6 is positively associated with the initial return. I show
pt
that the stock market’s price-to-book-value (P/BV) ratio has a positive effect on the initial
ce
return.7 Overall, these findings suggest that the stock market development and stock market
conditions drive the initial return of the Thai IPOs upwards, providing further support to
Ac
Adams et al. (2009), who find that industry P/BV ratio at the end of the year prior to the IPO
6
The natural logarithm of the level of market capitalization of the SET is used as a proxy for the stock market
development. The baseline results remain robust to an alternative measure of the stock market development. See
Section 4.4.
7
These positive relationships suggest that if investors are optimistic about future opportunities for existing firms
in the stock market, firms that go public during these periods will experience higher initial returns. Alternatively,
one may argue that a high stock market valuation (as indicated by a high market P/BV ratio) indicates
overperformance (or overpricing) of common stocks, which in turn leads to high short-run returns of the IPO.
6
Page 8 of 55
is positively associated with the one-day initial return of the mutual thrifts IPO in the United
States.
With respect to the effect of bond market development on the initial return of the IPOs, I
find that the amounts of outstanding T-bills, outstanding government bonds, and outstanding
private debt securities (in natural logarithms) at the end of the month prior to the IPO date,
t
which are used as a proxy for bond market development,8 are negatively associated with the
ip
initial return. These results suggest that the bond market development results in lower initial
cr
returns. Furthermore, controlling for the stock market development, the amount of new T-
us
bills issuance and new government bond issuance (both in natural logarithms) for the month
prior to the IPO date has a negative effect on the initial return, suggesting that Thai IPOs that
an
go public during periods of a high volume of new T-bills issuance or new government bond
interest-rate differentials, and exchange rates on the initial return. I find that in the full
ed
sample, after controlling for stock market development, the domestic interest rate has a
negative effect on the initial return, but the interest-rate differential between Thailand and the
pt
United States, which is used as a proxy for the domestic-foreign interest rate parity, has no
ce
effect on the initial return. The negative effect of the domestic interest rates suggests that
some investors appear to be unwilling to participate in the IPO market when domestic interest
Ac
rates are high, thereby shifting the demand curve for the IPOs inwards, and thus lowering the
initial return. A plausible explanation for no linkage between the interest-rate differential and
the initial return in the full sample is that the interest-rate differentials might not be a
8
These variables measure the bond market development in a narrow sense and the financial market
development in a broader sense.
7
Page 9 of 55
sufficient condition for investors to use carry-trade strategies in the IPO markets. That is, the
With respect to the third research question, I show that the interest-rate differential is
positively associated with the initial return during periods of positive interest-rate
differentials, suggesting the asymmetric effect of the interest-rate differential on the initial
t
return. This finding is consistent with the notion that carry-trade strategies are likely to be
ip
implemented only when interest-rate differentials are positive. Furthermore, real effective
cr
exchange rate (REER) of the THB is associated with the initial return only when interest-rate
us
differentials are positive. However, I find that the interest-rate differential has a negative
effect on the initial return in a sample of “hot” IPOs, but it has no effect on the initial return
an
in a sample of “cold” IPOs.9 One plausible explanation for the first result is that firms might
expect a strong demand for the IPO during periods of stock market booms and large interest-
M
rate differentials and thus increase their IPO prices, thereby reducing the initial returns. A
plausible explanation for the latter is that the investors who undertake carry-trade strategies in
ed
the IPOs are very selective and do not participate in “cold” IPOs due to small expected
payoffs. I also find that term spread has no effect on the initial return. This finding is
pt
somewhat inconsistent with Schrimpf (2010), who reports that the term spread has a
ce
As investors were less optimistic during the financial crises, lower IPO initial returns in
Ac
the aftermath of the 2008 financial crisis are therefore expected. The fact that the average
initial return for the Thai IPOs issued in 2008 and 2009 is 1.16% and 0.90%, respectively,
9
Following Brau and Fawcett (2006), I define a “hot” IPO as one with an initial return greater than 10% and a
“cold” IPO otherwise. There is a substantial difference between this definition and the one used by Derrien and
Womack (2003) and other scholars, who rather focus on whether the IPO markets are “hot” or “cold”. That is,
Brau and Fawcett (2006) emphasize on the initial return of an IPO in isolation, rather than on the return of IPOs
that are listed within a period of time, whereas Derrien and Womack (2003) focus on a pattern of IPOs in the
sense that a “hot” market is defined as a period in which many IPOs go public at almost the same time and the
average underpricing cost is high, whereas a “cold” market is defined as a period in which there are very few
IPOs that go public and the average underpricing cost is low.
8
Page 10 of 55
appears to support an alternative hypothesis that poor market sentiments, rather than lower
levels of asymmetric information, account for lower initial returns of the Thai IPOs over this
period. In a closely related study, Boonchuaymetta and Chuanrommanee (2013) show that
the IPO allocation to institutional investors has a positive effect on the IPO’s initial return,
while the issue size, the hot issue market, and the length of the lock up period have a positive
t
effect on the initial return in a sample of 153 IPOs in Thailand during 2001–2011. For
ip
comparison purposes, depending on model specifications, the adjusted R2 of my simple
cr
model with stock market variables (see Table 5) that is ranging between 0.23 and 0.30 is
us
higher than the adjusted R2 of 0.18 reported by Boonchuaymetta and Chuanrommanee
(2013), indicating that the explanatory power of models with the market-level variables
an
appears to be greater than that of models with only firm-level variables, as in
Boonchuaymetta and Chuanrommanee (2013). A likely explanation for this finding is that
M
market conditions tend to play a relatively greater role in determining the IPO’s short-run
Prior studies using the US data, such as Ritter (1991), Schultz (2003), and Chahine and
Goergen (2014), report that IPOs in the United States are underpriced. For instance,
Ac
Aggarwal and Conroy (2000) report that the average initial return for a sample of 188 Nasdaq
IPOs over May-October 1997 period is 19.5%. Likewise, Bradley and Jordan (2002) report
that the average return for a sample of 3,325 IPOs in the United States between 1990 and
1999 is 22.6%, and suggest that the high average initial return in their study is attributable to
the high average return (i.e. greater than 70%) for IPOs in 1999. Other studies that examine
the initial returns of IPOs in developed countries also report evidence of IPO underpricing.
9
Page 11 of 55
For instance, for the IPOs in the United Kingdom, Levis (1993) shows that the underpricing
of the IPOs during 1980–1988 is about 14%, whereas Hoque (2014) finds that the
Evidence of IPO underpricing in small- and medium-sized developed markets has also
been documented. For instance, Saunders and Lim (1990) report the underpricing of IPOs in
t
Singapore, whereas Kunz and Aggarwal (1994) find that there is evidence of IPO
ip
underpricing in a sample of 42 IPOs in Switzerland between 1983 and 1989. In a more recent
cr
study, Drobetz et al. (2005) show that IPOs in Switzerland between 1983 and 2000 are
us
underpriced by approximately 35%.
Several scholars examining IPO data in emerging market countries report that IPOs in
an
these countries are substantially underpriced. For example, Chen et al. (2004) find that the
average adjusted initial return for A-share IPOs in China during the period 1992–1997 is
M
145%. Hensler et al. (2000) find that IPOs in Mexico between 1987 and 1993 are initially
underpriced and that the initial underpricing of bank IPOs is larger than that of non-bank
ed
IPOs. In a recent study, Su and Bangassa (2011) find that the average market-adjusted initial
return for Chinese IPOs during the period 2001-2008 is 118.62%. Allen et al. (1999) report
pt
that the mean initial return for a sample of 150 Thai IPOs during the period 1985–1992 is
ce
63.49%. Kim et al. (2004) find that the mean initial return of the IPOs in Thailand during the
period 1987–1993 is 67.66%. Vithessonthi (2008) reports that the mean initial return of Thai
Ac
IPOs during the period 2000–2005 is 19.97%. In a more recent study, Chorruk and
Worthington (2010) show that the underpricing of Thai IPOs during the period 1997–2008 is
17.60%. Kosala (2011) finds that the mean initial return of 76 IPOs in Indonesia during the
There are a number of explanations for the IPO initial underpricing. First, Rock (1986)
suggests that the extent to which the IPO is underpriced depends upon the ex ante uncertainty
10
Page 12 of 55
about the true value of the IPO firm, thereby suggesting that higher levels of asymmetric
information between better informed investors (e.g., managers) and less informed investors
will lead to higher initial returns. As the level of information asymmetry between firms and
investors in emerging market countries appears to be greater than that of advanced countries,
we therefore expect a larger initial return of the IPOs in emerging market countries, relative
t
to the IPOs in developed countries. The findings that the average initial returns of Chinese
ip
IPOs are substantially higher than that of US IPOs provide support for the information
cr
asymmetry argument. Second, Ritter (1991) proposes another plausible explanation for the
us
IPO underpricing by arguing that if investors are systematically too optimistic about the
prospects of the IPO firms, then IPOs will be underpriced as investors bid up prices of the
an
IPOs. This argument would suggest that high initial returns of IPOs in emerging market
countries are largely driven by optimistic investors rather than by the potentially high level of
M
asymmetric information in the market.
ed
I collect data for all Thai IPOs that began trading on the SET between January 1, 2000
ce
and December 31, 2012 from the SETSMART database, which contains the most
comprehensive information of the securities listed on the SET and Market for Alternative
Ac
Investment (mai). The SET was established in 1975, and is the largest stock exchange in
Thailand, whereas the mai is the second stock market in Thailand, which was established in
1999.10 Given that the IPOs listed in the mai are substantially smaller than the IPOs listed on
10
The main purpose of the establishment of the mai was to serve small firms, but with high growth potential,
that could not meet the listing requirements of the SET. More specifically, the SET’s listings require a minimum
paid-up capital of at least 300 million THB, whereas the mai’s listings require a minimum paid-up capital of 20
million THB.
11
Page 13 of 55
the SET11 and that small firms in the SET at times become illiquid, I mainly focus on the
analysis of the IPOs listed on the SET and exclude all IPOs in the mai so as to be
conservative and reduce noise in the data set. To provide an overview of the development of
the Thai IPO market, Figure 1 illustrates the number of IPOs listed on the SET between 1975
and 2012.
t
ip
[INSERT FIGURE 1 ABOUT HERE]
cr
us
I arbitrarily select year 2000 as the beginning year in this study for two reasons: First, the
Thai IPO market was practically nonexistent in the aftermath of the 1997 financial crisis (i.e.
an
between 1997 and 1999). Second, a new wave of Thai IPO listings began to emerge again in
2000. The initial sample consists of 200 IPOs in Thailand between January 2000 and
M
December 2012. Consistent with Ritter (1991) and Ljungqvist and Wilhelm (2003), I exclude
unit offers and closed-end funds (e.g., property funds) from the sample and firms that do not
ed
disclose the IPO price. Accordingly, the final sample consists of 187 IPOs.
Over the 13-year sample period 2000–2012, the total value of the 187 IPOs at the offer
pt
price is 1,031 billion THB (or approximately 33 billion USD at the exchange rate of 31 THB
ce
per USD), and the total value of the 187 IPOs at the end of the first-trading day price in the
sample is 1,160 billion THB (or approximately 37 billion USD). These figures suggest that
Ac
the IPO firms over the sample period leave 128 billion THB on the table. The largest IPO is
PTT Plc, with an issuing value of 97.93 billion THB at the IPO price, while the smallest IPO
11
For instance, the average issue size for the sample of IPOs in the SET in 2003 is 1,071 million THB versus
127 million THB for the sample of the IPOs in the mai in 2003. That is, the average IPO in the SET is roughly
743% [(1,071-127)/127] larger than the average IPO in the mai. The difference in size remains substantial
throughout the study period as almost all IPOs in the mai are generally not large enough to be listed on the SET.
The average issue size for the IPOs in the SET in 2008, 2011, and 2012 is 2,298 million THB, 1,264 million
THB, and 2,147 million THB, whereas the average issue size for the IPOs in the mai in the respective years is
125 million THB, 166 million THB, and 148 million THB.
12
Page 14 of 55
is Srivichaivejvivat Plc, with an issuing value of 0.168 billion THB at the IPO price. The five
largest IPOs have the combined issuing value of 365.69 billion THB at the IPO price,
accounting for 35.44% of the total value of all IPOs in the sample.
Similar to Ritter (1984), Ljungqvist and Wilhelm (2003), and Cook et al. (2006), I use an
initial return and a market-adjusted initial return as proxies for the short-run performance of
t
the Thai IPOs. The initial return for each IPO is measured as the difference between the first-
ip
day closing price of an IPO and its corresponding offer price relative to the offer price (i.e.
cr
(close price – offer price)/offer price). The market-adjusted initial return, which is also
us
known as the “abnormal” initial return, is then calculated as the difference between the initial
return and the market return on the IPO date. In this study, the return on the SET index12 is
4.1. Are initial public offerings in Thailand underpriced in the short run?
pt
To test whether IPOs in Thailand are underpriced in the short run, I adopt the standard
ce
event-study method. That is, I test the hypothesis that the average initial return for the IPOs is
equal to zero. To test this hypothesis, I employ the t test statistics as follows:
Ac
AIRi
t IR , (1)
(IRi )
n
12
The SET index is the value-weighted market portfolio that includes all common stocks traded on the main
board of the SET.
13
Page 15 of 55
where AIRi is the sample average initial return, calculated as the arithmetic average of initial
returns on all IPOs in the sample of n firms, and (IRi) is the cross-sectional sample standard
deviations of initial returns for the sample of n firms. I measure an initial return as the
difference between the first-trading day closing price and the IPO offer price divided by the
IPO offer price (i.e. (close price – IPO price)/IPO price), which is also known as offer-to-
t
close return, as shown in Aggarwal and Conroy (2000) and Cook et al. (2006). A positive
ip
initial return indicates the IPO underpricing.
cr
Table 1 presents average initial returns, standard deviations, minimum initial returns,
us
maximum initial returns, and percentages of positive initial returns for 187 Thai IPOs
between 2000 and 2012 by year of issuance (Panel A) and by industry (Panel B). The initial
an
returns are averaged across all IPOs for each year of issuance and the full sample. For the full
sample, the average initial return for the Thai IPOs is 18.00% and is statistically significant at
M
the 1% level. There appears to be a relatively wide range of initial returns evidenced by the
fact that the smallest initial return is –36.40% and the largest initial return is 200.00%.
ed
In comparison with IPO studies using the US data, the magnitude of the initial returns of
the IPOs in this study is relatively lower than that of US IPOs reported in the literature. For
pt
instance, Ljungqvist and Wilhelm (2003) report that the average initial return for the US IPOs
ce
during the hot markets are very high; that is, the average initial returns for IPOs in 1999 and
2000 are 73.3% and 57.7%, respectively. In a related study, Cook et al. (2006) show that the
Ac
average initial return for a sample of 3,026 US IPOs between 1993 and 2000 is 28.95%.
As Lam et al. (2007) suggest that the IPO underpricing tends to be lower in countries
with better investor protections, the finding that the mean initial return of Thai IPOs during
the period 2000–2012 is lower than that of Thai IPOs in the pre-1997 financial crisis period is
consistent with the view that financial and legal reforms in Thailand following the 1997
financial crisis had led to the introduction of better regulations protecting investors’ rights.
14
Page 16 of 55
I find that the number of positive initial returns accounts for 64.17% of the IPOs in the
sample. In addition, Table 1 shows that average initial returns for IPOs issued in 2012 is the
largest (69.38%) and is statistically significant the 5% level, and that average initial returns
for IPOs issued in 2003, 2004 and 2005 are positive and statistically significant at the 5%
level. Of all subsample periods, I find that the average initial returns for the Thai IPOs issued
t
in 2006, 2008, and 2009 are small (i.e. less than 2%) and statistically insignificant. Low
ip
initial returns for the IPO issuance in 2000 could be attributed to two reasons: First, the
cr
economy of Thailand had just recovered from the 1997 Asian financial crisis, causing the
us
initial wave of IPOs after the financial crisis to be less well received. Second, the number of
IPOs issued in 2000 is very small (i.e. 2 IPOs). A close inspection also shows that almost half
an
the IPO issuance in 2006 occurred after the military coup in Thailand took place on 19
September 2006, leading to several negative initial returns in that period. Overall, the
M
political instability in Thailand in 2006 could have affected the IPO markets. The low initial
returns for the IPO issuance in 2008 and 2009, respectively, are consistent with the falling
ed
stock markets across developed and developing countries following the 2007–2008 financial
As Ritter (1991) notes that the IPO firms in the U.S. between 1975 and 1984 are not
evenly distributed over all industries, and that there is substantial variation in the aftermarket
performance of the IPO firms across different industries, I examine whether the initial returns
of IPOs in Thailand differ across industries. To investigate the industry effect on the IPO
initial returns, I split the sample of 187 IPOs into eight subsamples, according to the SET’s
industrial classification.
15
Page 17 of 55
Panel B of Table 1 reports the mean initial returns for the sample IPOs between 2000 and
2012 by industry. As the sample size of the agro and food subsample is 6 and the subsample
size of the consumer products subsample is only 3, the results for these subsamples are not
meaningful, and thus are not discussed in detail. For other industries, I find that average
initial returns appear to vary across industries, ranging from the average initial return of
t
8.09% for the property and construction subsample to the average initial return of 35.35% for
ip
the financial subsample. In addition, the mean initial returns for five of the eight subsamples
cr
are positive and statistically significant.
us
Overall, the results reported in Table 1 suggest that IPOs in Thailand between January
2000 and December 2012 are on average underpriced by 18.00%, and that the underpricing
an
of the Thai IPOs varies by year of issuance and by industry.
I then test the hypothesis that the average market-adjusted initial returns for the sample
M
of Thai IPOs are not statistically different from zero by employing the t test statistics as
follows:
ed
AARi
pt
t AR , (2)
( ARi )
n
ce
Ac
where AARi is the sample mean market-adjusted initial return, calculated as the arithmetic
average of market-adjusted initial returns on all IPOs in the sample of n firms, and (ARi) is
the cross-sectional sample standard deviations of market-adjusted initial returns for the
return as an initial return less the one-day return on the market on the IPO date. Accordingly,
a positive market-adjusted initial return implies a better performance of the IPO relative to its
16
Page 18 of 55
Table 2 reports the mean market-adjusted initial returns for the sample IPOs by their year
of issuance (Panel A) and by industry (Panel B). The results reported in Table 2 are generally
consistent with the results reported in Table 1. The average market-adjusted initial return for
the full sample is 18.02% and is statistically significant. For subsample periods, I find that the
pattern of the market-adjusted initial return subsamples is largely consistent with that of the
t
initial returns reported in Table 1. For instance, the average market-adjusted initial returns for
ip
the IPO issuance in 2003, 2004, 2005, and 2012 subsamples remain positive and statistically
cr
significant at the 5% level or less. The average market-adjusted initial returns for the IPO
us
issuance in 2002 and 2010 subsamples are positive and statistically significant only at the
10% level. In comparison with other studies, the average market-adjusted initial return for the
an
Thai IPOs in this study is relatively smaller. For instance, Drobetz et al. (2005) report that the
Panel B of Table 2 reports the average market-adjusted initial returns for the IPOs by
pt
industry. The mean initial abnormal returns for five of the eight industries are positive and
ce
statistically significant. More specifically, the mean market-adjusted initial returns for the
financials and resources subsamples are 35.48% and 31.04%, respectively and statistically
Ac
significant at the 1% level. The mean market-adjusted initial returns for the property and
construction, services, and technology subsamples are 7.98%, 26.08%, and 21.68%,
Overall, the results in Tables 1 and 2 indicate that there is evidence of the IPO
underpricing during the period 2000–2012. I also find that the mean initial return of the Thai
IPOs is substantially lower than that of the Thai IPOs during the period 1987–1993, as
17
Page 19 of 55
reported by Kim et al. (2004), and during the period 1985–1992, as reported by Allen et al.
(1999). More importantly, the initial return of the Thai IPOs during the period 2000–2012 is
significantly lower than the that of the IPOs in China during the period 2001–2008, as
reported by Su and Bangassa (2011). In addition, the initial return of the Thai IPOs is very
close to that of the IPOs in more advanced markets in Asia, such as Hong Kong, as reported
t
by Mazouz et al. (2009). Furthermore, the magnitude of market-adjusted initial returns for the
ip
Thai IPOs varies across industries.
cr
us
4.2. Do financial development and interest-rate differentials affect IPO underpricing?
4.2.1. The effects of stock market conditions on the short-run stock returns of the IPOs
an
In this section, I estimate a series of regressions of market-adjusted initial returns on a set
of explanatory and control variables. As initial returns and market-adjusted initial returns are
M
almost identical, I therefore report my results based on using the market-adjusted initial
return as the dependent variable in order to conserve space. The issue size of the IPO can be
ed
used as a measure of the level of predisclosure information about the IPO in the market and
the level of information asymmetry (e.g., Christensen et al., 2004; Cox and Roden, 1999).
pt
Large firms are expected to have more resources for promoting their IPOs than smaller firms.
ce
Hence, the level of information asymmetry is expected to be relatively lower for larger firms
than smaller firms. Thus, issue size should be negatively related to the underpricing.
Ac
However, Cox and Roden (1999) find that issue size has a positive effect on the adjusted one-
day abnormal returns of the mutual thrift IPOs in the United States.
Empirical research shows that market conditions affect the short-run stock price
performance. For instance, Su and Bangassa (2011) find that market conditions, measured as
the three-month weighted average return on the market prior to listing, have a positive effect
on the market-adjusted initial return of the IPO in China. Likewise, Dey (2005) finds that the
18
Page 20 of 55
size of stock markets (i.e. market capitalization) is a determinant of portfolio turnover in a
sample of 48 stock exchanges during the 1995–2001 period. Kim and Ritter (1999) find that
market price-to-earnings (P/E) ratios and market-to-book (P/BV) ratios of comparable firms
could, to some extent, be used in IPO valuation. Prior studies, such as Bouis (2009), show
that the level of the stock market index as a measure of stock market valuations has a
t
predictive effect on the timing of IPOs. Therefore, I test the impact of stock market
ip
development and stock market conditions on the initial returns.
cr
Following Chinn and Ito (2006), who use the level of stock market capitalization as a
us
measure of equity market development, I use the level of market capitalization of the SET as
a proxy for stock market development (SMD).13 As larger stock markets tend to attract more
an
foreign investors, the size of the stock market should signal the overall attractiveness of the
stock market at the time of IPO listing. I use the stock market’s P/E ratio, P/BV ratio, and
M
yield as proxies for stock market conditions. Accordingly, I estimate the following
regressions:
ed
(3)
6 INDi i ,
ce
and
Ac
13
Some scholars such as Hondroyiannis et al. (2005) and Afonso and Jalles (2013) use the ratio of stock market
capitalization to GDP as a proxy for stock market development and the size of the stock market. I also test the
robustness of my results by using this proxy. See Section 4.4.
19
Page 21 of 55
where IRi denotes the initial return of IPO i; ISSUEi denotes the issue size of IPO i, measured
as the natural logarithm of the real gross proceeds of the IPO (in millions of THB)14; SMDi
denotes stock market development, measured as the natural logarithm of the real market
capitalization of the SET (in millions of the THB) on the IPO date; MKTPEi is the SET’s
price-to-earning (P/E) ratio on the IPO date; MKTPBVi is the SET’s price-to-book value
t
(P/BV) ratio on the IPO date; and MKTYIELDi is the SET’s yield on the IPO date. INDi is a
ip
vector of seven industry-dummy variables (i.e. from IND 1 through IND 7), and YEARi is a
cr
vector of 12 IPO year-dummy variables. All model specifications are estimated using the
us
OLS regression with the White’s heteroskedasticity consistent covariance matrix.
Table 3 presents the summary statistics of major variables in this study. The IPO prices
an
range from 1.25 THB to 82.19 THB, and the mean IPO price is 12.06 THB. The average
gross proceed and real gross proceed are 5,517 and 5,999.38 million THB, respectively, and
M
the average market capitalization of the SET on the IPO date is 4,676.43 billion THB. The
market P/E ratios range from 3.37 to 27.88, with the mean ratio of 10.68. The mean market
ed
P/BV ratio is 1.83, while the mean market yield is 3.21. The mean amount of the outstanding
government bonds is 1,343.77 billion THB, which is roughly one-third of the mean market
pt
capitalization of the SET. The mean amount of the outstanding private debt securities is
ce
625.45 billion THB, which is about 13.46% of the mean market capitalization of the SET. In
addition, the average volume of the outstanding T-bills is 145.05 billion THB, which is
Ac
approximately 3% of the mean market capitalization of the SET. These figures suggest that
the combined market of T-bills, government bonds, and private debt securities is on average
about half the size of the SET. For comparison purposes, I find that Thailand’s GDP in 2000,
2005, and 2010 is 5,242.9, 6,836.9, and 8,176.2 billion THB, respectively, thereby roughly
14
The issue size and stock market capitalization values are adjusted for inflation using the core consumer price
index (CCPI) with the base year of 2011 = 100.
20
Page 22 of 55
indicating that the outstanding amount of the government bonds relative to the size of the
GDP during the study period is not excessively high. Table 3 also shows that the mean one-
year government bond yield is slightly higher the mean US federal funds rate.
t
[INSERT TABLE 4 ABOUT HERE]
ip
cr
Table 4 presents the regression results of the one-day market-adjusted initial return for
us
IPOs on ISSUE, SMD, MKTPE, MKTPBV, MKTYIELD, industry-dummy variables, and year-
dummy variables. As MKTPE, MKTPBV, and MKTYIELD are somewhat correlated15, I add
an
each of them in Regression (3) separately in Models 1–3 of Table 4. Models 1–4 present the
estimation results for Regression (3) with industry-dummy variables. In Model 5, I replace
M
industry-dummy variables with year-dummy variables.
The most important observation from the results in Models 1–5 is that the coefficients on
ed
ISSUE in all regressions are negative and statistically significant, suggesting that the issue
size of the IPO has a negative effect on the initial return. This result is inconsistent with a
pt
study of Derrien (2005), reporting that the issue size of the IPO is not related to the initial
ce
return of the IPO in France. Likewise, Aggarwal (2000) also reports that the issue size of the
IPO exerts no effect on the market-adjusted initial returns. Furthermore, this finding is in
Ac
contrast with Samarakoon (2010), who finds that there is a positive and significant difference
in the initial return between small and large IPO issues. However, the significance of the
issue size in all models is consistent with Cox and Roden (1999). As discussed earlier,
empirical findings of the effect of gross proceeds on the initial return of the IPOs is mixed.
15
The correlation coefficient between the market P/E ratio and market yield 0.272 and is statistically significant
at the 1% level. The correlation coefficient between the market P/E ratio and P/BV ratio is 0.08 and is
statistically insignificant. The correlation coefficient between the market P/BV ratio and market yield is -0.393
and statistically significant at the 1% level.
21
Page 23 of 55
Kirkulak and Davis (2005) report that the insignificant effect of gross proceeds on the initial
The results in Table 4 show that the coefficient on SMD is statistically significant in only
one (i.e. Model 3) of the five models, suggesting that the size of the stock market appears to
have no effect on the initial return of the IPOs. In Model 1, the coefficient on MKTPE is
t
positive and statistically insignificant. The results in Model 2 show that the coefficient on
ip
MKTPBV is positive and statistically significant, suggesting that the stock market’s P/BV
cr
ratio as a proxy for the stock market conditions is positively associated with the short-run
us
stock price performance of the IPO. The results in Model 3 indicate that the coefficient on
MKTYIELD is negative and statistically significant, suggesting that stock market’s yield is
an
negatively associated with the initial return of the IPO.
Inconsistent with the results of Model 3, the coefficient on MKTYIELD is insignificant. The
ed
negatively correlated with MKTPBV and positively correlated with MKTPE, indicating the
ce
potential presence of multicollinearity among the stock market conditions variables in the
results of Regression (3) in Model 4. However, the VIF value for MKTPE, MKTPBV and
Ac
multicollinearity issue is not of great concern. The results in Model 5 show that both
MKTPBV ratio and MKTYIELD are positive and statistically significant, after controlling for
year effects.
Thus far, I show that the stock market development and stock market conditions (i.e. the
market’s P/BV ratio, P/E ratio and yield) are associated with the one-day market-adjusted
22
Page 24 of 55
initial return of the Thai IPOs, after controlling for industry or year effects. These results are
largely in line with Derrien and Womack (2003), who find that stock market returns have a
positive effect on the initial return. I also find that the issue size is negatively related to the
t
4.2.2. The effects of bond market development on the short-run stock returns of the IPOs
ip
A novel aspect of this paper compared to previous studies is that I examine whether bond
cr
market development explains the initial return of the IPOs (see Yong (2007) for a detailed
us
review of IPO research in Asia.). First, I test whether the amount of outstanding bonds has a
significant effect on the initial return. Given that the amount of the outstanding bonds signals
an
(1) the size of the bond markets and (2) the level of development of the bond markets, 16 it is
reasonable to expect that if bonds markets are sufficiently large, an IPO will be more
M
attractive to foreign investors as large financial markets lead to higher levels of liquidity in
the markets. Throughout the paper, financial markets refer to banking, bond and equity
ed
markets. It should be noted that Zagorchev et al. (2011) measure financial development size
by using the ratio of the sum of stock market capitalization, public and private bond market
pt
capitalization and depository money bank assets to GDP. My hypothesis is that there is a
ce
positive linkage between the size of the bond markets and the initial return. On the other
hand, it is also possible that when bond markets are large, firms that want to go public will
Ac
face a greater level of competition in the market for funds. Thus, my alternative hypothesis is
that the size of the bond markets has a negative effect on the initial return.
To test whether the bond market development has a significant effect on the short-run
stock price performance of the IPOs, I use the outstanding T-bills, government bonds, and
16
Several scholars, such as Afonso and Jalles (2013), use the ratio of bond market capitalization to GDP as a
proxy for bond market development. I also use this ratio to test the robustness of my results. See Section 4.4.
23
Page 25 of 55
private debt securities as proxies for the size of public and private bond markets.17 Hence, I
t
ip
and
cr
us
IRi 0 1 ISSUEi 2 SMDi 3TBMDi 4GBMDi 5 PBMDi
(6)
6YEARi i ,
an
where TBMDi denotes T-bills market development, measured as the natural logarithm of the
M
amount of the outstanding T-bills (in millions of the THB) at the end of the month prior to
the IPO date; GBMDi is government bond market development, measured as the natural
ed
logarithm of the amount of the outstanding government bonds (in millions of the THB) at the
end of the month prior to the IPO date; PBMDi is private bond market development,
pt
measured as the natural logarithm of the amount of the private debt securities (in millions of
ce
the THB) at the end of the month prior to the IPO date. The amounts of outstanding t-bills,
government bonds, and private debt securities can alternatively be labelled as t-bill market
Ac
17
Some scholars, such as Levine and Zervos (1998) and Boyd and Jalal (2012), use (1) the ratio of the private
bank lending to GDP and (2) the ratio of private bond outstanding to GDP to measure debt market development.
I use these measures in my robustness tests, which do not give different results. See Section 4.4.
24
Page 26 of 55
Table 5 presents the regression results of the one-day market-adjusted initial return for
IPOs on ISSUE, SMD, TBMD, GBMD, PBMD, industry-dummy variables, and year-dummy
variables. As TBMD, GBMD and PBMD are highly correlated, I add each of these variables
in the regression separately. Models 1–5 of Table 5 present the estimated parameters for
t
Regression (5) with industry dummy variables, whereas Model 6 in Table 5 presents the
ip
estimation results for Regression (5) with year dummy variables.
cr
Consistent with the results in Table 4, the estimated coefficients on ISSUE in Models 2,
us
3, 4, and 5 in Table 5 are negative and statistically significant at the 5% level or less. This
finding is consistent with the finding of Su and Bangassa (2011), who show that issue size
an
has a negative effect on the market-adjusted initial return of the IPO in China. I find that the
coefficient on GBMD is negative and significant. The results reported in Model 3 show that
ed
the coefficient on PBMD is negative and significant. As SMD is highly correlated with
GBMD (r = 0.826, p-value < 0.001) and PBMD (r = 0.843, p-value < 0.001), the results of
pt
Model 2 and 3 should be interpreted with caution. In Section 4.4, I replace SMD, TBMD,
ce
GBMD, and PBMD with the ratios of stock market capitalization, T-bills market
capitalization, bond market capitalization, and private bond market capitalization to GDP, I
Ac
find that these variables are no longer highly correlated. All of these results in Table 5
indicate that the bond market development has a negative effect on the initial return of the
IPOs.
The results in Model 4 show that the TBMD coefficient remains significant; however, the
coefficients on GBMD and PBMD are no longer significant. A plausible explanation for these
conflicting results is the presence of multicollinearity in Model 4, as indicated by the fact that
25
Page 27 of 55
VIF value for GBMD and PBMD is larger than 10. I therefore re-estimate Regression (5)
again by dropping the GBMD from the specification. In Model 5, the coefficients on both
TBMD and PBMD are negative and statistically significant at the 5% level. Overall, the
results in Models 1–5 of Table 5 suggest that after controlling for industry effects, the bond
market development leads to lower initial returns of the IPOs in Thailand. The results in
t
Model 6 indicate that after controlling for year effects, the bond market development
ip
variables are no longer associated with the initial return. One plausible explanation is that the
cr
bond market development variables are correlated with time, which removes the effect of the
us
bond market development on the initial return.
As the IPOs compete with other new securities for a supply of funds (i.e. other securities
an
might shift the demand curve for the IPOs inwards), I therefore test whether the amount of
new fixed-income issuances affects the initial return. Accordingly, I estimate the following
M
regressions:
ed
where NTBILLi denotes new T-bills, measured as the natural logarithm of the amount of new
T-bills issuance (in millions of the THB) for the month prior to the IPO date; NGBONDi is
Ac
new government bonds, measured as the natural logarithm of the amount of new government
bonds issuance (in millions of the THB) for the month prior to the IPO date; NPBONDi is
new private debt securities issuance, measured as the natural logarithm of the amount of
private debt securities (in millions of the THB) for the month prior to the IPO date; and other
26
Page 28 of 55
[INSERT TABLE 6 ABOUT HERE]
Table 6 presents the regression results of the one-day market-adjusted initial return for
the IPOs on ISSUE, SMD, NTBILL, NGBOND, NPBOND, and industry-dummy variables.
Models 1–6 of Table 6 report the estimated parameters for Regression (7).
t
In Model 1, the coefficient on NTBILL is negative and statistically significant, indicating
ip
that new T-bills issuance has a negative effect on the initial return. The coefficient on
cr
NGBOND in Model 2 is negative and statistically significant, suggesting that new
us
government bonds issuance has a negative effect on the initial return. The estimated
coefficient on NPBOND in Model 3 is positive and insignificant, indicating that new private
an
debt securities issuance has no effect on the initial return.
In Model 4, the coefficients on NTBILL, NGBOND, and NPBOND are not significant. As
M
NTBILL and NGBOND are highly correlated, I estimate Regression (7) by using only two of
the three new issuance variables at a time in Models 5 and 6. The results show that the
ed
coefficients on NTBILL and NGBOND in Models 5 and 6, respectively, are negative and
statistically significant, whereas the coefficients on NPBOND are not significant in both
pt
models. Overall, the findings suggest that the new T-bills and new government bond have a
ce
4.2.3. The effects of interest-rate differentials on the short-run stock returns of the IPOs
Do interest-rate differentials affect the short-run stock price performance of the IPOs? As
noted by Coudert and Mignon (2013) that in terms of carry-trade gains, the THB is one of the
most rewarding currencies18, my hypothesis is that the interest-rate differential influences the
18
It should be noted that the length between the IPO subscription date (i.e. the last subscription date) and its
corresponding first-trading date for most IPOs in the sample is fairly short (i.e. roughly about one week on
average), which is very similar to what was reported by Derrien and Womack (2003).
27
Page 29 of 55
IPO underpricing in these markets. In addition, Elyasiani and Mansur (1998) find that
changes in interest rate levels are associated with the distribution of stock returns of banks,
whereas Koch and Saporoschenko (2001) report that increases in interest rate levels have a
negative effect on stock returns of Japanese keiretsu financial firms. Hence, I examine
t
Thailand and the United States, is associated with the initial return. The positive relation
ip
between the interest-rate differential and the initial return might indicate that carry-trade
cr
strategies are carried out in the IPO markets as well. That is, investors participate in the IPO
us
markets by taking a short position in low-yield currencies when high-yield currencies are not
expected to depreciate. This type of a carry-trade strategy is, however, relatively riskier than
an
typical carry-trade strategies that take a long position in high-yield bonds in one currency and
that the term spread on government bonds can be used as a predictor of real activity in the
ed
economy. Using data over the 1980–1999 period in South Korea, Paya and Matthews (2004)
find that the term spread has a positive effect on future economic growth in South Korea,
pt
providing further evidence of the predictive power of the term spread on future real economic
ce
activity in Asia. In addition, Schrimpf (2010) finds that the term spread has a predictive effect
on stock market returns. Borrowing an argument from Ivanov and Lewis (2008) that the term
Ac
spread could be used as a proxy for the cost of capital and that a lower term spread implies
that the long-term return on a risk-free asset is falling, I expect that the term spread, measured
as the difference between the yield on the 10-year government bond and the yield on the one-
As Katechos (2011) reports that exchange rate returns are significantly associated with
stock market returns, I use the real effective exchange rate (REER) of the THB, as a measure
28
Page 30 of 55
of the value of the THB, to control for the currency effect. An increase in the REER indicates
the appreciation of the THB, which will in turn deteriorate the competitiveness of Thailand in
t
6 REERi 7 INDi i ,
ip
cr
where INTi denotes the domestic interest rate, measured as the yield on a short-term (one-
us
year) Thai government bond at the end of the month prior to the IPO date (i.e. due to missing
data on the one-month Thai government bond yield in early periods in the sample, the one-
an
year Thai government bond yield is therefore used as a proxy for the short-term interest rate);
TSi denotes the term spread on government bonds, measured as the difference between the
M
yield on a 10-year government bond and the yield on the one-year government bond at the
end of the month prior to the IPO date; INTDIFi denotes the interest-rate differential,
ed
measured as the difference between a yield on the one-year government bond in Thailand and
the US federal funds rate at the end of the month preceding the IPO date; and REERi denotes
pt
the real effective exchange rate of the THB at the end of the month prior to the IPO date.
ce
returns of the IPOs. The interest-rate differentials are shown on the x-axis, while the one-day
market-adjusted initial returns are plotted on the y-axis. Interestingly, Figure 2 appears to
show that IPOs tend to be listed during periods of positive interest-rate differentials. Table 7
presents the regression results of the one-day market-adjusted initial return for IPOs on
ISSUE, SMD, INT, TS, INTDIF, and industry-dummy variables. Models 1–5 in Table 7 report
29
Page 31 of 55
[INSERT FIGURE 2 ABOUT HERE]
t
that higher domestic interest rates are associated with lower initial returns. The estimated
ip
coefficient on TS in Model 1 is statistically insignificant, indicating that the term spread has
cr
no effect on initial returns. This result is inconsistent with the finding that the term spread on
us
government bonds is associated with stock market returns reported by Schrimpf (2010). I also
find that the estimated coefficients on INTDIF and REER in Model 1 are statistically
an
insignificant, indicating that the interest-rate differential and real effective exchange rate of
the THB have no effect on the initial return. As INTDIF and TS variables are highly
M
correlated (r = –0.70, p-value = 0.001), I re-estimate Model 1 without TS in the model
specification. In untabulated results, I find that the INT coefficient remains negative and
ed
As the effect of the interest-rate differential should be evident during periods of high
pt
interest-rate differentials, I first add an interactive term between INT and ID1, which takes a
ce
value of one if the interest-rate differential is larger than the median value and zero
otherwise, in Model 2. The results show that the estimated coefficient on the interaction term
Ac
is not statistically significant. I repeat the regression estimation for Model 2 again without the
term spread varaible. The results show that domestic interest rate is negative and significant
at the 5% level. The estimated coefficients on the interest-rate differential and the interactive
term involving interest-rate differential and the ID1 dummy variable become significant at
the 10% level. I then add an interaction term involving INTDIF and ID2, an indicator equal to
30
Page 32 of 55
one if the interest rate differential is positive, and zero otherwise, in Model 3. I find that the
To further check whether the effect of the interest-rate differential on the initial return is
conditional on whether the interest-rate differential is positive, I spilt the sample into two
subsamples: the negative and the positive interest-rate differential subsamples. In Model 4 of
t
Table 7, I estimate Regression (8) using the negative interest-rate differential (INTDIF < 0)
ip
subsample. The estimated coefficient on INTDIF is not significant in Model 4. In Model 5, I
cr
estimate Regression (8) using the positive interest-rate differential (INTDIF ≥ 1) subsample.
us
As expected, after controlling for INT, TS, REER, the INTDIF coefficient is positive and
significant, suggesting that the interest-rate differential affects the initial return during
an
periods of positive interest-rate differentials. Furthermore, in Model 5, the REER coefficient
is negative and significant, suggesting that while high interest-rate differentials potentially
M
attract foreign investors into the IPO markets, a strong currency appears to have a negative
In this section, I test whether the same set of variables equally drives the initial returns
for “hot” and “cold” IPOs. Following Brau and Fawcett (2006), I spit the sample into a “hot”
Ac
IPO subsample and a “cold” IPO subsample by defining a hot IPO firm when a firm’s initial
return is greater than 10% and a cold IPO firm when a firm’s initial return is less than 10%.
The “hot” IPO subsample then consists of 76 IPOs, with the mean initial return of 49.27%
and the mean issue size of 4,456 million THB. The “cold” IPO subsample consists of 111
IPOs, with the mean initial return of –3.42% and the mean issue size of 6,245 million THB.
31
Page 33 of 55
[INSERT TABLE 8 ABOUT HERE]
I first re-estimate Regressions (3)–(8) using the market-adjusted initial return as the
dependent variable. Due to the smaller size of the subsamples, I do not estimate regressions
that include the year-dummy variables. For the sake of brevity, I do not report all tabulated
t
results. In Model 1 of Table 8, I show that in the full sample, the ISSUE, SMD, MKTPBV
ip
coefficients are statistically significantly. In Models 2–3, I show that the ISSUE, SMD,
cr
MKTPBV, TBMD, PBMD, INT, INTDIF, and REER variables have no effect on the initial
us
returns in the “cold” IPO subsample (see Model 3) but the effects of the ISSUE, SMD,
MKTPBV, and INT variables on the initial return become statistically significant in the “hot”
an
IPO subsample. In Models 4–5, I re-estimate Models 2–3 with the year-dummy variables and
find that the effect of the interest-rate differential on the initial return remains evident in the
M
“hot” IPO subsample.
One implication of my domestic interest rate and interest-rate differential results pertains
ed
to the IPO timing. It might be that high interest rates either shift the demand curve for the
IPOs inwards or lower the slope of the demand curve for the IPOs. All else equal, both
pt
changes would separately lead to lower initial returns. A second, perhaps more plausible,
ce
explanation is that during periods of high interest-rate differentials, foreign investors (or
investors who have access to international capital sources) participate more in the IPO
Ac
markets, leading to greater demand for the IPOs, which not only make up for a weaker
demand for the IPOs from domestic investors (due to higher domestic interest rates) but also
lead to higher initial returns. The fact that both domestic interest rates and interest-rate
differentials have no effect on the initial return for a sample of “cold” IPOs are consistent
with the view that investors are unlikely to undertake carry-trade strategies when the
expected return on the investment (e.g., the IPOs in this case) is low. The fact that the
32
Page 34 of 55
average initial return for the “cold” IPOs is –3.42% simply rule out the carry-trade activities,
leading to no observed effects of the domestic interest rates and interest-rate differentials on
t
ip
In this section, I repeat a number of regression analyses using alternative measures for
cr
some variables under study and using different model specifications. First, I re-estimate
us
Regression (8) without the term spread variable for the full sample using different measures
for the interest-rate differential: (1) the interest-rate difference between the one-year Thai
an
government bond yield and the one-month LIBOR and (2) the interest-rate difference
between the one-year Thai government bond yield and the one-year LIBOR. As before, all of
M
these rates are recorded at the end of the month prior to the IPO date. For brevity, I do not
report the tabulated results. Interest-rate differential between the one-year Thai government
ed
bond yield and the one-month LIBOR has a positive effect on the initial return in the positive
interest-rate differential subsample; however, the interest-rate differential between the one-
pt
year Thai government bond yield and the one-year LIBOR is not significant. These findings
ce
suggest that the interest-rate differential results are sensitive to proxies for foreign interest
rates. A plausible explanation is that in my sample, the one-year LIBOR rate (2.86% on
Ac
average) is higher than both the US federal fund rate (2.24% on average) and the one-month
LIBOR rate (2.33%), which would in turn make carry-trade strategies in Thailand less
profitable. As participants in the IPO markets are like to take a long position in the IPOs for a
short period of time, using the short-term interest rate (e.g., the one-month LIBOR rate),
rather than the long-term interest rate (e.g., the one-year LIBOR), as a measure for the
33
Page 35 of 55
To make the results of this study comparable to prior studies in the financial
development literature and to address concerns about the multicollinearity issue (i.e. between
stock market development and bond market development variables in the original regression
analyses), I re-estimate the regressions in Tables 5 and 8 using the ratio of stock market
capitalization, T-bills market capitalization, bond market capitalization, and private bond
t
market capitalization to GDP as defined in the financial development literature. To measure
ip
equity market development, I use the ratio of stock market capitalization to GDP. To measure
cr
t-bill market development, I use the ratio of t-bills market capitalization to GDP. To measure
us
government bond development, I use the ratio of government bond capitalization to GDP. I
use the ratio of private debt market capitalization to GDP to measure private bond market
an
development. Since the GDP data are at the annual level, I use a one-year lagged GDP in the
estimation of these ratios. I find that the T-bills market development has a negative effect on
M
the initial return of the IPOs, whereas the private bond market development has a positive
While there is evidence in Tables 4 and 8 that the positive effect of stock market
conditions, measured by market P/BV ratio, on the initial return is robust to different model
pt
rather than market P/BV ratio, were used instead. I use four market return variables (AR20,
AR40, BHR20, and BHR40), which are conceptually similar to those used by Derrien and
Ac
Womack (2003). I construct the market return variables using the daily returns on the SET
index in the pre-IPO period for each IPO. AR20 is the simple 20-day accumulative market
returns prior to the IPO, which is the sum of the daily returns on the SET index during a 20-
trading day period prior to the IPO date for each IPO. I repeat this estimation for a 40-day
period prior to the IPO date to construct AR40. To make my results more comparable with
Derrien and Womack (2003), I construct two market return variables, BHR20 and BHR40 as
34
Page 36 of 55
20-day and 40-day buy-and-hold returns of the SET index prior to the IPO date, respectively.
I rerun Model 2 of Table 4 and all Models of Table 8 using each of four market return
variables separately as a measure for stock market conditions. The mean value for AR20,
AR40, BHR20 and BHR40 is 1.02%, 2.92%, 0.89% and 2.75%, respectively. In unreported
results, I find a similar pattern for the coefficients; that is, the positive impact persists, and the
t
coefficients are positive and statistically significant, after controlling for industry effects.
ip
However, I notice that the adjusted R2 for almost all models are smaller than those reported in
cr
Tables 4 and 819, so I feel safe in concluding that market P/BV ratio is a reliable measure for
us
stock market conditions in this context and produces the higher adjusted R2 than do the four
alternative market return variables used in the robustness tests. More importantly, it is a
an
simple measure to be used, requiring fewer calculations. It is natural then to ask whether
lagged market P/BV ratios, rather than the market P/BV ratios on the IPO date, would alter
M
the results. I reestimate Model 2 of Table 4 and all models of Table 8 using the one-day
lagged market P/BV ratio (i.e. at t = –1 relative to the IPO date) as a proxy for the stock
ed
market condition. I find that the coefficients on the one-day lagged market P/BV ratio are
positive and significant as those reported in Tables 4 and 820, thereby providing additional
pt
evidence to suggest that the market P/V ratio, both lagged and contemporaneous values, is a
ce
determinant of IPO underpricing, and, more importantly, can be used as a simple alternative
measure to stock market returns prior to the IPO date as the stock market condition variable.
Ac
19
With respect to Model 2 of Table 4, the adjusted R2 for all regressions using the market return variables are
not larger than 0.18 (compared with 0.23 in Table 4). Likewise, With respect to Model 1 of Table 8, the adjusted
R2 for all regressions (with industry-dummy variables) using the market return variables are not larger than 0.21,
whereas the adjusted R2 for Model 1 is 0.27. It is noteworthy that while my results are not directly comparable
to Derrien and Womack (2003), the adjusted R2 in Model 2 of Table 4 is almost as large as the largest R2 for
similar models their study (see, e.g., Table 3, p. 42).
20
A key difference is that the use of the one-day lagged market P/BV ratio produces the relatively smaller
adjusted R2 for all models in Table 8. For instance, it falls from 0.27 to 0.25 in Model 1, and from 0.21 to 0.20 in
Model 2.
35
Page 37 of 55
5. Conclusion
With the exception of a few earlier studies, such as Hensler et al. (2000), Chen et al.
(2004), Mazouz et al. (2009), Kosala (2011), Su and Bangassa (2011), and Boonchuaymetta
and Chuanrommanee (2013), we have limited knowledge of the IPOs in emerging market
t
countries in recent years, especially after the 1997 Asian financial crises. Using a sample of
ip
187 IPOs in Thailand between January 2000 and December 2010, I investigate whether the
cr
financial development, domestic interest rates, and interest-rate differentials simultaneously
us
explain the variation in the short-run IPO underpricing.
The effects of the stock market development on the initial return are mixed. Overall,
an
these results seem to suggest that stock market development drives the initial return of the
IPOs by (1) lowering the level of information asymmetry and (2) attracting foreign investors
M
into the IPO market. The stock market’s P/BV ratio is positively related to the initial return. It
is possible that the development of stock markets generally results in lower initial returns,
ed
while favorable stock market conditions lead to higher initial returns as they shift the demand
To the best of my knowledge, this paper is one of the first to examine the effects of bond
ce
market development on the IPO underpricing. The T-bills market capitalization, government
bond market capitalization, and private bond market capitalization, which are used as an
Ac
indicator for the development of the bond markets, have a negative effect on the initial return.
In addition, the amounts of new T-bills and government bond issuances have a negative
effect on the initial return, suggesting that all else equal, the development of bond markets
might cause an inward shift of the demand curve for the IPOs, thereby lowering the initial
returns. These findings are novel in the sense that they have not been previously explored.
36
Page 38 of 55
Overall, my results are consistent with the view that the development of the bond markets
The results show that interest-rate differentials has a positive effect on the initial return
when the interest-rate differential is positive, suggesting that IPOs listed in the high interest-
rate differential environment are likely to experience higher initial returns. These results are
t
robust to different measures of the interest-rate differentials. In addition, domestic interest
ip
rates have a negative effect on the initial return in the full sample as well as in the positive
cr
interest-rate differential subsample.
us
Collectively, my findings suggest that the development of the financial markets and
interest-rate differentials are related to the short-run performance of the IPOs. While my
an
results are specific to Thailand, they provide new insights that might be applicable to other
emerging market countries, especially those that have undertaken major financial and legal
M
reforms following financial crises.
ed
References
pt
Adams, B., Carow, K.A., Perry, T., 2009. Earnings management and initial public offerings:
ce
The case of the depository industry. Journal of Banking & Finance 33, 2363-2372.
Afonso, A., Jalles, J.T., 2013. Growth and productivity: The role of government debt.
Ac
Aggarwal, R., 2000. Stabilization activities by underwriters after initial public offerings.
Aggarwal, R., Conroy, P., 2000. Price discovery in initial public offerings and the role of the
37
Page 39 of 55
Allen, D.E., Morkel-Kingsbury, N.J., Piboonthanakiat, W., 1999. The long-run performance
Baillie, R.T., 2011. Possible solutions to the forward bias paradox. Journal of International
t
Thailand. Journal of Multinational Financial Management 23, 272-284.
ip
Bouis, R., 2009. The short-term timing of initial public offerings. Journal of Corporate
cr
Finance 15, 587-601.
us
Boyd, J.H., Jalal, A.M., 2012. A new measure of financial development: Theory leads
Chahine, S., Goergen, M., 2014. Top management ties with board members: How they affect
99-115.
ce
Chen, G., Firth, M., Kim, J.-B., 2004. IPO underpricing in China’s new stock markets.
Chinn, M.D., Ito, H., 2006. What matters for financial development? Capital controls,
Chorruk, J., Worthington, A.C., 2010. New evidence on the pricing and performance of
initial public offerings in Thailand, 1997–2008. Emerging Markets Review 11, 285-
299.
38
Page 40 of 55
Christensen, T.E., Smith, T.Q., Stuerke, P.S., 2004. Public predisclosure information, firm
Cook, D.O., Kieschnick, R., Van Ness, R.A., 2006. On the marketing of IPOs. Journal of
t
Coudert, V., Mignon, V., 2013. The “forward premium puzzle” and the sovereign default
ip
risk. Journal of International Money and Finance 32, 491-511.
cr
Cox, S.R., Roden, D.M., 1999. Initial public offerings by mutual thrifts: The regulatory
us
impact. Journal of Economics and Finance 23, 113-122.
Derrien, F., 2005. IPO pricing in 'hot' market conditions: Who leaves money on the table? .
Dey, M.K., 2005. Turnover and return in global stock markets. Emerging Markets Review 6,
ed
45-67.
Drobetz, W., Kammermann, M., Wälchli, U., 2005. Long-run performance of initial public
pt
offerings: The evidence for Switzerland. Schmalenbach Business Review 57, 253-
ce
275.
Elyasiani, E., Mansur, I., 1998. Sensitivity of the bank stock returns distribution to changes in
Ac
the level and volatility of interest rate: A GARCH-M model. Journal of Banking &
Estrella, A., Hardouvelis, G.A., 1991. The term structure as a predictor of real economic
39
Page 41 of 55
Hensler, D.A., Herrera, M.J., Lockwood, L.J., 2000. The performance of initial public
Hochradl, M., Wagner, C., 2010. Trading the forward bias: Are there limits to speculation?
t
Hondroyiannis, G., Lolos, S., Papapetrou, E., 2005. Financial markets and economic growth
ip
in Greece, 1986–1999. Journal of International Financial Markets, Institutions and
cr
Money 15, 173-188.
us
Hoque, H., 2014. Role of asymmetric information and moral hazard on IPO underpricing and
lockup. Journal of International Financial Markets, Institutions and Money 30, 81-
105.
an
Ivanov, V., Lewis, C.M., 2008. The determinants of market-wide issue cycles for initial
M
public offerings. Journal of Corporate Finance 14, 567-583.
Kamara, A., 1997. The relation between default-free interest rates and expected economic
ed
Katechos, G., 2011. On the relationship between exchange rates and equity returns: A new
pt
approach. Journal of International Financial Markets, Institutions and Money 21, 550-
ce
559.
Kim, K.A., Kitsabunnarat, P., Nofsinger, J.R., 2004. Ownership and operating performance
Ac
in an emerging market: Evidence from Thai IPO firms. Journal of Corporate Finance
10, 355-381.
Kim, M., Ritter, J.R., 1999. Valuing IPOs. Journal of Financial Economics 53, 409-437.
Kirkulak, B., Davis, C., 2005. Underwriter reputation and underpricing: Evidence from the
40
Page 42 of 55
Koch, T.W., Saporoschenko, A., 2001. The effect of market returns, interest rates, and
exchange rates on the stock returns of Japanese horizontal keiretsu financial firms.
Kosala, R., 2011. Intraday patterns of initial public offerings in hot and cold markets: An
t
Kunz, R.M., Aggarwal, R., 1994. Why initial public offerings are underpriced: Evidence
ip
from Switzerland. Journal of Banking & Finance 18, 705-723.
cr
Lam, S.-S., Tan, R.S.-K., Wee, G.T.-M., 2007. Initial public offerings of state-owned
us
enterprises: An international study of policy risk. Journal of Financial and
Ljungqvist, A., Wilhelm, W.J., Jr., 2003. IPO pricing in the dot-com bubble. Journal of
Mazouz, K., Saadouni, B., Yin, S., 2009. Offering methods and issuer-oriented underpricing
ce
costs: Evidence from the Hong Kong IPO market. Journal of International Financial
Moore, M.J., Roche, M.J., 2012. When does uncovered interest parity hold? Journal of
Paya, I., Matthews, K., 2004. Term spread and real economic activity in Korea: Was the
Ritter, J.R., 1984. The "hot issue" market of 1980. Journal of Business 57, 215-240.
41
Page 43 of 55
Ritter, J.R., 1991. The long-run performance of initial public offerings. Journal of Finance
46, 3-27.
Rock, K., 1986. Why new issues are underpriced. Journal of Financial Economics 15, 187-
212.
Samarakoon, L.P., 2010. The short-run underpricing of initial public offerings in the Sri
t
Lankan stock market. Journal of Multinational Financial Management 20, 197-213.
ip
Sarno, L., Schneider, P., Wagner, C., 2012. Properties of foreign exchange risk premiums.
cr
Journal of Financial Economics 105, 279-310.
us
Saunders, A., Lim, J., 1990. Underpricing and the new issue process in Singapore. Journal of
Su, C., Bangassa, K., 2011. The impact of underwriter reputation on initial returns and long-
Vithessonthi, C., 2008. Stock price performance of initial public offerings: The Thai
Yong, O., 2007. A review of IPO research in Asia: What's next? Pacific-Basin Finance
Zagorchev, A., Vasconcellos, G., Bae, Y., 2011. Financial development, technology, growth
and performance: Evidence from the accession to the EU. Journal of International
42
Page 44 of 55
Table
Table 1
Initial returns for Thai IPOs between 2000 and 2012 categorized by year of issuance and by
industry.
Average Minimum Maximum %
Initial Initial Initial Positive
Returns Returns Returns Initial
Year N (%) Std. Dev. p-value (%) (%) Returns
Panel A: By year
2000 2 -1.44 15.64 0.917 -12.50 9.62 50.00
t
2001 7 36.78 61.48 0.165 -17.50 166.67 85.71
ip
2002 18 15.55 31.77 0.053 -13.48 106.67 50.00
2003 22 46.95 44.17 0.000 -36.40 151.75 86.36
2004 36 14.05 31.97 0.012 -23.20 98.75 63.89
cr
2005 38 10.09 23.15 0.011 -23.08 90.91 57.89
2006 16 1.44 11.07 0.610 -16.88 36.89 50.00
us
2007 7 22.50 54.94 0.320 -5.96 146.67 71.43
2008 12 1.16 11.96 0.743 -23.91 26.56 50.00
2009 11 0.90 12.31 0.814 -30.00 16.16 63.64
2010
2011
2012
7
3
8
17.63
8.80
69.38
21.81
16.77
74.02
an 0.076
0.459
0.033
-1.96
-3.67
-9.52
59.26
27.86
200.00
85.71
66.67
75.00
All issues 187 18.00 37.29 0.000 -36.40 200.00 64.17
M
Panel B: By industry
AGR 6 1.13 6.52 0.689 -7.10 12.11 66.67
CON 3 7.40 16.74 0.524 -4.38 26.56 66.67
d
initial return is the sample average initial return, computed as the arithmetic average of initial
returns on all IPOs in the sample of n firms. According to the classification of the SET, firms
are listed in one of the following industry groups: (1) agro & food (AGR), (2) consumer
products (CON), (3) financials (FIN), (4) industries (IND), (5) property & construction
(PRO), (6) resources (RES), (7) services (SER), and (8) technology (TEC).
1
Page 45 of 55
Table 2
Market-adjusted initial returns for Thai IPOs between 2000 and 2012 categorized by year of
issuance and by industry.
%
Average Minimum Maximum Positive
Market- Market- Market- Market-
Adjusted Adjusted Adjusted Adjuste
Initial Initial Initial d Initial
Returns Returns Returns Returns
Year N (%) Std. Dev. p-value (%) (%) (%)
t
Panel A: By year
ip
2000 2 -0.55 14.11 0.965 -10.53 9.43 50.00
2001 7 37.14 60.69 0.157 -18.56 164.63 85.71
cr
2002 18 15.52 31.60 0.053 -13.64 107.26 61.11
2003 22 46.86 43.81 0.000 -36.64 149.77 86.36
us
2004 36 14.16 32.18 0.012 -23.35 101.44 63.89
2005 38 9.90 23.06 0.012 -22.92 89.77 63.16
2006 16 1.52 11.00 0.587 -19.12 35.96 56.25
2007 7 22.94 54.59 0.309 -4.78 146.22 71.43
2008
2009
12
11
1.08
1.24
11.92
12.43
an 0.759
0.748
-22.28
-29.93
25.78
16.43
58.33
72.73
2010 7 17.65 264.58 0.077 -0.64 58.73 71.43
M
2011 3 8.53 15.89 0.450 -3.09 26.64 66.67
2012 8 69.28 74.24 0.033 -9.16 199.52 75.00
All issues 187 18.02 37.20 0.000 -36.64 199.52 67.38
d
Panel B: By industry
AGR 6 0.60 6.01 0.817 -6.96 10.83 50.00
te
This table presents average market-adjusted initial returns for Thai IPOs during the 2000–
2012 period by year of issuance (Panel A) and by the Stock Exchange of Thailand’s
industrial classification (Panel B). IPO prices, stock prices, SET indices are obtained from the
SETSMART database. Average market-adjusted initial return is the sample average market-
adjusted initial return, computed as the arithmetic average of market-adjusted initial returns
on all IPOs in the sample of n firms. According to the classification of the SET, firms are
listed in one of the following industry groups: (1) agro & food (AGR), (2) consumer products
(CON), (3) financials (FIN), (4) industries (IND), (5) property & construction (PRO), (6)
Resources (RES), (7) services (SER), and (8) technology (TEC).
2
Page 46 of 55
Table 3
Summary statistics of the main variables.
Variable Mean Median Minimum Maximum Std. Dev.
Initial return (%) 18.00 3.13 -36.40 200.00 37.29
Market-adjusted initial returns
(%) 18.02 3.46 -36.64 199.52 37.20
IPO price (in THB) 12.06 8.80 1.25 82.19 12.25
Gross proceeds (in millions
THB) 5,518 1,900 169 97,904 13,045
Real gross proceeds (in
t
millions THB) 5,999 2,034 165 110,004 14,235
ip
Stock market capitalization (in
millions THB) 4,676,427 4,654,038 1,303,353 11,376,814 1,942,668
Real stock market
cr
capitalization (in millions
THB) 5,006,415 5,049,173 1,464,442 11,142,815 1,874,650
us
Market P/E 10.68 9.40 3.37 27.88 4.17
Market P/BV 1.83 1.87 0.95 2.71 0.33
Market yield (%) 3.21 3.14 1.60 6.28 0.76
T-bills outstanding (in millions
THB)
Government bonds outstanding
(in millions THB)
145,053
1,429,599
an
144,040
1,280,536
0
568,316
300,000
2,948,762
61,571
510,138
Private debt securities
M
outstanding (in millions THB) 625,426 473,695 215,888 1,719,409 364,454
New T-bills (in millions THB) 42,790 42,135 0 100,000 22,445
New government bonds (in
millions THB) 28,251 18,000 0 291,599 36,321
d
3
Page 47 of 55
t
ip
cr
Table 4
The effects of stock market conditions on the level of market-adjusted initial returns for IPOs in Thailand between 2000 and 2012.
Variable Model 1 Model 2 Model 3 Model 4 Model 5
us
Coefficient S.E. Coefficient S.E. Coefficient S.E. Coefficient S.E. Coefficient S.E.
Intercept 4.208* 2.341 7.789*** 2.273 3.187 2.123 8.709*** 2.481 4.407 6.527
ISSUE -1.719** 0.722 -2.033*** 0.686 -2.033*** 0.667 -2.115*** 0.667 -1.269** 0.552
an
SMD 0.068 0.081 -0.160* 0.084 0.246*** 0.084 -0.197 0.173 -0.166 0.450
MKTPE 0.004 0.006 0.016*** 0.006 0.017 0.012
MKTPBV 0.520*** 0.096 0.482*** 0.171 1.192*** 0.296
MKTYIELD -0.199*** 0.044 -0.050 0.063 0.306*** 0.102
M
Industry dummies Yes Yes Yes Yes No
Year dummies No No No No Yes
R2 0.151 0.279 0.272 0.306 0.364
Adjusted R2
d
0.103 0.238 0.230 0.258 0.300
F-statistic 3.136*** 6.803*** 6.560*** 6.385*** 5.689***
e
N 187 187 187 187 187
This table presents the regression results of the one-day market-adjusted initial return for IPOs. The sample period is from January 2000 to
pt
December 2012. Issue size (ISSUE) is measured as the natural logarithm of IPO real gross proceeds. Stock market development (SMD) is
measured as the natural logarithm of the real market capitalization of the Stock Exchange of Thailand (in millions of the THB) on the IPO date.
MKTPE, MKTPBV and MKTYIELD are the market P/E ratio, market P/BV ratio, and market yield of the Stock Exchange of Thailand on the IPO
ce
date, respectively. Industry-dummy variables are included in Models 1–4, whereas year-dummy variables are included in Model 5. All OLS
regressions are estimated with the White’s heteroskedasticity consistent covariance matrix. The symbols ***, **, and * indicate that the
coefficient is significant at 1%, 5%, and 10% levels, respectively.
Ac
4
Page 48 of 55
t
ip
cr
Table 5
The effects of bond market development on the level of market-adjusted initial returns for IPOs in Thailand between 2000 and 2012.
Variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
us
Coefficient S.E. Coefficient S.E. Coefficient S.E. Coefficient S.E. Coefficient S.E. Coefficient S.E.
Intercept 4.958** 2.331 5.602** 2.414 3.517 2.349 6.167* 3.495 4.532* 2.458 -6.883 5.333
ISSUE -1.260* 0.668 -1.908*** 0.726 -1.867** 0.718 -1.433** 0.691 -1.379** 0.673 -0.841* 0.494
an
SMD 0.075 0.076 0.302*** 0.101 0.336*** 0.110 0.296*** 0.106 0.274** 0.107 1.039*** 0.296
TBMD -0.180** 0.073 -0.160** 0.068 -0.155** 0.067 -0.255 0.212
GBMD -0.311** 0.133 -0.246 0.271
PBMD -0.222*** 0.081 -0.064 0.178 -0.194** 0.085 -0.213 0.395
M
Industry
dummies Yes Yes Yes Yes Yes No
Year
dummies No No No No No Yes
R2
d
0.164 0.170 0.174 0.190 0.185 0.281
Adjusted R2 0.116 0.123 0.127 0.133 0.133 0.211
e
F-statistic 3.366*** 3.602*** 3.711*** 3.310*** 3.515 4.034***
N 182 pt 187 187 182 182 182
This table presents the regression results of the one-day market-adjusted initial return for IPOs. The sample period is from January 2000 to
December 2012. Issue size (ISSUE) is measured as the natural logarithm of the IPO real gross proceeds (in millions of The THB). Stock market
development (SMD) is measured as the natural logarithm of the real market capitalization of the Stock Exchange of Thailand (in millions of The
ce
THB) on the IPO date. TBMD, GBMD, and PBMD are measured as the natural logarithm of the amount of outstanding T-bills, outstanding
government bonds, and outstanding private debt securities at the end of the month prior to the IPO date, respectively. Industry-dummy variables
are included in Models 1–4, whereas year-dummy variables are included in Model 5. All OLS regressions are estimated with the White’s
heteroskedasticity consistent covariance matrix. The symbols ***, **, and * indicate that the coefficient is significant at 1%, 5%, and 10%
Ac
levels, respectively.
5
Page 49 of 55
t
ip
cr
Table 6
The effects of bond market development on the level of market-adjusted initial returns for IPOs in Thailand between 2000 and 2012.
Variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
us
Coefficient S.E. Coefficient S.E. Coefficient S.E. Coefficient S.E. Coefficient S.E. Coefficient S.E.
Intercept 4.497* 2.389 3.487* 1.958 3.935* 2.301 4.032* 2.050 4.561* 2.416 3.600* 1.962
ISSUE -1.356* 0.715 -1.677** 0.674 -1.721** 0.724 -1.330* 0.676 -1.364* 0.718 -1.686** 0.675
an
SMD 0.043 0.071 0.142* 0.073 0.084 0.073 0.070 0.065 0.038 0.069 0.131* 0.070
NTBILL -0.087** 0.044 -0.058 0.046 -0.088** 0.044
NGBOND -0.052** 0.026 -0.042 0.028 -0.052** 0.026
NPBOND 0.007 0.012 0.004 0.012 0.004 0.011 0.007 0.012
M
Industry
dummies Yes Yes Yes Yes Yes Yes
Year
d
dummies No No No No No No
R2 0.129 0.190 0.150 0.159 0.129 0.191
e
Adjusted R2 0.077 0.144 0.102 0.098 0.072 0.141
F-statistic 2.484*** 4.139*** 3.117*** 2.607*** 2.252** 3.764***
N 179
pt 187 187 179 179 187
This table presents the regression results of the one-day initial return for IPOs. The sample period is from January 2000 to December 2012. Issue
size (ISSUE) is measured as the natural logarithm of the IPO real gross proceeds (in millions of the THB). Stock market development (SMD) is
ce
measured as the natural logarithm of the real market capitalization of the Stock Exchange of Thailand (in millions of the THB) on the IPO date.
NTBILL, NGBOND, and NPBOND are the natural logarithm of the amount of total new T-bills issuance, total new government bonds issuance,
and total new private debt securities issuance (in millions of the THB) for the month prior to the IPO date, respectively. Industry-dummy
variables are included in all models. All OLS regressions are estimated with the White’s heteroskedasticity consistent covariance matrix. The
Ac
symbols ***, **, and * indicate that the coefficient is significant at 1%, 5%, and 10% levels, respectively.
6
Page 50 of 55
t
ip
cr
Table 7
The effects of interest rates and on the level of market-adjusted initial returns for IPOs in Thailand between 2000 and 2012.
Variable Model 1 Model 2 Model 3 Model 4 Model 5
us
INTDIF < 0 INTDIF ≥ 1
Coefficient S.E. Coefficient S.E. Coefficient S.E. Coefficient S.E. Coefficient S.E.
Intercept 3.607 2.381 3.732 2.401 3.748 2.417 12.438 10.400 3.257 2.427
an
ISSUE -1.747** 0.719 -1.700** 0.701 -1.724** 0.706 -1.924 1.468 -1.229 0.756
SMD 0.141 0.092 0.155 0.095 0.150 0.095 -0.591 0.926 0.139* 0.083
INT -6.275** 3.145 -5.556* 3.274 -5.874* 3.187 -15.238 17.341 -6.661** 3.010
TS 0.290 4.523 0.395 4.544 0.222 4.574 -2.875 19.199 4.250 5.554
M
REER -0.001 0.005 -0.007 0.007 -0.005 0.007 0.035 0.044 -0.014** 0.007
INTDIF 0.739 3.354 -3.314 4.366 -2.684 4.402 22.846 27.562 17.578** 7.613
INTDIF*ID1 8.901 8.172
d
INTDIF*ID2 7.311 8.704
Industry dummies Yes Yes Yes Yes Yes
e
Year dummies No No No No No
R2 0.186 0.192 0.190 0.288 0.240
Adjusted R2 0.125
pt 0.127 0.124 -0.067 0.166
F-statistic 3.051*** 2.929*** 2.879*** 0.811 3.238***
N 187 187 187 40 147
ce
This table presents the regression results of the one-day market-adjusted initial return for IPOs on issue size (ISSUE), stock market development
(SMD), domestic interest rate (INT), term spread (TS), interest-rate differential (INTDIF), real effective exchange rate (REER), industry-dummy
variables, and year-dummy variables. The sample period is from January 2000 to December 2012. ISSUE is measured as the natural logarithm
Ac
of the IPO real gross proceeds (in millions of the THB). SMD is measured as the natural logarithm of the real market capitalization of the Stock
Exchange of Thailand (in millions of the THB) on the IPO date. INT denotes the yield on the one-year government bond at the end of the month
prior to the IPO date. TS is measured as the difference between a yield on the 10-year government bond and a yield on the one-year government
bond in Thailand. INTDIF is measured as the difference between the one-year Thai government bond yield and the US federal funds rate at the
end of the month preceding the IPO date. ID1 is a dummy variable, which takes a value of one if INTDIF is larger than the median value, and
zero otherwise. ID2 is a dummy variable, which takes a value of one if INTDIF is positive, and zero otherwise. REER is the real effective
exchange rate of the THB at the end of the month prior to the IPO date. Industry-dummy variables are included in all models. All OLS
regressions are estimated with the White’s heteroskedasticity consistent covariance matrix. The symbols ***, **, and * indicate that the
coefficient is significant at 1%, 5%, and 10% levels, respectively.
7
Page 51 of 55
t
ip
cr
Table 8
The effects of stock and bond market development on the level of market-adjusted initial returns for IPOs in Thailand between 2000 and 2012
for the hot- and cold-IPO subsamples.
us
Variable Model 1 Model 2 Model 3 Model 4 Model 5
Full sample Hot IPOs Cold IPOs Hot IPOs Cold IPOs
Coefficient S.E. Coefficient S.E. Coefficient S.E. Coefficient S.E. Coefficient S.E.
an
Intercept 12.488*** 3.567 13.728*** 5.113 -0.078 0.976 14.474 9.676 -4.378* 2.510
ISSUE -1.539** 0.596 -2.103* 1.144 -0.052 0.172 -1.118 0.984 0.065 0.143
SMD -1.091*** 0.416 -1.228** 0.577 -0.104 0.139 -0.637 1.011 0.002 0.155
MKTPBV 1.051*** 0.286 1.049*** 0.348 0.097 0.093 0.928*** 0.326 0.104 0.095
M
TBMD -0.010 0.053 -0.072 0.107 0.035 0.025 -0.057 0.240 0.111 0.075
PBMD 0.479* 0.284 0.630 0.430 0.065 0.100 -0.217 1.093 0.228 0.150
INT 1.074 2.568 4.857*** 4.922 0.121 0.952 17.417 10.880 -3.726* 2.143
d
INTDIF 1.391 3.490 -24.273* 8.747 -0.917 1.004 -64.409*** 21.975 2.860* 1.457
REER 0.010 0.007 0.025 0.013 0.004 0.003 0.006 0.044 0.003 0.007
e
Industry dummies Yes Yes Yes No No
Year dummies No No No Yes Yes
R2
pt
0.331 0.378 0.198 0.417 0.281
Adjusted R2 0.271 0.214 0.068 0.223 0.118
F-statistic 5.480*** 2.306** 1.526 2.146** 1.723**
ce
N 182 73 109 73 109
This table presents the results of the one-day market-adjusted initial return for IPOs on issue size (ISSUE), stock market development (SMD),
market P/BV (MKTPBV), T-bills market development (TBMD), private bond market development (PBMD), domestic interest rate (INT),
Ac
interest-rate differential (INTDIF), real effective exchange rate (REEF), industry-dummy, and year-dummy variables. The sample period is from
January 2000 to December 2012. ISSUE is measured as the natural logarithm of the IPO real gross proceeds (in millions of the THB). SMD is
measured as the natural logarithm of the real market capitalization of the Stock Exchange of Thailand (in millions of the THB) on the IPO date.
MKTPBV is the market P/BV ratio of the Stock Exchange of Thailand on the IPO date. TBMD and PBMD are measured as the natural logarithm
of the amount of outstanding T-bills and outstanding private debt securities at the end of the month prior to the IPO date, respectively. INT
denotes the yield on the one-year government bond at the end of the month prior to the IPO date. INTDIF is measured as the difference between
the one-year Thai government bond yield and the US federal funds rate at the end of the month preceding the IPO date. REER denotes the real
effective exchange rate of the THB at the end of the month prior to the IPO date. Industry-dummy variables are included in Models 1–3, whereas
8
Page 52 of 55
t
ip
cr
year-dummy variables are included in Models 4–5. All OLS regressions are estimated with the White’s heteroskedasticity consistent covariance
matrix. The symbols ***, **, and * indicate that the coefficient is significant at 1%, 5%, and 10% levels, respectively.
us
an
M
e d
pt
ce
Ac
9
Page 53 of 55
Figure
Figure 1
Development of initial public offerings listed on the Stock Exchange of Thailand from 1975
to 2012.
t
ip
cr
us
an
M
d
te
This figure illustrates the number of initial public offerings listed on the Stock Exchange of
Thailand by year of issuance during 1975–2012. Source: Adapted from the SETSMART
p
database.
ce
Ac
1
Page 54 of 55
Figure 2
Interest-rate differentials and market-adjusted initial returns.
t
ip
cr
us
an
M
d
This figure shows a scatter plot of interest-rate differentials (on the x-axis) and one-day
te
market-adjusted initial returns of the IPOs (on the y-axis). The interest-rate differential is
measured as the difference between the one-year Thai government bond yield and the US
federal funds rate.
p
ce
Ac
2
Page 55 of 55