Corporate Governance in Ipo Underpricing by hlp80151

VIEWS: 100 PAGES: 33

More Info
									      CORPORATE GOVERNANCE, DISCLOSURE QUALITY AND THE
                   UNDERPRICING OF IPOS




                                                       by


                                                Jean Bédard
                                         Professor, Université Laval

                                            Daniel Coulombe**
                                         Professor, Université Laval

                                             Lucie Courteau
                          Associate Professor, Free University of Bozen/Bolzano


                                                January 2007

        **Corresponding author
        Université Laval
        Département des sciences comptables
        Cité Universitaire, Québec
        Canada, G1K 7P4




Acknowledgements:


Keywords: Corporate governance, Initial public offerings; Underpricing; Audit committee

We would like to thank participants at the European Accounting Association Conference in Dublin in 2006, and at
the research workshop at l’Université de Lausanne for their comments. Special thanks are extended to Larry Weiss
for his helpful comments and suggestions on an earlier version of the paper. The authors acknowledge financial
support from the Certified General Accountants of Canada Research Foundation.
     CORPORATE GOVERNANCE, DISCLOSURE QUALITY AND THE
                  UNDERPRICING OF IPOS


                                           ABSTRACT

This paper presents a study of corporate governance and voluntary disclosure in IPO
prospectuses. We propose a three-part analysis. We first examine the possible determinants for
the creation of a separate audit committee in the context of an IPO. Second, we measure the
impact of governance attributes on the quality of disclosure, as measured by the quality of
management forecasts in the prospectus. Third, we study the impact of corporate governance and
audit committee characteristics on the cost of capital as measured by the underpricing at the time
of the IPO.
        The empirical analysis is performed on a sample of 246 IPO issued under Québec
securities regulation, over the period 1982 to 2002. This setting is ideal to study the relationship
between governance, disclosure and cost of capital since audit committees were not compulsory
over the sample period and since Canadian securities regulations allow the inclusion of earnings
forecasts in the IPO prospectus, which provides us with a proxy for disclosure quality.
       We find that the creation of an audit committee is positively related to the quality of the
board of directors and that the quality of the audit committee decreases significantly the level of
underpricing of the IPO. Contrary to previous studies in other settings, however, we find that the
governance attributes of Québec IPO firms have no significant effect on disclosure quality.




                                                                                                  2
1-INTRODUCTION

Accounting scandals such as Enron and WorldCom have triggered first an awareness of the
effects of weak corporate governance and then an increase in regulation about governance
mechanisms in the US and around the world. Some questions remain about the effectiveness of
these regulations (Romano 2005). We contribute to the debate by examining a setting where
governance best practices are known and available but not mandatory.

   In particular, we examine the choice of governance practices and their effect on disclosure and
cost of capital at the time of the IPO in firms who make a first issue in the Province of Québec,
which has a regulatory setting slightly different from that of the US and the rest of Canada, yet is
operating in the North American economic environment.

   IPOs are characterized by a large information asymmetry between the entrepreneur who has
private knowledge about his firm’s expected future cash flows and investors with whom he wants
to share the firm’s ownership and risk. Investors’ knowledge about IPO companies being limited,
they must place substantial reliance on the prospectus prepared by the entrepreneur. Prior studies
have shown, both analytically and empirically, that entrepreneurs use signals such as hiring
underwriter and/or auditors of prestige (Beatty and Ritter 1986 and Feltham et al. 1991 in the US;
Clarkson and Simunic 1994 in the Canadian context; Bédard et al. 2000 in the Québec setting),
the retention of a significant percentage of firm ownership (Leland and Pyle 1977; Datar et al.
1991; Courteau 1995), underpricing (Grinblatt and Hwang 1989; Welch 1989, Carter et al. 1998),
and the issue of earnings forecast (Lee et al. 2003) to communicate their private information to
investors. The entrepreneur has the choice between these different signals, or any combination of
them, and he reaches an equilibrium between the costs and benefits of the various signals.

   The success of the signalling strategy chosen by the entrepreneur depends very much on the
credibility of the information communicated, as perceived by investors. Corporate governance is
a set of mechanisms that ensures, among other things, the quality of the information (financial
and non financial) that is communicated to the markets through the prospectus in the specific case
of IPOs. The auditing of the information included in the prospectus by an independent auditor
represents one of these mechanisms, but the board of directors and the audit committee are also
important components of the system that ensures that the information communicated before the


                                                                                                  3
issue is credible and that the firm’s managers will continue to act in the best interest of all
shareholders and provide quality information even after the IPO. If investors are confident that
the governance mechanisms in place provides this assurance, they are likely to require a lower
cost of capital from the firm and this is evidenced by a lower level of underpricing of the issue.

   The capital market consisting of securities issued by Québec firms has some characteristics
that make it particularly useful to examine the relationship between governance, quality of
disclosure and underpricing of IPOs. Due to the lower level of the risk of legal costs, Québec
issuers, like other Canadian firms but unlike their US counterparts, can include earnings forecasts
in their prospectus. Not only does the inclusion of forecasts constitute an additional signal at the
disposal of the entrepreneur but the quality of these forecasts provides the researcher with a
measure of the quality of disclosure. Moreover, unlike other Canadian issuers, firms incorporated
under the Québec Incorporation Act can choose whether or not to establish an audit committee.
Since the proportion of firms that choose not to have such a committee is significant (39% in our
sample), something can be learned from this setting about the effect of audit committees on
disclosure quality and cost of capital.

   In this small North-American market, we first examine the decision to establish an audit
committee and then the effect of this decision and of other governance practices on disclosure
quality and cost of capital. We consider various board characteristics that have been examined in
prior studies (Finkle 1998, Certo et al. 2001, Filatotchev and Bishop 2002, among others), but we
add two audit committee characteristics: the financial expertise and the independence of its
members. We use earnings forecast errors as a measure of disclosure quality and the level of
underpricing of the issue as a measure of cost of capital.

   We find that the creation of an audit committee is positively related to the quality of the board
of directors and that the quality of the audit committee decreases significantly the level of
underpricing of the IPO. Contrary to previous studies in other settings, however, we find that the
governance attributes of Québec IPO firms have not significant effect on disclosure quality.

   The remainder of the paper is organized as follows. The next section elaborates on our
research questions, putting them into the context of the existing literature. The third section
presents the research design used to answer these questions, the sample selection procedures and




                                                                                                     4
the measurement of variables. Our empirical results are presented in the fourth section and a
conclusion follows.

2-RESEARCH QUESTIONS


Regulatory environment
  In Canada, the two levels of government (provincial and federal) make the regulation of IPOs
somewhat complicated. Each of the 10 provinces has a separate Company Law and there is also a
Canadian Company Law at the federal level. Canadian companies can choose to incorporate
under the Law of their province or under the federal Law. Moreover, each province has its own
Securities Commission that regulates public companies and the Canadian stock exchanges.
Finally, whereas the legal systems in nine of the provinces and at the federal level are based on
Common Law, Québec’s system is based on a Civil Code.

  Until 1999, companies that chose incorporation under the Québec law were mainly listed on
the Montreal Stock Exchange (MSE, hereafter) when they became public. The MSE was
regulated by the Québec Securities Commission. Following a major reorganization of Canadian
Stock Exchanges in 1999, all Canadian firms are now listed on the Toronto Stock Exchange
(TSX, hereafter) or the Venture Exchange, whether they are incorporated under the federal law or
that of any of the provinces. While private companies are not required to have an audit committee
even if they are preparing an IPO, requirements regarding public companies vary across
provinces and time.

  In Canada, the requirements regarding the audit committee depends on the incorporation Act
and the exchange or Securities Commissions regulations. The Canadian Securities Commissions
made independent audit committees mandatory for public companies only in 2004 (multilateral
instrument 52-110 (2.1)). Before 2004, once they became public, Québec corporations had to
create an audit committee with a majority of non-executive members only if they were
incorporated under the under the Canadian Business Corporation Act (CBCA section 171(1)) but
were not required to even have such a committee if they were incorporated under the Quebec
Incorporation Act. Between 1995 and 2004 the TSX and MSE Corporate Governance Guidelines
required every listed company incorporated in Canada or a province of Canada to disclose any
divergence from the stated guidelines, including guidelines #13 which state that firms must have


                                                                                               5
an audit committee composed only of outside directors (TSX Company Manual, Sec. 473 and
474). Compliance with these guidelines was not mandatory, as long as the non-compliance was
disclosed. Since 2004 all Canadian public companies must have an audit committee composed of
a minimum of three independent and financially literate members, but the regulations still provide
an exemption for the members’ independence requirement for IPO firms. For these firms, the
requirement does not apply for a period of up to one year if the majority of the audit committee
members are independent.

    In summary, Québec corporations which are incorporated under the Québec law have the
choice to have an audit committee or not, while those incorporated under the Canadian law must
have such a committee once they are listed (but not necessarily before the IPO). This set of
regulations makes Québec IPOs an excellent terrain for examining the choice of whether to create
an audit committee at the time of the IPO, because it is not mandatory.1

    On the topic of disclosure, Québec corporations, like those of the rest of Canada, are allowed
(but not required) to include earnings forecasts in their IPO prospectus. Before 1989, these
forecasts were reviewed by the external auditors but a 1989 recommendation of the CICA
requires that all forward-oriented financial information included in the prospectus be audited. The
choice to include forecasts and the requirement that those issued after 1988 be audited apply to all
Canadian issuers, whatever the jurisdiction under which they are incorporated or listed.

    In this setting where the creation of an audit committee and the issue of earnings forecasts are
voluntary, we examine these choices in the context of the entrepreneur’s communication strategy
at the time of the IPO. These choices are in interaction with many other signals that have been
found to be relevant in prior studies, and we expect them to affect the outcome of the IPO process
and the level of underpricing, which we use as a measure of the firm’s cost of equity capital. In
particular, we seek to answer the following questions: (1) why IPO firms elect to create an audit
committee before going public, (2) what is the effect of governance quality on the quality of




1
  All publicly traded firms in Quebec and Canada must have a Board of directors. For the period under study there
are no specific requirements as to the size of the board or the independence of directors except for firms incorporated
under the Canadian Business Corporation Act where the board must be composed of a minimum of three members
and at least two non-executives. The information about the board of directors of IPO firms must be included in the
prospectus.

                                                                                                                     6
disclosure, and (3) what is the effect of the characteristics of the board of directors and the audit
committee on the equilibrium pricing of securities in the context of an IPO.

Why create an audit committee?
Our first research question is related to the decision to create a separate audit committee (AC,
hereafter). Menon and Williams (1994) note that ACs are a useful component of the governance
structure that monitors managers and ensure that they act in the interest of the firm’s
shareholders. The committee can enhance the monitoring function of the board of directors by
assuming the board’s responsibility in the areas related to financial reporting and auditing. Its
independence insulates internal and external auditors from the influence of managers which may
be quite strong on the board. Moreover, the AC may increase the efficiency of the board in
matters of control, especially when the number of directors is large. Beasley and Salterio (2001)
also point to the role of the AC in the communication of financial information to the board as the
committee members meet internal and external auditors to discuss the main issues related to the
firm’s financial reporting.

   Several studies have identified firm characteristics that are associated with the voluntary
creation of ACs. Pincus et al. (1989) and Menon and Williams (1994), for example, examine US
firms which were not subject to stock exchange regulations requiring them to form audit
committees. Since the 1990s, ACs have become mandatory for listed firms in most markets and
more recent studies have examined factors affecting AC characteristics and structure rather than
its mere existence. For example, Beasley and Salterio (2001) examine the choice by Canadian
firms to give their AC more than the minimum characteristics required by regulation in terms of
size, independence and financial competence, while Klein (2002) concentrates on AC
independence and studies the economic factors that are associated with it. The advantage of the
study presented here is that it examines a sample of firms for which the creation of an AC was not
compulsory.

   At the time of the IPO, the presence of an AC in the firm’s governance structure may be
considered by investors as a signal about firm quality. According to signalling theory, a signal
allows investors to distinguish between firms of various quality if its costs outweigh its benefits
for low-quality firms while benefits exceed costs for high-quality firms (Ackerlof 1970; Leland
and Pyle 1977). Indeed, the creation of an AC is costly for the entrepreneur, both in terms of


                                                                                                   7
direct disbursements and by the fact that a well-functioning committee constitutes a constraint on
the firm managers’ ability to manage earnings in the future. The costs related to this constraint are
higher for low-quality issuers who may need earnings management to present a performance
which is deemed satisfactory by investors. The benefit from the signal is the reduction of the cost
of capital at the time of the IPO, i.e. the level of underpricing of the issue, which is similar for
both types of firms if the signalling strategy is successful.

   The preparation of an IPO is one occasion when a firm must review its governance structure
and decide on changes that need to be done before confronting itself with the market’s
judgement. The firms that went public in Québec before 2004 and that were incorporated under
the Québec Incorporation Act were under no constraint as to the choice to establish an audit
committee. This characteristic allows us to examine the factors which affect the voluntary
adoption of such a control mechanism in a legal environment that has been described as non-
threatening.

   Based on previous studies we identify three sets of factors that may affect the decision to
establish an audit committee: the characteristics of the firm’s corporate governance structure, its
level of risk and the alternative signals it may be using to communicate firm-specific information
to investors at the time of the IPO.

   When studying the choice to create an AC, Menon and Williams (1994) and Pincus et al.
(1989) find that board independence increases the likelihood of ACs in firms that are under no
obligation to have them. In her study of AC independence, Klein (2002) finds that board
independence and board size are positive determinants whereas Beasley and Salterio (2001) find
that the likelihood of Canadian firms choosing a level of AC independence which is higher than
the strict minimum required by regulation increases in board size and independence and decreases
when the CEO chairs the board of directors. We expect these results to extend to the decision to
set up an AC at the time of the IPO. In particular, we expect larger boards and boards with a
larger percentage of outside members and not chaired by the CEO to be associated with the
presence of an audit committee at the time of the IPO.

   Firms that exhibit higher levels of risk may benefit most from the creation of an AC since
investors are likely to require a higher cost of capital from them. While most studies on AC




                                                                                                   8
creation and independence control for leverage, only Pincus et al. (1989) find a significantly
positive effect.

   At the time of the IPO, firms have to make a certain number of decisions related to control and
governance. In addition to the creation of an AC, they must, among other things, decide what
auditor to hire for the prospectus and what underwriter will promote the issue. The principal
shareholders must also decide how much of their shares to include in the public offering. As these
elements are also part of the governance structure of the firm, they may have an influence on the
creation of an audit committee. Indeed, Certo et al. (2001) find that the creation of an audit
committee represents a signal that the entrepreneur can use to communicate information about
firm quality, in addition to auditor and underwriter choice and to retained ownership by the firm’s
principals. It is possible that firms can compensate for the choice of a lower-quality external
auditor or underwriter by creating a competent audit committee to supervise the audit work and
the quality of financial disclosure. On the other hand, high-quality auditors and underwriters may
encourage their clients to strengthen their governance structure by creating an AC, in which case
the AC, the auditor and the underwriter would be complements rather than substitutes. In contexts
other than that of the IPO, both Menon and Williams (1994) and Pincus et al. (1989) have found a
positive association between the quality of the auditor and the decision to create an AC. Given the
possibility of counterbalancing effects, we make no prediction as to the direction of the effect of
auditor and underwriter quality on the likelihood of the creation of an AC in preparation for the
IPO.

Impact of good governance on the quality of financial information
   The quality of the governance has been found to have a positive effect on the quantity and
quality of financial information disclosure. Karamanou and Vafeas (2005) examine the relation
between governance quality and the issue or update of management earnings forecast and find
that the percentage of outsiders on the board and the AC increase the likelihood that a firm’s
managers make an earnings forecast or update a forecast they have previously made. Similarly, in
a sample of Hong Kong firms, Gul and Leung (2004) find that CEO duality is negatively related
to the level of voluntary disclosure, as measured by a disclosure score. Hence, it seems that the
quality of a firm’s governance structure has a positive effect on the level of its voluntary
disclosure.


                                                                                                 9
   Klein (2002) and Bédard et al. (2004) examine the relationship between board and AC
characteristics and the quality of financial disclosure and both find a negative association between
the independence and financial expertise of ACs and the level of discretionary accruals.
Examining the disclosure and updating of management earnings forecasts, Karamanou and
Vafeas (2005) find that the effectiveness of boards and audit committees is related with the
accuracy of the forecasts. Using the informativeness of financial disclosure as a measure of
disclosure quality, Beekes and Brown (2006) find that governance quality, as measured by
corporate governance rankings for Australian firms, is related to the level of informativeness of
financial disclosure.

   Our second research question is concerned with the relationship between governance quality
and the quality of financial disclosure. Information quality is difficult to measure in general, but
the Canadian setting of our study provides us with a measure that is not available for IPOs in the
US and in several other countries. Indeed, Canadian disclosure regulations allow the inclusion of
management earnings forecasts in IPO prospectuses. We expect that Karamanou and Vafeas’s
(2005) result on the association between governance structure and accuracy of management
earnings forecast extend to forecasts made at the time of the IPO.

   As close to 48% of our sample firms issue a management earnings forecast in their prospectus,
we use the forecast error as a measure of the quality of the information disclosed and test the
following hypothesis.

H1: The effectiveness of the board of directors and the audit committee is negatively
    associated with the error in the earnings forecasts included in the IPO prospectus.

Impact of good governance on underpricing

   Our third research question concerns the effect of governance effectiveness on the cost of
capital required by investors at the time of the IPO. Prior studies have shown that entrepreneurs
attempt to communicate to investors their private information about the quality of their firm by
choosing from a set of signals such as ownership retention (Leland and Pyle 1977), auditor and
underwriter reputation (Datar et al. 1991; Hughes 1986) and underpricing (Grinblatt and Hwang
1989; Welch 1989). Each of these signals generates a cost to the issuer that is high enough to
discourage low-quality firms from incurring them and mimicking the behaviour of higher-quality


                                                                                                 10
firms. In preparation for the issue, the entrepreneur chooses a combination of signalling devices
that maximise the expected issue proceeds, net of issuing and signalling costs.

   Certo et al. (2001) examine the role played by the quality of governance structure as a signal
to communicate information about firm quality. They find a significant negative relationship
between underpricing and board size and reputation in the context of IPOs, which leads them to
conclude that effective governance is part of the communication strategy in firms preparing to go
public. On the other hand, they find that the proportion of outsiders on the board seems to
increase, rather decrease, the level of underpricing. They explain this counterintuitive result by
the fact that investors prefer to see on the board of small start-up firms a large proportion of
directors who are knowledgeable about the firm’s operations and growth opportunities and they
are less concerned about control.

   Canadian IPOs have proven to constitute a well-suited environment in which to test signalling
theories because of the low level of litigation risk, as compared with the US setting (Clarkson and
Simunic 1994).

   In the relatively mild legal environment of Québec IPOs, we are interested to see whether
board and audit committee characteristics are related to IPO underpricing. Underpricing and the
establishment of an effective governance structure can be considered as two signalling devices
that can be used as part of the IPO firm’s signalling strategy. Moreover, governance effectiveness
is expected to increase investor confidence about the quality of current and future financial
information. Hence we test the following hypothesis.

H2: The effectiveness of a firm’s board of directors and audit committee is negatively
    associated with the level of underpricing of its IPO.

3-METHODOLOGY AND SAMPLE FORMATION


Audit committee decision
To answer our first research question, we examine the relationship between the decision to create
an audit committee and the factors that have been shown in previous studies to affect this
decision. We first consider factors related to governance quality, then firm-specific factors related




                                                                                                  11
to the riskiness of the firm and finally various control variables that are likely to affect the
decision to create an AC in preparation for an IPO. We use the following logistic model:2

    AC i = α + β1 Bsize i + β 2 Bind i + β 3 Bdual i + β 4 Age i + β 5 Growth i + β 6 Lev i + β 7 Subordinate i
                                                                                                                  (1)
              + β 8 Dey i + β 9 LnAssets i + β10 % Re t i + β11 Aud - h i + β12 UWi + ε i

where ACi is a dichotomous variable which takes value 1 if firm i has an AC and 0 otherwise.

Board characteristics

Quality boards are more likely to create audit committees to increase the overall governance
(Beasley and Salterio 2001). The percentage of outside members on the board of directors (Bind),
the number of directors (Bsize) and the separation of the roles of CEO and chairperson (1–Bdual)
have been shown to increase the likelihood of the creation or the quality of ACs (Menon and
Williams 1994; Klein 2002).3

Riskiness

Riskier firms have more to gain by establishing a stronger governance structure to reassure
prospective investors. The latter are likely to require lower expected returns from firms with an
AC whose main task is to overview the production and disclosure of quality financial
information. Because it is difficult to estimate the riskiness of a firm that is presenting financial
statements for the first time, we use four variables to measure it. Firms which have been in
operations for many years (Age) are likely to have already passed the most risky stages of their
development, so we expect older firms to be less likely to need an AC to convince investors of
the quality of the information they disclose. On the other hand, older firms are more likely to be
familiar with good governance practices and have had more time to improve their governance
structure, so they are more likely to have created an AC before the IPO. Given these two
opposing forces, we make no prediction about the direction of the net effect of age.




2
  The full description of all variables is provided in Table 1.
3
  Following Gompers et al. (2003) and Defond et al (2005), we also replace these three factors with a board quality
index that builds on the following 3 dimensions: board size larger that the median gets a 1, board with more than
50% outside members gets a 1, and board with separate chair and CEO functions gets a 1. Hence, each firm gets an
index value between 0 and 3, the higher the index, the higher the board quality

                                                                                                                  12
   Growth firms are more risky than those that have reached a more mature stage in their
development. We measure Growth as the average sales growth over the 3 years preceding the
issue and expect it to be positively related to the decision to create an AC. Our third measure of
riskiness, Lev, is the financial leverage (total debt/total assets) at the last balance sheet date before
the IPO. Finally, we take into account that subordinate-voting shares (Subordinate=1) are more
risky for their holders than common shares with the normal 1:1 relationship between cash flow
rights and voting rights.

Other control variables

In 1994, the TSE published a report entitled “Where were the directors?”, often referred to as “the
Dey Report”, by the name of its author. The report contains 14 recommendations for good
governance that were later adopted by the Exchange. The report had a major educational impact
in Canada with respect to governance practices. We add a control indicator variable (Dey) to
capture the effect of the Dey Report on the governance choices in the IPO issues that occurred
after its publication, as we believe that after this date firms were more likely to create a separate
audit committee prior to their IPO.

   Firm size is likely to affect the decision to create an AC before the IPO. Larger firms are
usually older, more developed and have more stakeholders who are pushing for better
governance. Eichenseher and Shields (1985) and Pincus et al. (1989) find that larger firms are
more likely to have an AC and Beasley and Salterio (2001) find a positive relation between firm
size and the likelihood that the AC quality be higher than the strict minimum required. We expect
large firms to be more likely to have established an AC before the IPO. LnAssets is the log of
total assets expressed in constant dollars.

   We also expect that a higher level of retained ownership (%Ret) after the IPO be associated
with the preponderance of the entrepreneur in his firm, and that such firms are less likely to form
an AC. Finally, we expect that firms who hire high-quality auditors (Aud-h) and underwriter
(UW) are also more likely to have high-quality governance, including the creation of an AC. We
define high-quality auditors (Aud-h=1) as Big eight/Big six audit firms and high-quality
underwriters (UW=1) as those with the largest number of employees in Canada.




                                                                                                      13
Information quality
To examine the relationship between disclosure quality and governance, we take advantage of the
fact that Canadian IPO firms have the choice to include an earnings forecast in their prospectus
and, as Karamanou and Vafeas (2005), we use the forecast error as proxy for information quality.
We test our hypothesis H1 with the following cross-sectional regression model:

                 LnFerri = α + β1 Bsize i + β 2 Bind i + β 3 Bdual i + β 4 AC i + β 5 Fhypi
                           + β 6 Fhori + β 7 Faudit i + β 8 Age i + β 9 Growth i + β10 RiskFact i   (2)
                           + β11 Aud - h i + β12 LnAssets i + ε i

where LnFerri is the log of the percentage forecast error (the absolute value of (Actual –
Forecast)/Forecast). In addition to the governance characteristics which constitute our test
variables, the model includes characteristics of the forecast itself that have been shown to affect
forecast errors in other settings as well as other control variables such as riskiness, auditor choice
and firm size.

Governance attributes

As in Karamanou and Vafeas (2005), best practices board attributes (Bsize, Bind, Bdual=0) as
well as the existence of an AC prior to the IPO (AC=1) are expected to improve forecast quality
(i.e. lower LnFerr).

Forecast attributes

We also control for non-governance factors that have been found to be associated with forecast
error. In the context of Canadian IPOs, Clarkson (2000) finds that forecast errors are related
positively to the forecast horizon (Fhor) and negatively to the number of hypotheses mentioned
in the prospectus as basis for the forecast (Fhyp). In addition, the fact that the forecast is audited
(Faudit=1) as opposed to having only review-level assurance (Faudit=0) has been found to
reduce the error in forecasts made at the time of the IPO (Clarkson 2000; McConomy 1998; and
Davidson and Neu 1993).

Other control variables

Firms with more operating experience (Age) are expected to have lower forecast errors (Clarkson
2000) while growth firms (Growth), because their future earnings are more difficult to forecast,


                                                                                                     14
are likely to have higher errors. Similarly, larger firms have been found to have lower forecast
errors (Clarkson 2000). Finally, given the association of auditors with earnings forecasts we
control for the quality of auditors (Aud-h).

Cost of capital
In order to test the effect of governance quality on the cost of capital, we examine the relationship
between the level of underpricing of the issue, as a measure of cost of capital, and the board and
AC attributes. In our test of H2, we must control for other variables which have been found to
affect the underpricing of IPOs. In the very extensive literature on IPOs, a large number of factors
have been found important in the determination of the pricing of the issue. While it is not possible
to include them all in our model, we consider two types of variables to control for these effects:
Alternative signals that the entrepreneur can choose to communicate his private information to
prospective investors and attributes that are specific to the firm and to the issue itself. Our choice
of control variables is similar to that of Certo et al. (2001) in a similar context. We test the
hypothesis with the following cross-sectional regression model.

                   Undpri = α + β1 Bqual i + β 2 AC i + β 3 ACqual i + β 4 % Re t i + β 5 UWi
                           + β 6 Aud - h i + β 7 Age i + β 8 Lev i + β 9 RiskFact i + β10 Unit i                   (3)

                            + β11 IPOprice i + β12 LnAssets i + β13 TaxDed i + ∑ γ k Yearik + ε i
                                                                                      k


where Undpri is the initial return from investing in the new issue of firm i and is measured as the
difference between the closing price of the first trading day and the issue price, as a proportion of
the closing price.

Governance attributes

The underpricing of IPOs is widely documented in the current accounting and finance literature
and a number of factors have been shown to affect the pricing of IPOs.5 If the choice of
governance structure is part of the entrepreneur’s information strategy in preparation for the IPO,
we expect the variables over which the choices are made to affect the level of underpricing of the
issue in a manner similar to that of other signalling variables.


5
    See Welch and Ritter (2002) for a recent review of the literature on IPO activity, pricing, and allocations.



                                                                                                                   15
    Thus we expect lower underpricing for firms with large boards (Bsize) made up of
independent directors (Bind) and a separate audit committee (AC=1) with a majority of
independent members (ACind=1) and with at least one financial expert (ACcomp=1), and higher
underpricing for firms where the CEO chairs the board of directors (Bdual=1). In alternative
models, we regroup these attributes into two quality indices (Gompers 2003): Bqual which
measures board quality and ACqual which represents AC quality.

Alternative signals

In 1977 Leland and Pyle identified retained ownership as a signal entrepreneurs can use to
communicate their firm’s type to prospective investors. Our variable %Ret represents the
percentage of common shares retained by the original retained by the main pre-IPO owners listed
in the prospectus. While the signalling literature predicts that this measure is negatively
associated with IPO underpricing (Downes and Heinkel 1982; Datar et al. 1991; Firth and Liau-
Tan 1998), Beatty and Welch (1996) report that the sign on this parameter depends on whether
the sample consists of young/small company IPOs or of older/larger ones. No prediction is made
on the sign of the parameter.

    Hughes (1986) and Willenborg (1999) suggest that the reputation of the underwriter hired for
the issue serves as a signal of firm quality. Because the market for investment bankers in Québec
is characterised by the presence of large firms that are present only in the province, we use
Bédard et al.’s (2000) classification in which underwriter quality (UW) is defined as a categorical
variable which equals 1 if the underwriter is a large Canadian firm, 2 if the underwriter is one of
the largest Québec firms, and 3 if the underwriter is a small Canadian firm.6 Larger Canadian
firms are expected to be associated with lower levels of underpricing.

    The last signal we include as control variable is the quality of the auditor chosen by the firm
(Aud-h). Higher quality auditors are expected to be associated with lower levels of underpricing.




6
 There is no official classification of underwriters in Canada. Bédard et al.’s (2000) classification is based on the
number of employees of each underwriting firm in Canada and in Québec as published every year in Les affaires
500. Moreover, all issues are underwritten on a firm commitment basis.

                                                                                                                  16
Other control variables

The effect of the issuing firm’s riskiness in the IPO process has been the focus of a large number
of studies.7 As in Equation (2) we use three measure of risk: Financial leverage (Lev) a proxy for
cash flow risk, Age which proxies for the stage of development of the firm and the number of risk
factors listed in the prospectus (RiskFact). We expect lower levels of risk to be associated with
lower levels of underpricing.

      Firms that go public via an IPO may choose to issue units (i.e., bundles combining common
shares and other securities such as warrants) rather than common shares alone. Unit IPOs
(Unit=1) have been found to be issued by firms that are smaller and younger than those who issue
common share IPOs and to experience significantly higher levels of underpricing (Schultz 1993;
Hogan 1997). We also control for share price-related effects by using the inverse of the IPO price
per share, expressed in constant dollars, as a control variable (IPOprice) and for firm size
(LnAssets) (Ibbotson et al. 1988; Tinic 1988; and Guenther and Willenborg 1999).

      The last control variables in Equation (3) are related to the specificity of our sample. Québec
residents pay income taxes at both the federal and provincial levels. Individuals who are resident
in the province of Québec at the end of the taxation year are entitled, in computing their taxable
income for Québec income tax purposes for that year, to deduct a stipulated percentage of the
cost of qualifying shares purchased during the year and included in a stock savings plan. The
deduction rates range from 0 to 150 percent, depending on the class of shares issued, the size and
type of the issuer, as well as the year of the issue. Bédard et al. (2006) examine the issue of
implicit taxes and the sharing of the benefits between the entrepreneur and investors and they find
that the tax deduction is an important element in explaining underpricing in this particular market.
We use the deduction rate (TaxDed) to control for this implicit tax effect.

      Finally, because the IPOs included in the sample are spread out over 21 years, we include a
series of dummy variables (Yeark) to indicate the year of each issue and to control for inflation
and economic growth over the sample period.




7
    Feltham et al. (1991), Clarkson and Simunic (1994) and Bédard et al. (2000), among others.

                                                                                                  17
Sample
We examine our three research questions and test our hypotheses on a sample of 246 IPOs of
Québec-based corporations during the period 1982 to 2002. All issues that pre-date the 1999
reform of Canadian stock exchanges were made on the Montreal Stock Exchange, while the
issues made after 1998 were listed on the TSE but under the Québec SEC rules. The sample
includes all firms that have made an IPO in the province of Québec in that period, except those in
the mining and resources sector because their prospectus is based more on geological than on
financial data (Lee et al. 2004). Of the 246 sample firms, 145 are incorporated under the
Canadian Business Corporation Act (CBCA) and 101 under the Québec Incorporation Act.

   All accounting data as well as information about governance, forecasts and the characteristics
of the issues were hand-collected from the IPO prospectuses. Market prices were collected from
the Financial Post.

   Table 2 presents descriptive statistics for the sample of IPOs. A brief look at the statistics
shows that board features are significantly different between the firms with an audit committee
(61% of the sample) and those without. The median board size is 7 members for both subsamples,
but the mean is significantly higher when there is an audit committee. Further, the average board
is composed of a majority of independent directors (60% and 53%). We also find that in a large
proportion of our sample firms the CEO plays a dual role by serving also as chairperson of the
board (Bdual=1), and this dual role is more prevalent in firms without an audit committee (79%
vs. 66%). Given the relatively small size of the sample firms (with median assets of 16.6 and 11.2
millions $CAN), this prevalence of dual role of the CEO is to be expected. The average audit
committee is composed of 3 directors with an average 49% of independent members (ACind) and
39% of the committees include at least one member with financial expertise (ACcomp). Finally,
25% of the committees have a majority of independent members and at least one with financial
expertise (ACqual).
                                              Table 2
   Issuers with an AC are on average larger (although the difference is not significant) and
disclose significantly more risk factors in their prospectus (RiskFact). The other characteristics do
not seem to differ significantly across the two groups. Both subsamples have an equivalent
proportion of firms issuing earnings forecasts, except that those with an AC disclose more
forecast assumptions (Fhor), and their forecasts are audited in a larger proportion (Faudit). Panel

                                                                                                  18
B of Table 2 indicates that although our sample IPOs are clustered in the years leading to the
stock market crash of 1987, the presence of audit committees is fairly well distributed over time.
In all years except 5, there are issuers with and without a separate audit committee.

   Table 3 presents the correlation between the various variables.
                                               Table 3

4- EMPIRICAL RESULTS

The decision to create an Audit Committee
Table 4 reports the results of estimating logistic regression (1) exploring the decision to create an
audit committee. Because the legal basis of incorporation (federal vs. provincial) has an incidence
on the requirements to create an AC, we run the regression first on the full sample and then on
two subsamples: firms incorporated under the CBCA and those incorporated under the Québec
Incorporation Act.
                                               Table 4
   The results of the full sample analysis show that the model is significant (Likelihood Ratio of
34.18) and has an adjusted pseudo R2 of 17.59%. The result show board size (Bsize) and the dual
role played by the CEO have a significant effect (at the 5% and 10% levels respectively) on the
decision to create an AC in preparation for the IPO. Board independence (Bind), however, is not
significant although in the right direction. Of the control variables, only the Dey report variable is
significantly associated with the decision at the 5% level. All the other control variables have the
expected sign although none is significant.

   When we split the sample on the basis of the law under which each firm is incorporated, a
clearer picture emerges. Firms incorporated under the CBCA have an obligation to create an AC
when they are listed. While most of them did it before the IPO, 27% of them did not. As in the
first regression, board size and the dual role of the CEO seem to affect the decision to create an
AC or not, but none of the control variables is significant. For these firms, the push for change in
governance structures generated by the Dey report does not seem to have had any effect.
Moreover, the low values of the Likelihood Ratio and the adjusted R2 indicate that the model is
not very powerful in explaining the creation of ACs in this subsample.

   The last two columns relate to the subsample of firms incorporated under Québec laws, the
subset of firms which were under no obligation to create an AC, even after being listed on an

                                                                                                   19
exchange. This is reflected in the lower percentage of firms with an AC in this subsample (44%)
than for the firms incorporated under the CBCA (73%). The increase in the performance of the
model both in terms of Likelihood Ratio and of R2 is remarkable, compared with the full sample
and the first subsample regressions. This seems to indicate that this subsample is the best
indicated to examine the determinants of the creation of an AC.

   The results of the regression are as expected and consistent with those of previous studies.
Both the size (Bsize) and the independence of the board (Bind) have a significantly positive effect
on the likelihood of the creation of an AC. This is consistent with results obtained by Menon and
Williams (1994) and Pincus (1989). On the other hand, the dual role of the CEO (Bdual) seems to
have no impact at all on the decision. This is contrary to what was found by Beasley and Salterio
(2001) that this dual role has a negative effect on the likelihood that the AC be of a quality higher
than the strict minimum required by regulations.

   Of the control variables, the publication of the Dey report (Dey) seems to have had an
important impact on the decision by firms incorporated under Québec law to create an AC.
Surprisingly, the presence of a quality auditor (Aud-h) is negatively related to the creation of an
AC, suggesting that for these firms AC and audit quality seem to be substitutes rather than
complements.

   Our results seem to indicate that the board of director plays a central role in the decision to
create an audit committee, whatever the basis of incorporation. IPO firms with larger boards and
a higher proportion of independent directors tend to create audit committee in an IPO context and
the Canadian Dey report of the TSE seems to have had a major impact on the creation of ACs,
especially for firms that were under no regulatory obligation to create them.

Governance and Disclosure Quality
Table 5 reports the results of estimating our regression model (2) examining the relation between
governance quality and disclosure quality as measured by management forecast errors. Contrary
to our expectations, neither board attributes (Model 2a) nor the presence of an AC (Model 2b) are
significantly associated with disclosure quality. This result is not consistent with those of
Karamanou and Vafeas (2005) who find a negative association between board and AC
independence and their measure of disclosure quality, the likelihood that managers update their
earnings forecasts. Our result may be explained by the fact that earnings are very difficult to

                                                                                                  20
predict at the time of an IPO and that the forecast error is due more to factors related to firm
operations and characteristics than to the quality of its governance structure.
                                               Table 5
   This conjecture is supported by the fact that the variables which are significant in the
regression are related to the firm (Growth) and to the characteristics of the forecast itself, the
length of the forecast period (Fhor) and the fact that it is audited or not (Faudit). These results are
consistent with those of other studies on the determinants of forecast errors (McConomy 1998;
Clarkson 2000). It seems that H1 is not supported in the sample of this study and that the quality
of governance has no significant impact on the quality of the forecast, over and above the
characteristics of the forecast itself.

Governance and IPO Underpricing
Table 6 shows the results of testing H2. In the first three models, we separate the effects of the
attributes of the board of directors from those of the AC, while Model 3d includes both the
quality of the board (Bqual) and that of the AC (ACqual) as explanatory variables. The results of
estimating Model 3a seem to suggest that the mere existence of an AC (AC=1) has no significant
effect on the level of underpricing of the issue, and thus on the cost of capital of Québec IPO
firms. Similarly, the quality of the board of directors as measured by the index Bqual is not
significant in either Model 3c or Model 3d, suggesting that board attributes have no effect on cost
of capital either.

   The significantly negative coefficient on ACqual, an index of AC overall quality, suggests that
it is the quality of the AC that can convince investors to require a lower return on the firm’s
shares, decreasing the level of underpricing. Indeed, in both Model 3b where ACqual is the only
governance variable included and in Model 3d where it is added to the board quality index Bqual,
its coefficient is negative at -0.089 and significant at the 5% level.

   In all four versions of the model, the riskiness of the firm as measured by Lev seems to
increase significantly the level of underpricing while larger firms (LnAssets) seem to suffer from
less underpricing and the tax advantage of the issue for investors (TaxDed) decreases the required
rate of returns requested on the newly issued shares. Finally, consistent with results of other
studies on IPO underpricing, the quality of the auditor seems to reassure investors and to be
negatively related to underpricing.


                                                                                                    21
   Thus, contrary to Certo et al. (2001), we find no clear evidence of reduction in underpricing
caused by the characteristics of the board of directors or the existence of an audit committee. It
appears that, at least in our context, board quality does not add anything to other signals that are
found to be relevant in prior studies. One possible explanation could be that all firms have a
board of directors and the difference in quality is perceived as marginal by investors. Our results,
however, suggest that it is not the existence of an audit committee, nor the quality of the board of
directors that matters, but the quality of the audit committee. The audit committee quality play an
important role with a decrease of up to 8.9% in underpricing for IPO firms with an independent
and financially competent audit committee.
                                              Table 6




5-CONCLUSION

We contribute to the debate surrounding the effectiveness of governance practices by examining
IPO firms’ decision to voluntarily create an audit committee, and the effect of governance
characteristics on the quality of disclosure as measured by management forecast errors and on the
cost of capital as measured by IPO underpricing. Our results seem to indicate that the creation of
an audit committee is more likely when the board of directors is larger, a higher proportion of its
members are independent and it is not chaired by the firm’s CEO. On the other hand, neither the
quality of the board nor the presence of an AC seem to have any effect on the forecast error when
earnings forecasts are included in the IPO prospectus. Governance characteristics do not seem to
affect the quality of financial information issued by IPO firms. Finally, the cost of capital as
measured by the level of underpricing seems to be negatively related to the quality of the AC
more than to the quality of the board or the mere existence of an AC at the time of the IPO.

   The setting in which the tests are performed is particular in that the sample firms had some
freedom as to their choice of governance structure. In fact, the effect of governance quality on the
likelihood of the creation of an AC is strongest in the subsample of firms who were under the
least regulatory constraints, those incorporated under the Québec Incorporation Act. Given the
low level of litigation risk in Canadian markets, and particularly in Québec, we can examine the



                                                                                                 22
effect of governance in a setting that is different from that of most studies which are set in the US
context. We find that our results are mostly consistent with those of previous studies.




                                                                                                  23
References

Ackerlof, G. 1970. The market for ‘Lemons’. Quarterly Journal of Economics, 74, pp.488-500.

Beasly M. and S. Salterio (2001) The relationship between board characteristics and voluntary
  improvements in audit committee composition and experience. Contemporary Accounting
  Research, v 18 n 4, pp 539-70.

Beatty, R.P., and I. Welch (1996) Issuer expenses and legal liability in initial public offerings.
  Journal of Law and Economics, 39, pp. 545-602.

Beatty, R.P., and J.R. Ritter (1986) Investment banking, reputation and the underpricing of initial
  public offerings. Journal of Financial Economics, 15, January/February, pp. 13-232.

Bédard, J., D. Coulombe, and L. Courteau (2000) Demand and supply of auditing in IPOs: An
  empirical analysis of the Québec market. International Journal of Auditing, 4, pp. 227-245.

Bédard, J., D. Coulombe, and S. Paquette (2006) Tax Incentives on Equity and Firms’ Cost of
  Capital: Evidence from the Quebec Stock Savings Plan, Contemporary Accounting Research.
  Forthcoming.

Bédard, J., S. Marakchi-Chtourou and L. Courteau (2004) “The Effect of Audit Committee
  Expertise, Independence, and Activity on Aggressive Earnings Management”. Auditing: A
  Journal of Practice and Theory 23, p. 13–35.

Beekes, W. and P. Brown (2006). Do Better-Governed Australian Firms Make More Informative
  Disclosures? Journal of Business Finance & Accounting, 33(3) & (4), 422–450,

Carter, R., F. Dark, and A. Singh (1998) Underwriter reputation, initial returns, and the long-run
  performance of IPO stocks. Journal of Finance, 53, pp. 285-312.

Certo T. S., C. M. Daily and D. R. Dalton (2001), Signalling Firm Value Through Board
  Structure: An Investigation of Initial Public Offerings, Entrepreneurship Theory & Practice,
  26/2, pp33-50.

Clarkson, P.M., (2000), Auditor Quality and Accuracy of Management Earnings Forecasts,
    Contemporary Accounting Research, v17 n4, pp 595-622.

Clarkson, P.M., and D.A. Simunic. (1994) The association between audit quality, retained
  ownership, and firm-specific risk in the US vs. Canadian IPO markets. Journal of Accounting
  and Economics, 17, January, pp. 207-228.

Courteau, L. (1995) Under-diversification and retention commitments in IPOs. Journal of
  Financial and Quantitative Analysis, 30, December, pp. 487-508.

Datar S.M., G.A. Feltham, and J.S. Hughes (1991) The role of audits and audit quality in valuing
  new issues. Journal of Accounting and Economics, 14, March, pp. 3-49.

                                                                                                24
Davidson, R., D. Neu, (1993), A Note on the Association Between Audit Firm Size and Audit
  Quality, Contemporary Accounting Research, v9, pp 479-488.

Downes, D., and R. Heinkel (1982) Signalling and the valuation of unseasoned new issues.
  Journal of Finance, 37, pp. 1-10.

Eichenseher, J. and Shields, D. Spring 1985. Corporate director liability and monitoring
   preferences. Journal of Accounting and Public Policy 4(1):13-31.

Feltham, G.A., J.S. Hughes, and D.A. Simunic (1991) Empirical assessment of the impact of
   auditor quality on the valuation of new issues. Journal of Accounting and Economics, 14,
   December, pp. 375-399.

Filatotchev I. and K. Bishop (2002) Board Composition, Share Ownership, and Underpricing of
   U.K. IPO Firms, Strategic Management Journal, 23, pp941-955.

Finkle, T. A. (1998), The Relationship Between Boards of Directors and Initial Public Offerings
   in the Biotechnology Industry, Entrepreneurship Theory & Practice, 22, pp 5-29.

Firth, M., and C.K. Liau-Tan (1998) Auditor quality, signalling, and the valuation of initial public
   offerings. Journal of Business Finance and Accounting, January/March, 25, pp. 145-166.

Gompers, P., J. Ishii, A. Metrick, (2003) Corporate Governance and Equity Prices, The Quarterly
  Journal of Economics, February, pp.107-154.

Grinblatt, M., and C.Y. Hwang (1989) SignalingSignalling and the pricing of new issues. The
   Journal of Finance, 44, June, pp. 393-420.

Guenther, D.A., and M. Willenborg (1999) Capital gains tax rates and the cost of capital for small
  business: evidence from the IPO market. Journal of Financial Economics, 53, September,
  pp.385-408.

Gul, F.A. and S. Leung, (2004). Board leadership, outside directors expertise and voluntary
  corporate disclosures. Journal of Accounting and Public Policy 23, p. 351–379

Hogan, C.E. (1997) Cost and Benefits of Audit Quality in the IPO Market: A Self-Selection
  Analysis. The Accounting Review, 72, January, pp. 67-86.

Hughes, Patricia J. (1986) Signaling by direct disclosure under asymmetric information. Journal
  of Accounting and Economics, 8, June, pp. 119-142

Ibbotson, R., J. Sindelar, and J. Ritter (1988) Initial public offerings. Journal of Applied
   Corporate Finance, 1, pp. 37-45.

Karamanou, I. and N. Vafeas (2005), The Association Between Corporate Boards, Audit
  Committees, and Management Earnings Forecasts: An Empirical Analysis, Journal of
  Accounting Research, v 43 n 3, pp453-86.



                                                                                                 25
Klein, A. 2002. Audit committee, board of director characteristics, and earnings management.
  Journal of Accounting and Economics 33, p.375-400.

Lee, P., D. Stokes, S. Taylor and T. Walter (2003). The association between audit quality,
  accounting disclosures and firm-specific risk: Evidence from initial public offerings. Journal
  of Accounting and Public Policy 22 (2003) 377–400

Leland, H. and D. Pyle. 1977. InformationsInformation asymmetries, financial structure, and
   financial intermediation. Journal of Finance, 32, pp. 371-387.
McConomy, B., (1998), Bias and Accuracy of Management Earnings Forecasts: An Evaluation of
   the Impact of Auditing, Contemporary Accounting Research, v15, pp 167-195.

Menon, K. and J.D. Williams (1994). The Use of Audit Committees for Monitoring. Journal of
  Accounting and Public Policy, 13, p. 121-139.

Pincus, K., Rusbarsky, M. and Wong, J. Winter 1989. Voluntary formation of corporate audit
   committees among NASDAQ firms. Journal of Accounting and Public Policy 8(4):239-265.

Romano, R. 2005.The Sarbanes-Oxley Act and the Making of Quack Corporate Governance.
  Yale Law Journal 114 (June): 1521-1612.

Schultz, P. 1993. Unit initial public offerings : A form of staged financing. Journal of Financial
  Economics 34 (October) 199-219.

Tinic, S. (1988) Anatomy of initial public offerings of common stock. Journal of Finance, 43, pp.
   789-822.

Welch, I. (1989) Seasoned offerings, imitation costs, and the underpricing of initial public
  offerings. The Journal of Finance, 44, June, pp. 421-499.

Welch, I., and J. Ritter (2002) A review of IPO activity, pricing, and allocations. Yale ICF
  Working Paper No. 02-01, available from http://papers.ssrn.com/abstract=296393.

Willenborg, M. (1999). Empirical analysis of the economic demand for auditing in the initial
  public offerings market. Journal of accounting research, 37, spring p.225-238.




                                                                                               26
                                      TABLE 1
                              Description of variables
  Bsize =      Number of directors on the board;
  Bdual =      Categorical variable that equals 1 if the CEO also chairs the board and 0
               otherwise;
   Bind        % of members on the board who are not managers of the firm and who have
               no business or family relationship with the firm or its managers;
  Bqual        Categorical variable that equals 3 if Bdual=0, Bsize>7 (the median board
               size of the sample) and Bind>50%, equals 2 (1) if only 2 (1) of these
               conditions hold and 0 otherwise
     AC        Categorical variable that equals 1 if the firm has an audit committee at the
               time of the IPO and 0 otherwise;
ACcomp         Categorical variable that equals 1 if at least one member of the audit
               committee has an accounting title, is a CFA or has experience as CFO and 0
               otherwise;
  ACind        Categorical variable equal to 1 if more than 50% of the audit committee
               members are independent
 ACqual        Categorical variable equal to 1 if ACind = 1 and ACcomp = 1 and 0
               otherwise
   %Ret    =   Percentage of the firm’s shares retained by initial owners;
    Age    =   Number of years from the foundation or incorporation to the IPO;
  Assets   =   Pre-IPO total assets in millions of Canadian dollars;
  Aud-h    =   Categorical variable equal to 1 if the auditor at the time of the IPO is a Big
               Eight/Big Six Firm (highest quality), and 0 otherwise;
    Dey        Categorical variable that equals 1 if the prospectus is issues after the Dey
               report and 0 otherwise;
  Faudit       Categorical variable equal to 1 if the forecast was audited and 0 if it was
               only reviewed by the auditor
   Fhor        Number of months between the date of approval of the prospectus by the
               board and the end of the forecast period
   Fhyp        Number of assumptions on which the earnings forecast is based, as listed in
               the prospectus
Forecast       Categorical variable equal to 1 if the prospectus includes an earnings
               forecast and 0 otherwise
  Growth       Average growth in revenue in the three years preceding the issue
IPOprice =     The inverse of the IPO offer price per share or per unit, in constant dollars;
     Lev =     Financial leverage, defined as total debt over total assets.
LnAssets       Natural log of pre-IPO total assets in constant Canadian dollars
  LnFerr       Log of the absolute value of the error in the earnings forecast included in the
               prospective as a percentage of forecasted earnings

                                                                                           27
                                TABLE 1 (continued)
                               Description of variables
   RiskFact   Number of risk factors listed in the prospectus
Subordinate   Categorical variable equal to 1 if shares offered have a ratio of voting to
              ownership rights different from 1 and zero otherwise
    TaxDed    Tax deduction rate allowed for Québec income tax purposes on the IPO
              issue
     Undpr = Underpricing, computed as the difference between the closing price for the
              first day of trading following the issue and the initial offer price, as a
              percentage of the closing price;
       Unit = Categorical variable equal to 1 if the IPO is a unit offering and 0 otherwise;
       UW = Categorical variable which takes the value of 1 if the firm’s underwriter is a
              large Canadian firm, 2 if the underwriter is considered one of the largest
              Québec firms, and 3 if the underwriter is a small Canadian firm;
      Yeark   Categorical variable equal to 1 if the IPO was made in year k and 0
              otherwise, k = 1982,...,2001




                                                                                          28
                                           Table 2.
           Descriptive Statistics of the Sample, Partitioned by audit committeea
                     Panel A: Mean and median of explanatory variables

                              Firms with an audit                      Firms without an audit
                                  committee                                  committee                                Difference
                                   (n=150)                                     (n=96)                                 in meansb
                              Mean              Median                     Mean                   Median
Undpr                          0.05              0.00                      0.04                    0.00                0.02
Bsize                          7.71              7.00                      6.82                    7.00                0.89***
BInd                           0.60              0.60                      0.53                    0.57                0.07**
Bdual                          0.66              1.0                       0.79                    1.0                -0.13**
ACind                          0.49              0.42                      --                      --
ACcomp                         0.39              0.0                       --                      --
ACqual                         0.25              0.0                       --                      --


Age (years)                   22.28             11.59                   20.50                     14.17                1.78
Assets (M $CAN)             61.04               16.63                   35.84                     11.22               25.20
Growth                       1.36                0.30                    0.76                      0.27                0.60
RiskFact                    11.60                7.00                    8.98                      6.00                2.62*
%Ret                         0.68                0.71                    0.69                      0.72               -0.01
Aud-h                        0.53                1.00                    0.46                      0.00                0.07
UW                           1.90                2.00                    2.07                      2.000               0.17
Forecast                     0.47                0.00                    0.52                      1.00               -0.05
Fhyp                        10.01                8.00                    8.02                      7.00                1.99*
Fhor                         7.74                7.00                    7.92                      8.00               -0.18
Faudit                       0.31                0.00                    0.04                      0.00                0.27***


Panel B: Distribution of IPOs across years

Yearc           82   83   84     85   86   87    88     92   93   94   95       96   97      98    99    00      01    02
No audit
                1    3    5      14   40   12    2      0    3    0    0        2    3       5     2     3       0     1
committeec
Audit
                0    1    6      19   49   19    0      1    14   4    3        10       7   1     5         7   3     1
committee
Total           1    4    11     33   89   31    2      1    17   4    3        12   10      6     7      10     3     2
a
 Sample of primary issues in the Province of Québec between 1983 and 2002. See variable definitions in Table 1.
b
  Test statistics are for F-tests except for categorical variables for which a χ2 test for independence is used.
* Significant at the 10% level; **Significant at the 5% level; ***Significant at the 1% level (two-tailed tests).
c
    Number of issues in each year where the issuing firm did not have an audit committee.



                                                                                                                                   29
                                                                                                TABLE 3
      Correlation coefficients between variables (Pearson (Spearman) correlation coefficients are above (below) the diagonal)a

                                                                                                         Risk                       IPO
               Taxded Unit        Age      UW       LnAssets Growth %Ret        Aud-h Dey       Fhor     fact       Faudit Fhyp     price       Lev       Bsize   Bind     Bdual    AC       ACQual BQual
    Taxded           1 -0,142     0,050    0,177      -0,082   -0,089   0,070 -0,176 -0,393 -0,110 -0,406 -0,221 -0,386 -0,158 -0,179 -0,051 -0,206 -0,028 -0,044                             -0,031 -0,096
    Unit        -0,148        1   0,006    0,100      -0,141   -0,021 -0,008 -0,141 -0,067       0,018 -0,016 -0,183        0,086    0,166 -0,041         0,019   0,017     0,010 -0,010      -0,043 -0,038
    Age          0,040   0,011          1 -0,223      0,317    -0,186   0,084 -0,001 -0,155 -0,076 -0,204 -0,027            0,016 -0,319 -0,142           0,015 -0,168 -0,044 -0,051          0,051 -0,072
    UW           0,220   0,100 -0,215           1     -0,471   0,020    0,027 -0,093 -0,128      0,169 -0,055 -0,201 -0,132          0,430      0,095 -0,148 -0,053         0,120 -0,097      0,016 -0,160
    lnAssets    -0,164 -0,172     0,345 -0,504            1    -0,006   0,103   0,175 -0,004 -0,206 -0,042          0,286   0,304 -0,608 -0,367           0,306   0,038 -0,144      0,110     -0,032   0,192
    Growth       0,161 -0,102 -0,262       0,120      -0,203       1 -0,116     0,063   0,130 -0,033      0,348     0,050 -0,138     0,005 -0,028         0,038   0,149 -0,038      0,086     0,115    0,088
    %Ret        -0,052 -0,022     0,078 -0,021        0,174    -0,029      1 -0,168 -0,338 -0,236 -0,266 -0,241 -0,222 -0,163 -0,018                      0,018 -0,290      0,138 -0,032      -0,033 -0,217
    Aud-h       -0,085 -0,141     0,001 -0,094        0,205    -0,003 -0,176       1    0,390    0,046    0,269     0,232   0,114 -0,054        0,080     0,000   0,060 -0,153      0,067     0,102    0,178
    Dey         -0,262 -0,067 -0,176 -0,129           0,064    0,078 -0,338     0,390       1 -0,035      0,703     0,946   0,620    0,177      0,122 -0,116      0,176 -0,032      0,179     0,146    0,077
    Fhor        -0,042   0,026 -0,084      0,178      -0,247   -0,081 -0,234    0,037 -0,025           1 -0,059 -0,063      0,219    0,179      0,062 -0,156      0,248 -0,042 -0,018         -0,004   0,180
    Riskfact    -0,251   0,040 -0,237 -0,036          -0,182   0,159 -0,286     0,239   0,718 -0,059            1   0,604   0,560    0,110      0,075 -0,032      0,217 -0,075      0,104     0,171    0,156
    Faudit      -0,116 -0,183 -0,055 -0,201           0,304    0,071 -0,288     0,232   0,946 -0,058      0,554         1   0,546 -0,188        0,176 -0,076      0,149 -0,046      0,340     0,196    0,026
    Fhyp        -0,273   0,110    0,068 -0,232        0,350    -0,214 -0,156    0,095   0,551    0,172    0,363     0,518         1 -0,107      0,106     0,024   0,168 -0,095      0,210     0,117    0,164
    IPOprice     0,163   0,233 -0,285      0,620      -0,706   0,101 -0,089 -0,187 -0,061        0,209    0,075 -0,196 -0,164               1   0,317 -0,191 -0,009         0,165 -0,038      -0,107 -0,168
    Lev         -0,008 -0,078     0,131 -0,034        0,292    -0,085   0,016 -0,031 -0,130      0,082 -0,174       0,180   0,189 -0,013              1   0,022   0,091     0,051 -0,101      -0,035   0,043
    Bsize       -0,074 -0,062 -0,010 -0,169           0,313    -0,128 -0,008    0,076 -0,104 -0,056 -0,111 -0,140           0,174 -0,254        0,148         1   0,246 -0,176      0,177     0,174    0,536
    Bind        -0,166   0,013 -0,162 -0,058          0,038    -0,103 -0,322    0,086   0,199    0,278    0,135     0,144   0,198 -0,080        0,045     0,246          1 -0,348   0,155     0,259    0,698
    Bdual       -0,035   0,010 -0,052      0,120      -0,113   0,213    0,128 -0,153 -0,032 -0,072 -0,005 -0,046 -0,065              0,148      0,006 -0,175 -0,371            1 -0,142       -0,184 -0,674
    AC          -0,063 -0,010 -0,066 -0,097           0,098    0,037 -0,055     0,067   0,179 -0,032      0,135     0,340   0,220 -0,070 -0,015           0,156   0,139 -0,142           1    0,337    0,141
    ACQual      -0,038 -0,043     0,043    0,016      -0,034   -0,002 -0,061    0,102   0,146    0,025    0,154     0,196   0,126 -0,098 -0,107           0,129   0,261 -0,184      0,337         1    0,210
    BQual       -0,097 -0,027 -0,072 -0,155           0,189    -0,225 -0,244    0,180   0,076    0,186    0,057     0,034   0,243 -0,170        0,080     0,588   0,724 -0,651      0,140     0,206       1
a
    All variables are defined in Table 1

Bold significant at 5% level
                                                     TABLE 4
                     Logistic regression of the choice to create an audit committee
        AC i = α + β1 Bsize i + β 2 Bind i + β 3 Bdual i + β 4 Age i + β 5 Growth i + β 6 Lev i + β 7 Subordinate i
                  + β 8 Dey i + β 9 LnAssets i + β10 % Re t i + β11 Aud - h i + β12 UWi + ε i

                                               All                  Canada onlyb                     Quebec only
                                   Coefficient                   Coefficient                    Coefficient
Variablesa               Pred.       estimate        p valuesc    estimate p valuesc            estimate    p valuesc
Bsize                      +           0.094          0.019        0.110       0.053              0.211         0.003
Bind                       +           0.440          0.168       -0.460       0.513              1.053         0.081
Bdual                       -         -0.295          0.081       -0.578       0.050              0.000         0.999
Age                        ?           0.002          0.299        0.003        0.645             0.010         0.078
Growth                     +           0.043          0.134        0.024        0.264             -0.025        0.995
Lev                        +          -0.043          0.435       -0.184        0.681             -0.024        0.834
Subordinate                 -          0.099          0.682        0.066        0.835             0.113         0.804
Dey                        +           0.714          0.013        0.218        0.226             2.330       <0.001
LnAssets                   ?          -0.045          0.540       -0.153        0.186             0.216         0.114
%Ret                        -          0.655          0.259        0.891        0.235             0.532         0.645
Aud-h                      ?           0.297          0.282        0.206        0.450             -0.717        0.052
UW                         ?           0.008          0.473       -0.047        0.764             0.277         0.120


Intercept                             -1.000          0.289        1.341        0.317             -5.571        0.002


Number of
observations                             246             ns          145                            101
% with an AC                            61%                         73%                            44%
Likelihood Ratio
Test                                   34.18          0.001        10.05         0.61             41.36       <0.001
Adjusted R2                         17.59%                         9.74%                        45.06%



______________________________
a
  The variables are defined in Table 1.
b
  Canada only and Québec only refer to subsamples containing IPO firms incorporated under the Federal and Provincial
company laws, respectively.
c
  We report one-tailed tests where prediction is supported.
                                          TABLE 5
          GLM regression of forecast errors on board attributes and control variables

                 LnFerri = α + β1 Bsize i + β 2 Bind i + β 3 Bdual i + β 4 AC i + β 5 Fhypi
                            + β 6 Fhori + β 7 Faudit i + β 8 Age i + β 9 Growth i + β10 RiskFact i
                            + β11 Aud - h i + β12 LnAssets i + ε i

                                                   Model 2a                               Model 2b
                                           Coefficient                            Coefficient
Variablesa               Prediction         estimate      p valuesb                estimate      p valuesb
Bsize                            -                -0.072              0.38
Bind                             -                 0.943              0.27
Bdual                            +                 0.170              0.66
AC                               -                                                      -0.038        0.90


Fhyp                             +                 0.039              0.40               0.031        0.49
Fhor                             +                 0.102             <0.01               0.118       <0.01
Faudit                           -                -1.242              0.02              -1.110        0.04
Age                              -                -0.007              0.32              -0.006        0.35
Growth                           +                 0.342              0.01               0.345       <0.01
RiskFact                         +                 0.050              0.23               0.055        0.17
Aud-h                            -                -0.173              0.58              -0.197        0.53
LnAssets                         ?                 0.031              0.84               0.023        0.88
Intercept                                         -3.091              0.03              -2.986        0.03


Number of observations                               116                                   116
Model F Value                                       3.37             <0.01                3.93       <0.01
             2
Adjusted R                                       18.49%                               18.65%
a
  The variables are defined in Table 1.
b
  We report two-tailed tests




                                                                                                             32
                                                                      TABLE 6
                        GLM regression of underpricing on governance attributes and control variables
     Undpri = α + β1 Bqual i + β 2 AC i + β 3 ACqual i + β 4 % Re t i + β 5 UWi + β 6 Aud - h i + β 7 Age i + β 8 Lev i + β 9 RiskFact i + β10 Unit i
             + β11 IPOprice i + β12 LnAssets i + β13 TaxDed i + ∑ γ k Yearik + ε i
                                                                   k


                                            Model 3a                  Model 3b                     Model 3c                         Model 3d
                                    Coefficient                 Coefficientp valuesb        Coefficient                     Coefficient
Variablesa                Pred.       estimate    p valuesb      estimate                     estimate      p valuesb         estimate      p valuesb
Bqual                       -                                                                  -0.003         0.826           -0.0008         0.961
AC                          -         0.0001         0.96
ACqual                      -                                      -0.089           0.04                                       -0.089          0.05
%Ret                        ?          0.048        0.63           0.058            0.56       0.042           0.68            0.057            0.58
UW                          +          0.030        0.15           0.033            0.12       0.029           0.16            0.033            0.12
Aud-h                       -         -0.092       <0.01          -0.089           <0.01      -0.090          <0.01           -0.089            0.01
Age                         -         -0.010        0.54          -0.007            0.65      -0.011           0.48           -0.007            0.66
Lev                         +          0.002       <0.01           0.002           <0.01       0.002          <0.01            0.002          <0.01
RiskFact                              0.0001        0.96          0.0005            0.83      0.0002           0.92           0.0005            0.82
Unit                                  -0.022        0.58          -0.023            0.56      -0.021           0.60           -0.022            0.57
IPOprice                    +         -0.019        0.79          -0.036            0.61      -0.019           0.78           -0.036            0.61
LnAssets                    -         -0.027        0.06          -0.030            0.04      -0.025           0.09           -0.030            0.05
TaxDed                      -         -0.196       <0.01          -0.196           <0.01      -0.193          <0.01           -0.195          <0.01
Years                                               ns                               ns                        ns                               ns
Intercept                              0.464        0.14           0.468            0.13        0.449          0.15             0.489         0.117
N.                                       246                         246                          246                             246
Model F Value                           3.33       <0.01             3.48        <0.01           3.28         <0.01              3.34         <0.01
Adjusted R2                            21.1%                       22.0%                       20.7%                           21.7%
a
    The variables are defined in Table 1.
b
    We report two-tailed tests




                                                                              33

								
To top