Boards of Directors: An Empirical Study of
New Zealand Initial Public Offering Firms
Y. T. Mak
NATIONALUNIVERSITY OFSINGAPORE
M. L. Roush
VICTORIAUNIVERSITY OFWELLINGTON
This study examines the associations between the characteristics of boards ship) (e.g., Jensen, 1993), have been argued by researchers to be better able to monitor management.
of directors of initial public offering (IPO) firms and the availability of
growth opportunities and level of inside share ownership of these firms. Regulatory agencies in different countries view certain char-acteristics of boards of directors as being important
determi-Three characteristics of boards of directors are examined: board size,
proportion of outside directors, and the separation of the CEO and chair- nants of their effectiveness. For example, in the United States (U.S.), the New York Stock Exchange (NYSE) requires that
person’s roles (dual leadership). The impact of firm size on board
charac-teristics is controlled for in the empirical tests. Based on a sample of 110 listed companies have audit committees made up only of outside directors. In addition, the Securities and Exchange
New Zealand firms that made initial public offerings of equity securities
over the period 1983 to 1987, this study finds that firms that have lower Commission (SEC) recommended that nominating and com-pensation committees should be dominated by outside
direc-inside share ownership tend to employ larger boards. In addition, the
proportion of outside directors is positively related to the extent of growth tors (Wolfson, 1984). Brickley, Coles, and Jarrell (1994) also discuss calls in the U.S. by the United Shareholders’
Associa-opportunities available to a firm and negatively related to inside share
ownership. Finally, firms with relatively more growth opportunities are tion, large public pension funds, legislators and regulators, and the financial press for dual leadership.
likely to have dual leadership. Although some contrary results are found,
the findings from this study provide some support for arguments that Initial empirical research on board composition has tended to examine the association between board characteristics and
firms with greater agency problems, attributable to low inside share
ownership and significant growth opportunities, are likely to choose boards firm performance (e.g., Baysinger and Butler, 1985; Rechner and Dalton, 1991), and results from these studies are mixed.
of directors that are more effective at mitigating these problems. J BUSN
RES2000. 47.147–159. 1999 Elsevier Science Inc. These studies assume that boards with certain characteristics
(such as dual leadership or more outside directors) are better able to carry out their functions, regardless of the characteristics of the firms employing these boards, and therefore, that firms adopting these board characteristics will outperform those that
T
he corporate governance role of the board of directorsdo not. In contrast, several recent studies have examined the has become an issue of great interest to both researchers
associations between firm characteristics and board character-and regulatory agencies. The board of directors is often
istics, based on the assumption that there is no board structure seen as serving a monitoring function, protecting the interests
that is optimal for all firms and that firms will choose a of various stakeholders against management’s self-interests
board structure that is appropriate to its circumstances (e.g., (Fama, 1980; Fama and Jensen, 1983). Larger boards (e.g.,
Finkelstein and D’Averi, 1994; Rediker and Seth, 1995). Zahra and Pearce, 1989), boards with more independent,
The present study adopts the more recent perspective in outside directors (e.g., Fama and Jensen), and those with the
examining the determinants of board characteristics of New CEO’s and chairperson’s roles being separated (dual
leader-Zealand (N.Z.) firms making initial public offerings (IPOs) of equity securities (IPO firms). It tests hypotheses that firms Address correspondence to Dr. Y. T. Mak, Department of Finance and Ac- associated with greater agency problems are likely to choose counting, National University of Singapore, 10 Kent Ridge Crescent,
Singa-pore 119260, SingaSinga-pore. E-mail: [email protected] board characteristics that enhance a board’s monitoring ability.
Journal of Business Research 47, 147–159 (2000)
1999 Elsevier Science Inc. All rights reserved. ISSN 0148-2963/00/$–see front matter
Previous research suggests that the availability of growth op- interests. In evaluating the IPO, they will, therefore, take into account any monitoring expenditures they will have to incur portunities (e.g., Smith and Watts, 1992; Gaver and Gaver,
to ensure that inside owners act in their interests, and the 1993) and inside share ownership (e.g., Jensen and Meckling,
cost of the inside owners diverging from their interests that 1976) are important determinants of agency problems. The
cannot be cost-effectively eliminated through monitoring. This present study investigates associations between these two
vari-“price protection” can take the form of a discount in the ables and the board characteristics of board size, proportion
offering price of the securities, or as an ultimate form of price of outside directors, and dual leadership. The impact of firm
protection, investors may refuse to purchase shares offered by size is also controlled for in the empirical tests because of its
the firm. To reduce the price protection demanded, inside likely impact on board characteristics (e.g., Pfeffer, 1973;
Za-owners of IPO firms may voluntarily put in place mechanisms hra and Pearce, 1989).
for limiting such divergence of interests. The next section The present study contributes to the literature on the
deter-provides an overview of the model underlying the study, minants of board characteristics in several respects. First, apart
which shows the sources of agency problems of IPO firms from Beatty and Zajac (1994), previous studies have not
con-and the board characteristics that can control these agency sidered board characteristics of IPO firms. In the IPO context,
problems. agency problems are likely to be significant and to have an
important influence on board characteristics. Second, previous
studies have mainly been conducted in the U.S., where regula-
Model Overview
tory requirements or pressure means that the choice of boardSeveral researchers argue that the board of directors is an characteristics, particularly the use of outside directors, is not
element of corporate governance, particularly for monitoring completely voluntary. In contrast, regulatory agencies in N.Z.
the behavior of top managers. For example, the board of have not required or applied pressure on firms to choose
directors has been described as perhaps the ultimate internal boards with particular characteristics, such as using relatively
monitor of top managers (Fama, 1980) and as the “common more outside directors. Consequently, any use of these
moni-apex” of the decision control systems of organizations charac-toring techniques on the part of N.Z. firms is completely
terized by a separation of ownership and decision-making voluntary. Third, previous studies have generally not
exam-(Fama and Jensen, 1983). Similarly, the board of directors is ined the association between the nature of a firm’s investment
likely to represent an important mechanism for limiting the opportunities and its board characteristics.
divergence of interests between inside owners and potential investors for IPO firms.
Agency Problems in the
Inside owners choose the board of directors of IPO firmsat the time of the offering of securities. That is, regardless of
IPO Context
the number or proportion of shares that may eventually be Previous researchers argue that tender offer bids (Byrd and held by outside investors as a result of the security issuance, Hickman, 1992) and management buyout (MBO) transactions inside owners decide the number and composition of the (Lee, Rosenstein, Rangan, and Davidson, 1992) are character- board of directors and whether the CEO should also be the ized by significant agency problems. Similarly, we argue that chairperson of the board. However, the ability of potential there are significant agency problems when initial public offer- outside investors to “price-protect” provides a strong incentive ings are made. The proposed issue of securities by IPO firms for inside owners to increase the monitoring ability of the closely resembles the situation described by Jensen and Meck- board where agency problems are significant.
ling (1976), where an existing owner-manager is attempting to We argue that although the board of directors serves insti-sell equity claims to outside investors. The sale of equity claims tutional, governance, and strategic functions that may conflict to outside investors leads to a divergence of interests between (Goodstein, Gautam, and Boeker, 1994), the governance func-the owner-manager and outside investors, because func-the fact tion is likely to be particularly important at the time of a that the owner-manager will no longer own 100% of the firm public offering of securities. Consequently, we focus on the means that the owner-manager will not bear the full costs of monitoring role of the board and study the associations be-actions that reduce the value of the firm. In the case of IPO tween variables designed to capture cross-sectional differences firms, major agency problems are likely to exist between poten- in agency problems and board characteristics that can control tial investors and parties closely associated with these firms who agency problems. Figure 1 shows the model underlying the have the ability to expropriate wealth from outside investors to present study.
Figure 1. Model of relationship between board characteristics and firm characteristics.
ment in growth opportunities is less observable to outside Guatam, and Boeker, 1994; Judge and Zeithaml, 1992). Simi-larly, the few published empirical studies that have examined investors, and managerial discretion is correspondingly greater
(Smith and Watts, 1992). Consequently, there is greater scope board size have done so mainly from the resource dependence (e.g., Pfeffer, 1973) and strategic decision-making perspectives for managers of firms with relatively more investment in
growth opportunities to expropriate wealth from outside in- (e.g., Goodstein, Guatam, and Boeker). The results of these studies seem to indicate that larger boards are beneficial from vestors to themselves. The percentage of shares retained by
insiders affects agency problems, because the larger the pro- the resource-dependence perspective but dysfunctional from portion of shares insiders hold, the closer their interests are the strategic decision-making perspective. There is little empiri-aligned to outside investors (Jensen and Meckling, 1976). cal research that has specifically examined the determinants However, three characteristics of the board can help control or consequences of board size from a monitoring perspective. these agency problems and they are: larger board size (e.g.,
Singh and Harianto, 1989), more outside directors (e.g., Fama
Outside Directors
and Jensen, 1983) and dual leadership (e.g., Jensen, 1993). According to Fama and Jensen (1983), the effectiveness of This is discussed further in the following section the board for decision control can be enhanced by including
outside directors on the board. Outside directors have incen-tives to monitor management on behalf of shareholders,
be-Board Characteristics as a
cause the demand for directors’ services, and therefore, the
Response to Agency Problems
value of their human capital, is dependent upon their
effective-Board Size
ness as decision control specialists (Fama and Jensen, 1983). Many studies have examined the association between the Board size has been argued as affecting the monitoring abilityuse of outside directors and financial performance, based on of boards. Larger boards are often believed to be more capable
the assumption that firms using relatively more outside direc-of monitoring the actions direc-of top management, because it is
tors will show better financial performance. The results from more difficult for CEOs to dominate larger boards. This
en-these studies are mixed. However, apart from the difficulty in hances the independence of the board from the CEO, which
capturing the effects of using outside directors in overall firm increases the board’s ability and willingness to use its decision
performance, the rationale underlying these studies has been control powers to ratify or refute decisions made by the CEO
challenged. For example, Hermalin and Weisbach (1991) argue (Zahra and Pearce, 1989). Similarly, Singh and Harianto (1989)
that if firms have boards that are optimally weighted between suggest that larger boards can make it more difficult for the
insiders and outsiders, or if they reduce agency problems to CEO to obtain consensus for taking actions that harm
share-approximately the same level, it may be impossible to find a holders’ interests.
relationship between firm performance and board composition, Some authors argue that smaller boards may be more
effec-even if board composition helps control agency problems. tive, but they generally do so from the view of the board’s
impor-tance of outside directors for corporate governance in different being derived from incentives, and therefore, presumably higher agency costs, are more likely to have dual leadership. contexts, including as a substitute for takeover restriction
In summary, previous theoretical and empirical research (Brickley and James, 1987), removing the CEO for poor firm
generally suggests that firms that with greater agency problems performance (Weisbach, 1988), in tender offer bid situations
are likely to employ boards that are larger, have relatively more (Byrd and Hickman, 1992), and in management buyout
situa-outside directors, or dual leadership, because such boards are tions (Lee, Rosenstein, Rangan, and Davidson, 1992). A
num-more effective in controlling agency problems. Although larger ber of studies have found that agency-related factors affect
board size, more outside directors, and dual leadership may the use of outside directors. Brickley and James (1987) find
reduce a board’s effectiveness in carrying out other functions a negative relationship between the degree of concentration
(e.g., Goodstein, Guatam, and Boeker, 1994; Judge and Zei-of ownership and number and proportion Zei-of outsiders, but
thaml, 1992), the importance of controlling agency problems only in nonacquisition states. This is consistent with
concen-in an IPO context leads to the prediction concen-in the present study tration of ownership and outside directors being substitute
that firms with greater agency problems are more likely to devices for controlling managerial behavior. Beatty and Zajac
employ boards having these characteristics. (1994) find that firms whose managers owned relatively fewer
shares use a large proportion of outside directors. Rediker and Seth’s (1995) results indicate that the use of outside directors is
Sources of Agency Problems
negatively related to the presence of large outside shareholders,
and Hypotheses
insider ownership, and the mutual monitoring potential of top
management. Bathala and Rao (1995) find that, consistent with
Inside Share Ownership
the predictions of agency theory, the use of outside directorsFor an IPO firm, where potential investors face possible wealth is negatively associated with the proportion of managerial
transfers to inside owners, agency costs are likely to be depen-share ownership, the dividend payout ratio, and the ratio of
dent upon share ownership by these owners. The smaller the long-term debt to the sum of total debt and equity.
proportion of shares held by inside owners, the lower the costs that they would bear from taking actions that reduce
Dual Leadership
firm value (Jensen and Meckling, 1976). Therefore, it isex-It may also be advantageous from a monitoring perspective pected that the lower the proportion of shares retained by not to have the CEO of the organization also performing the inside owners, the greater the monitoring ability of the board chairperson’s role (Jensen, 1993). Given that chairpersons are employed by an IPO firm. Previous research indicates that the heads of boards of directors, they are likely to be able to managerial ownership is negatively related to the proportion affect the operations of the boards significantly by developing of outside directors (e.g., Bathala and Rao, 1994; Beatty and operating procedures, influencing the configuration of com- Zajac, 1994) and dual leadership (Beatty and Zajac). Although previous studies have not examined the association between mittees, and serving as the ultimate persons responsible for
share ownership and board size, a negative relationship be-performance of the boards. The chairperson often schedules
tween these two variables is expected on the basis that larger and sets the agenda for meetings and is likely to be responsible
boards are likely to be more effective at controlling agency for communicating information to external directors prior to
problems (e.g., Singh and Harianto, 1989). Therefore, the board meetings (Rechner, 1989).
following hypothesis is tested: Studies that have examined the association between dual
leadership and performance have found conflicting results H1: The proportion of inside share ownership is negatively (e.g., Rechner and Dalton; 1991; Donaldson and Davis, 1991; related to board size, proportion of outside directors, Mallette and Fowler, 1992). However, these studies have the and dual leadership.
same potential weaknesses as those studies that have examined the association between the use of outside directors and
perfor-mance. Some studies attempt to provide more direct evidence
Availability of Growth Opportunities
with greater investment in growth opportunities) increases IPOs remained relatively infrequent for much of the period after December 1988. The following types of firms are ex-the potential for wealth transfers from potential investors to
inside owners (Gaver and Gaver, 1995). cluded from the study:
Little empirical research has examined the association
be-1. oil and gas exploration firms and mining firms—these tween the availability of investment opportunities and board
typically are no-liability firms that are involved in highly characteristics. Anderson, Francis, and Stokes (1993) find that
speculative activities, which may affect their compara-relative expenditure on monitoring from directors is positively
bility with other firms; related to the availability of growth opportunities. Bathala
2. firms making offerings of securities that are unavailable and Rao (1995) note that firms with more future growth
for subscription by the general public; and opportunities can be argued to use more outside directors on
3. unit trusts—the nature of these entities, and, therefore, the board to control the higher agency problems inherent in
their governance structures, may not be comparable to such firms. Other researchers have also hypothesized that
other types of firms. firms with greater growth opportunities adopt mechanisms
and corporate policies (e.g., compensation plans) that better A total of 110 firms that met the selection criteria are included control agency problems, and there is substantial empirical in the study.
support for these predictions (e.g., Smith and Watts, 1992;
Gaver and Gaver, 1993).
Variable Measurement
Based on previous studies, we argue that firms with more Except where indicated otherwise, information for measuring growth opportunities will adopt boards that are better able the variables employed in the study was obtained from the to monitor the decisions and actions of inside owners, to prospectuses issued by the sample firms.
ensure that they are in the interests of potential investors.
BOARD SIZE (BRDSIZE). This variable is measured by the total This leads to the following hypothesis:
number of members on the board of directors. H2: The availability of growth opportunities is positively
related to board size, proportion of outside directors, PROPORTION OF OUTSIDE DIRECTORS (OUTDIR). The definition of outside directors adopted in this study includes only inde-and dual leadership.
pendent outside directors. According to Byrd and Hickman
(1992, p. 199): “The traditional two-way classification scheme
Control Variable
of ‘insider’ (corporate employee) or ‘outsider’ (nonemployee)FIRM SIZE. Previous agency theory-based studies have used directors fails to consider potential conflicts of interest when firm size as a proxy for agency costs and have argued that directors are not full-time employee but have affiliations with larger firms are more likely to use various monitoring mecha- the firm.”
nisms, such as external auditing (Chow, 1982) and audit Under the definition of outside directors adopted in the committees (e.g., Pincus, Rusbarsky, and Wong, 1989). Simi- present study, directors with the following affiliations with the larly, larger IPO firms facing greater agency problems are more firm are not deemed to be outside directors: (1) an employee of likely to employ boards that are more effective in monitoring the firm; (2) an individual or member of a firm who has sold managers. However, firm size may capture other factors, and or proposes to sell assets or other business interests to the firm; it is difficult to interpret relationships involving firm size as (3) an individual who provides consulting or other services, or being unambiguously attributable to agency-related factors. who is a member of a firm that provides such services; (4) a It is, however, important to include firm size as a control major shareholder or a director appointed by a major share-variable in the present study, because relationships between holder; or (5) a director or employee of a related firm. Firms firm size and board size (e.g., Pfeffer, 1973; Cotter, Shivdasani, are considered to be related to the IPO firms where: (1) the and Zenner, 1997), proportion of outside directors (e.g., Cot- new firm is formed to take over some of the activities of the ter, Shivdasani, and Zenner), and dual leadership (e.g., Fin- other firm, and shares are issued as consideration; (2) the firm kelstein and D’Aventi, 1994; Boyd, 1995) have been suggested being issued the shares is the promoter of the new firm; or (3)
or documented by other researchers. there are cross-shareholdings between the two firms. The
exclusion of a major shareholder or a director appointed by a major shareholder from the definition of outside directors
Data Collection
is consistent with Rosenstein and Wyatt (1990). We believe
Sample Selection
this exclusion is appropriate for our study, because, as we explained earlier, the major concern of potential investors in The sample for this study is based on N.Z. firms that madeinitial public offerings of equity securities between 1983 and the IPO context is that actions will be taken that benefit inside owners at the expense of potential investors (outside or new 1987, and which subsequently listed on the N.Z. Stock
is likely to be perceived as being more aligned to the interests through the conversion of an existing unlisted firm to a pub-of inside owners than potential investors. Note that we are licly listed firm with little change in operations. In this case, referring to major shareholders at the time of the offering, the major purpose of the IPO may be to allow the firm to who are, therefore, part of the group of inside owners of the increase its access to outside capital or its spread of sharehold-firm. The proportion of outside directors is calculated by ing to meet listing requirements, or to allow existing share-dividing number of outside directors by the total number of holders to reduce their shareholding. Alternatively, a new firm
directors. may be formed by merging or acquiring existing firms, with
these existing firms often becoming subsidiaries of the new
DUAL LEADERSHIP (DUALEAD). A binary variable is used to
firm. In both cases, the IPO firms have previous operating measure the existence of dual leadership. Firms that have
histories and potential investors in these firms will be acquir-CEOs also serving as chairpersons of the board are coded as
ing claims in a combination of future growth opportunities 0, and those with dual leadership as 1.
and existing assets-in-place.
AVAILABILITY OF GROWTH OPPORTUNITIES (GROWTH). Previ- On the other hand, an IPO firm may be formed to com-ous studies operationalize growth opportunities in several mence operations for the first time. Potential investors in such ways, including market value to book value of total assets start-up firms, which do not have past operating histories, (e.g., Smith and Watts, 1992), market value to book value of will be acquiring claims in future growth opportunities, be-equity (e.g., Gaver and Gaver, 1993), earnings to price ratio cause these firms do not have existing assets-in-place. It is (e.g., Gaver and Gaver, 1993), and variability of returns (e.g., expected that firms without operating histories will have rela-Smith and Watts, 1992). However, none of these studies tively more growth opportunities and that these firms will
pro-focused on IPO firms. vide greater monitoring through their board of directors.
For the IPO firms included in the present study, some of The existence of an operating history (HISTDUM) is mea-these measures cannot be used. For example, many IPO firms sured by a binary variable, with firms having operating histories do not have previous earnings histories that rule out the use coded as 1, and those without operating histories coded as 0. of the earnings to price ratio. In this study, four alternative
YEARS OF OPERATING HISTORY. Another measure used in this measures are used to measure the availability of growth
oppor-tunities and these are discussed below. study that is closely related to the previous operating history measure is the number of years of operating history. That is,
VARIANCE OF RETURNS. Variance of returns (VARRET) has
rather than measuring operating history as a binary variable, been used in many previous studies to operationalize the
operating history here is measured by a continuous variable. availability of growth opportunities (e.g., Smith and Watts
The assumption underlying this measure is that relatively 1992; Gaver and Gaver, 1995). The use of variance of returns
more established firms are likely to have more assets-in-place is based on the argument that the value of investment options
compared to growth opportunities. Both measures of op-increases with the variability of cash flows (Chung and
Cha-erating history have the advantage of being ex ante measures roenwong, 1991; Kester, 1986), which is, in turn, expected
of the availability of growth opportunities. to be reflected in the variance of returns. However, because
Years of operating history (HISTCON) is measured by the price and returns data for IPO firms are only available after
number of years between the date of formation of the IPO listing, this measure can only be calculated on an ex post basis
firm and the prospectus date. for these firms. Consequently, there are limitations associated
with this particular measure. OFFERING SIZE. Another potential measure of the availability
VARRET is measured by the variance of the difference be- of growth opportunities for IPO firms is the relative size of tween the firm’s equity return and the market return; that is, the offering. Assuming that IPOs are designed primarily to Var(ri2rm). Returns for the first 20 trading days after listing, fund growth opportunities, the relative size of the offering
excluding the initial return, are used to calculate this variable. (OFFER) can proxy for the extent of growth opportunities This is similar to the method used by Beatty (1989) to calculate
available to the IPO firm. OFFER is measured by the amount the ex post measure of ex ante uncertainty for IPO firms, except to be raised through the IPO divided by firm size (see below). that returns are adjusted for market movements. This
adjust-This measure has the advantage of being ex ante in nature ment is important, because the study extends over several years,
but has the major limitation of ignoring growth opportunities where significant differences in market conditions may have
funded by existing resources. Furthermore, although relative existed. The measure used assumes that all IPO firms have
offering size is likely to affect the need to enhance the reputa-the same market risk, an assumption consistent with most
tion of the offeror, and, therefore, the monitoring ability of previous studies involving IPO firms (Saunders, 1990).
the board of directors, it may be a relatively weak measure of availability of growth opportunities.
OPERATING HISTORY. For IPO firms, another potential
mea-Each of the three measures is likely to partially and imper-sure of the availability of growth opportunities is the existence
Fol-Table 1. Descriptive Statistics for Independent and Dependent Vari-lowing Gaver and Gaver (1993, 1995), a combined measure
ables (n5110) is derived by factor analyzing the three continuous measures
of VARRET, HISTCON, and OFFER. As Gaver and Gaver Variable Mean SD Median
(1993, p. 133) note: “The investment opportunity set is
inher-OUTDIRa 0.58 0.26 0.60
ently unobservable and is likely to be imperfectly measured
DUALEADb 0.85 — —
by any single empirical proxy” (p. 133), and factor analysis BRDSIZEc 5.28 1.53 5.00
allows the isolation of “the underlying construct that is com- VARRETd 24.98 61.89 10.17
ln(VARRET) 2.34 1.21 2.32
mon to all of the measures” (p. 137). According to Hair,
HISTDUMe 0.60 — —
Anderson, Tatham, and Black (1995, p. 373): “Variables for
HISTCONf 9.09 16.39 2.0
factor analysis are generally assumed to be of metric
measure-ln(HISTCON)g 1.45 1.27 1.39
ment.” Therefore, in this study, the dummy variable, HIST- OFFERh 0.46 0.59 0.33
DUM, is not included in the factor analysis. To assess the INOWNi 0.53 0.23 0.58
SIZE (NZ$m)j 20.52 70.76 7.01
robustness of the results, the relationship between selected
ln(SIZE) 16.23 1.23 16.08
individual measures and board characteristics is also
exam-ined. This is discussed further in the Data Analysis and Results aProportion of outside directors.
section. bDual leadership, zero if no dual leadership, one if there is dual leadership. For this
variable, the mean represents the proportion of firms with dual leadership.
cTotal number of directors.
INSIDE SHARE OWNERSHIP (INOWN). First, the number of ordi- dVariance of (Ri–Rm) over the first 20 trading days after listing, excluding the first
trading day.
nary shares retained by inside owners is computed. This is
eExistence of operating history, coded 0 for firms without operating histories, and 1 done by adding the following: shares to be issued to existing for firms with operating histories. For this variable, the mean represents the proportion
of firms wth an operating history.
shareholders, promoters, management, directors, staff, and
fNumber of years of operating history.
associates (including related firms), and shares outstanding gNatural logarithm of (HISTORY11). The constant term was added to eliminate zero
values of HISTORY prior to taking log (Wall, 1986, p. 145).
before the initial public offering. The number of ordinary hAmount of offering divided by firm size. shares retained by inside owners is then divided by the total iProportion of shares retained by inside owners.
jFirm size, measured by book value of total assets. number of ordinary shares outstanding after the completion
of the issue (assuming all shares to be issued were subscribed).
it is evident from Table 1 that IPO firms on average employ
FIRM SIZE (SIZE). Firm size is measured by the book value of
a majority of these directors. Also, most of the IPO firms have total assets, which is calculated by adding the book value of
dual leadership. These characteristics may be indicative of total assets immediately prior to the IPO plus the proceeds from
significant agency conflicts existing in the IPO context. Jensen new shares issued in the IPO. The analyses are rerun using an
(1993) argues that it may be easier for the CEO to control alternative measure of firm size, calculated as the market value
boards that are too large, which he defines as having more of ordinary shares (the total number of shares after the issue
than seven to eight people. It should be noted that only three multiplied by the first listing price), plus book value of total
out of the 110 sample firms had more than eight directors debt and preference shares immediately prior to the IPO.
on their boards. Therefore, most of the companies had director Because the results are unaffected by the size measure used,
numbers not exceeding the maximum board size suggested only the results using the first measure are presented.
by various authors (Jensen; Lipton and Lorsch, 1992; Patton and Baker, 1987). In other words, we do not seem to be dealing with boards that are so large that effective decision
Data Analysis and Results
making and decision control are inhibited.
Descriptive Statistics
Table 1 shows descriptive statistics for the independent, de-
Factor Analysis
pendent, and control variables included in the study. The Following Gaver and Gaver (1993, 1995), the continuous mea-distributions of VARRET, HISTCON, and SIZE are positively sures of GROWTH—ln(VARRET),ln(HISTCON), and OFFER— skewed, and this is reflected in the mean values for these are factor analyzed to derive an aggregate GROWTH measure. variables being significantly larger than their median values. Principal component factor analysis is used to decompose Therefore, the natural log transformation is applied to these each of the three measures into one or more common factor(s) three variables in order to reduce the skewness in their distri- and a factor unique to each of the individual measures. butions before conducting the empirical tests. The distribu- Table 2 presents the results of the factor analysis. Only tions of the logged measures are approximately normal, and one factor had a eigenvalue greater than one, and this first this is supported by Table 1, which shows that the mean and factor explained 55% of the total variance. All three individual
median values are very close. measures had factor loadings of 0.60 or better for this factor.
Table 2. Principal Components Factor Analysis of Three Measures CORE. Note, however, that the high correlations between
of Growth Opportunities GRSCORE, and INOWN andln(SIZE), can affect the ability
to make inferences from the estimates of the coefficients of
Results for Extraction of Component Factors
these variables. Because Table 3 also indicates that one of the
% Cumulative %
individual measures of growth opportunities, ln(VARRET),
Factor Eigenvalue Variance Variance
is not as highly correlated with INOWN andln(SIZE), the
1 1.6630 55.4 55.4 sensitivity of the potential impact of multicollinearity can be
2 0.7789 26.0 81.4 assessed by rerunning the analysis withln(VARRET)
substi-3 0.5581 18.6 100.0 tuted for GRSCORE.
Factor Matrix
Factor 1
Multivariate Tests
BOARD SIZE. Ordinary least-squares (OLS) regression is used
In(VARRET) 0.7873
to examine the relationship between board size and the
ex-OFFSIZE 0.6450
planatory variables, and the following model is estimated:
In(HISTCON) 20.7919
Descriptive Statistics of the Common Factor Extracted from BRDSIZE5 a1b1GROWTH1b2INOWN 1b3ln(SIZE)
Three Measures of Growth Opportunities
Three separate models are estimated, using three alternative
Maximum 4.691 measures of growth opportunities, GRSCORE,ln(VARRET),
Minimum 22.055 and HISTDUM. Recall that HISTDUM is the categorical
mea-Median 0.099
sure of growth opportunities not used in the factor analysis
Mean 0.000
to derive GRSCORE, whileln(VARRET) is used, because it is less highly correlated with the other explanatory variables.
The results of this regression are shown in Table 4. All factor is used as the primary measure of availability of growth three estimated models are significant at p , 0.01. For all opportunities in the subsequent analyses. three models, inside share ownership is significantly related to board size in the expected direction atp,0.05 or better.
Correlations
Not surprisingly, the relationship between firm size and board size is positive and significant (p,0.01). However, the direc-Table 3 presents the Pearson correlations among thecontinu-ous explanatory variables, including correlations between the tion of the associations for the three measures of growth opportunities are in the opposite direction to that predicted, aggregate GRSCORE and the three underlying individual
mea-sures of availability of growth opportunities. In view of the with only HISTDUM being significant atp,0.05. This result may reflect the problems that large boards have in making significant correlations among the explanatory variables,
GRS-CORE, INOWN and ln(SIZE), a multivariate analysis should timely strategic decisions. (Goodstein, Gautam, and Boeker, 1994; Judge and Zeithaml, 1992). That is, although larger be used to analyze the relationship between these variables
and each board characteristic. A multivariate analysis allows boards may be more effective in controlling agency problems (as reflected in the significant negative relationship between an assessment of the relative importance of each explanatory
variable after controlling for the correlations between them. board size and inside share ownership), growth firms may find it important to have boards that can make timely strategic The significant correlations between GRSCORE and the
underlying individual measures of growth opportunities pro- decisions and such firms may therefore prefer smaller boards. The need for both greater monitoring and more timely vide further support for the argument that the individual
measures contributes significantly to the variation in GRS- decision making in growth firms may lead to countervailing
Table 3. Pearson Correlations between Explanatory Variables (n5110)
Variables ln(VARRET) ln(HISTCON) OFFER GRSCORE INOWN
ln(HISTCON) 20.4418***
OFFER 0.2669** 20.2747**
GRSCORE 0.7873*** 20.7919*** 0.6450***
INOWN 20.0780 0.4226*** 20.4296*** 20.4048***
ln(SIZE) 20.2142* 0.3316** 20.4479*** 20.4319*** 0.4746***
Table 4. OLS Regression of Board Size on Explanatory Variablesa
Independent Expected
Variable Direction Model 1 Model 2 Model 3
GROWTH
GRSCORE 1 20.2420
(1.105)
HISTDUM 2 0.5937
(1.747)**
ln(VARRET) 1 0.0270
(0.228)
INOWN 2 20.0171 20.0206 20.0153
(2.518)** (2.855)*** (2.308)**
ln(SIZE) NA 0.4497 0.4356 0.5012
(3.432)*** (3.400)*** (3.892)***
R2 0.1411 0.1316 0.1009
5.803 5.354 3.965
F-value (p,0.001) (p,0.002) (p,0.01)
aFor all models, the coefficients and the associatedt-statistics (in parentheses) are shown.
*Significant atp,0.10 (one-tailed). **Significant atp,0.05 (one-tailed). ***Significant atp,0.01 (one-tailed).
effects on board size, resulting in the weak and inconsistent DUAL LEADERSHIP. Finally, because dual leadership is a bi-relationship. However, because the relationship between board nary variable, logit analysis is used to estimate the relationship size and availability of growth opportunities is only significant between this variable and the explanatory variables, and the when HISTDUM is used to measure the availability of growth following model is estimated:
opportunities, it is possible that operating history is capturing
DUALEAD5 a1 b1GROWTH1b2INOWN1 b3ln(SIZE)
factors other than the extent of growth opportunities. For example, it may be that as firms become more established (and,
The results of the logit analysis are shown in Table 6. All therefore, having longer operating histories), more directors are
three models are significant atp, 0.05 or better. All three added to the board, perhaps through the promotion of the
measures of growth opportunities are related to dual leader-CEO and other senior executives to the board. In addition,
ship in the expected direction, with GRSCORE andln (VAR-more established firms may be able to afford larger boards.
RET) being statistically significant (p,0.01). However, the Overall, there is only weak support for the argument that
relationship for HISTDUM is weak (p,0.10). Inside share IPO firms with greater agency costs are likely to use larger
ownership is not significantly related to dual leadership. boards of directors to attempt to reduce these costs.
Therefore, there is only weak support for the agency theory predictions of choice of dual leadership. Consistent with
PROPORTION OF OUTSIDE DIRECTORS. OLS regression is also
Boyd’s (1995) observation, there is also a significant negative used to examine the multivariate relationship between the
relationship between firm size and dual leadership. proportion of outside directors and the explanatory variables,
and the following model is estimated: There are at least two potential explanations for the rela-tively weak findings for dual leadership. First, most firms (94 OUTDIR5 a1b1GROWTH1b2INOWN 1b3ln(SIZE) out of 110) in the sample have dual leadership, which makes
it difficult to find a statistical relationship for this variable. As Table 5 shows, all three models are significant atp,
Second, there are costs and benefits associated with dual lead-0.0001. The measure of growth opportunities in each of the
ership. It may be that the costs of dual leadership, which three models is significant atp,0.05, while the inside share
include agency problems associated with having a non-CEO ownership variable is significant atp,0.05 or better, with
chairperson, information costs, costs of changing the succes-the direction of succes-these associations being consistent with
expec-sion process, and other costs (Brickley, Coles, and Jarrell, tations. The results are consistent with the argument that firms
1994), outweigh the benefits of dual leadership. Boyd (1995) associated with greater agency costs are likely to use relatively
notes that dual leadership may not be appropriate under con-more outside directors for monitoring purposes. There is a
ditions of resource scarcity or high complexity. However, note weak positive relationship (p,0.10) between firm size and
that the findings on the relationship between firm characteris-proportion of outside directors, which is consistent with the
Table 5. OLS Regression of Proportion of Outside Directors on Explanatory Variablesa
Independent Expected
Variable Direction Model 1 Model 2 Model 3
GROWTH
GRSCORE 1 0.1103
(3.110)***
HISTDUM 2 20.1848
(3.350)***
ln(VARRET) 1 0.0543
(2.834)***
INOWN 2 20.0031 20.0023 20.0040
(2.783)*** (1.939)** (3.735)***
ln(SIZE) 1 0.0341 0.0320 0.0273
(1.604)* (1.539)* (1.307)*
R2 0.1860 0.1968 0.1743
F-value 8.075 8.656 7.459
(p,0.0001) (p,0.0001) (p,0.0001)
aFor all models, the coefficients and the associatedt-statistics (in parentheses) are shown.
*Significant atp,0.10 (one-tailed). **Significant atp,0.05 (one-tailed). ***Significant atp,0.01 (one-tailed).
and may not be generalizable to other countries with signifi- boards. In addition, consistent with predictions, the propor-cantly different regulatory requirements, such as the U.S. tion of outside directors is positively related to the extent of growth opportunities available to a firm and negatively related to inside share ownership. Finally, as predicted, firms with
Summary and Concluding
relatively more growth opportunities are likely to use dual
Comments
leadership, although there is no relationship between insideshare ownership and dual leadership. Overall, although there This study examines the impact of agency problems on
charac-are some contrary results, the findings provide some support teristics of boards of directors of IPO firms. After controlling
for arguments that firms with greater agency problems, attrib-for firm size, the results show that firms with lower inside share
utable to low inside share ownership and significant growth ownership tend to employ larger boards. However, contrary to
opportunities, are likely to choose boards of directors that are expectations, there is some evidence that firms expected to
have more growth opportunities tend to employ smaller more effective at mitigating these problems.
Table 6. Logistic Regression of Dual Leadership on Explanatory Variablesa
Independent Expected
Variable Direction Model 1 Model 2 Model 3
GROWTH
GRSCORE 1 1.2829
(2.496)***
HISTDUM 2 21.1572
(1.375)*
ln(VARRET) 1 0.6555
(2.307)***
INOWN 2 0.0166 0.0158 0.0050
(1.07) (0.958) (0.349)
ln(SIZE) 1 20.3706 20.4192 20.4066
(1.966)** (1.836)** (1.643)**
Model likelihood index 0.1343 0.0838 0.1263
Model likelihood statistic 12.255 7.646 11.528
(p,0.007) (p,0.06) (p,0.01)
aFor all models, the coefficients and the associatedt-statistics (in parentheses) are shown.
For example, in the U.S., various regulatory agencies have
Limitations
been made rules or recommendations that require firms to There are limitations associated with the present study. First,
use more outside directors (Wolfson, 1984). There has also in developing the study hypotheses, this study only considered
been pressure from various sources for firms to have dual the board of directors’ role in monitoring managers. Clearly,
leadership (Brickley, Coles, and Jarrell, 1994). Requiring all boards of directors also perform other roles. The omission of
firms to adopt particular board characteristics ignore that cer-other factors unrelated to the demand for monitoring may
tain firms may not require the greater monitoring provided explain some of the contrary results found in this study.
by having more outside directors, larger boards or dual leader-Second, some of the proxies used could capture other
ship. This imposes unnecessary costs on firms with low agency factors that may or may not be related to agency problems.
problems. Furthermore, other board functions not considered In particular, the measurement of availability of growth
oppor-in the present study, such as strategic decision makoppor-ing, may tunities is problematic. For example, variance of returns is
also require board characteristics that are quite different from clearly a risk measure and an imprecise measure of growth
those required for monitoring purposes. Again, mandating opportunities. Therefore, relationships involving this variable
that firms adopt particular board structures may result in these could reflect the impact of risk, rather than growth
opportuni-firms having boards that are ineffective for performing other ties, although risk itself can be a source of agency problems
functions. (Simunic and Stein, 1987). Similarly, operating history may
capture information asymmetry because investors in firms
without operating histories must rely mainly on information
Future Research
provided by inside owners to evaluate the IPO; whereas, in- The preponderance of firms in the present study with dual vestors in firms with operating histories have the previous leadership can be contrasted with the findings of Brickley, track records and financial information as additional sources Coles, and Jarrell (1994), who find that most firms in their of information. Although significant information asymmetry sample have unitary leadership. This difference may be attrib-is a key characterattrib-istic of growth firms (Gaver and Gaver, 1995) utable to the different types of firms included in the two and is an important source of agency problems (Whittred and studies. This explanation is supported by Beatty and Zajac’s Zimmer, 1992), relationships involving operating history may (1994) study of U.S. IPO firms, where they also find that dual simply reflect the impact of information asymmetry rather leadership is the dominant form of board leadership, although than availability of growth opportunities. In other words, al- the proportion of firms using dual leadership in their sample though the relationships involving the measures of growth is lower than the proportion found in the present study of opportunities may still reflect the impact of agency problems N.Z. IPO firms. In contrast, Brickley, Coles, and Jarrell (1994) on board characteristics, they may not be agency problems focus on large existing firms included in aForbessurvey of that arise from the availability of growth opportunities. Unfor- executive compensation. That is, Brickley, Coles, and Jarrell’s tunately, problems of developing appropriate proxies for im- (1994) sample firms are very different from those used in portant constructs, such as the availability of growth opportu- Beatty and Zajac’s and the present study.
nities, pervade the financial economics literature. Comparing the use of dual leadership in Brickley, Coles, Finally, this study only considered the board of directors and Jarrell (1994), Beatty and Zajac’s (1994), and the present for controlling divergence of interests between owner-manag- study suggests that firms move from dual leadership to unitary ers and potential investors. In reality, a firm has available other leadership as they become more established. This lends cre-mechanisms for limiting such divergence, including incentive dence to Brickley, Coles, and Jarrell’s claim that unitary leader-contracts (e.g., Haugen and Senbet, 1981) and market mecha- ship is the equilibrium. According to them, dual leadership nisms (e.g., Fama, 1980). The choice between alternative “signifies normal succession periods or extraordinary, transi-mechanisms is likely to be affected by the relative costs (and
tory events” (p. 5). Perhaps IPOs represent “extraordinary, benefits) of these mechanisms. If firms choose from a package
transitory events,” and these firms move toward unitary leader-of these mechanisms and where these alternative mechanisms
ship, because the benefits of unitary leadership outweigh the may be used as complements or substitutes, such
single-equa-costs of unitary leadership. This presents potential issues for tion models as those used in this study may not adequately
future research; for example, (1) Do IPO firms adopt dual capture the factors that affect the use of outside directors.
leadership and as they mature, move toward unitary leader-ship?; and (2) If so, when do these changes occur, and are
Implications
these changes related to changes in the costs and benefits of The major implication from this study is that because cross- dual leadership as the firms’ circumstances change?
sectional differences in the level of monitoring provided by the board of directors seem to be related to the severity of
We thank the participants at the Department of Finance and Accounting
agency problems faced by these firms, regulations mandating Seminar Series at the National University of Singapore, and especially the all firms to adopt certain board characteristics may result in two anonymous reviewers for their insightful comments and helpful sugges-tions that have considerably improved the paper. Funding provided by
tional University of Singapore Academic Research Fund is also gratefully Gaver, Jennifer J., and Gaver, Kenneth M.: Compensation Policy and the Investment Opportunity Set.Financial Management24 (Spring
acknowledged.
1995): 19–32.
Goodstein, Jerry, Gautam, Kanak and Boeker, Warren: The Effects
References
of Board Size and Diversity on Strategic Change.Strategic Manage-Anderson, Don, Francis, Jere R., and Stokes, Donald J.: Auditing, ment Journal15 (1994): 241–250.Directorships, and the Demand for Monitoring.Journal of Account- Hair, Joseph F. Jr., Anderson, Rolph E., Tatham, Ronald L., and ing and Public Policy12 (Winter 1993): 353–375. Black, William C.:Multivariate Data Analysis with Readings, 4th Bathala, Chenchuramaiah T., and Rao, Ramesh P.: The Determinants ed., Prentice-Hall, Englewood Cliffs, NJ. 1995, p. 373.
of Board Composition: An Agency Perspective.Managerial and Haugen, Robert A., and Senbet, Lemma: Resolving Agency Problems Decision Economics16 (January 1995): 59–69. of External Capital through Options.Journal of Finance36 (June Baysinger, Barry, and Butler, Henry: Corporate Governance and the 1981): 629–647.
Board of Directors: Performance Effects of Changes in Board Hermalin, Benjamin E., and Weisbach, Michael S.: The Effects of Composition.Journal of Law, Economics and Organization1 (1985): Board Composition and Direct Incentives on Firm Performance.
101–124. Financial Management20 (Winter 1991): 101–112.
Beatty, Randolph P.: Auditor Reputation and the Pricing of Initial Jensen, Michael C.: Presidential Address: The Modern Industrial Public Offerings.The Accounting Review64 (October 1989): 693– Revolution, Exit and the Failure of Internal Control Systems.
709. Journal of Finance48 (July 1993): 831–880.
Beatty, Randolph P., and Zajac, Edward J.: Managerial Incentives, Jensen, Michael C., and Meckling, William H.: Theory of the Firm: Monitoring, and Risk-Bearing: A Study of Executive Compensa- Managerial Behavior, Agency Costs, and Ownership Structure. tion, Ownership, and Board Structure in Initial Public Offerings. Journal of Financial Economics3 (October 1976): 305–360. Administrative Science Quarterly39 (June 1994): 313–335.
Judge, William Q. Jr., and Zeithaml, Carl P.: Institutional and Strate-Boyd, Brian K.: CEO Duality and Firm Performance: A Contingency gic Choice Perspectives on Board Involvement in the Strategic Model.Strategic Management Journal16 (1995): 301–312. Decision Process. Academy of Management Journal35 (October Brickley, James A., and James, Christopher M.: The Takeover Market, 1992): 766–794.
Corporate Board Composition, and Ownership Structure: The
Kesner, Idalene F., Victor, Bart, and Lamont, Bruce T.: Board Compo-Case of Banking.Journal of Law and Economics30 (April 1987): sition and the Commission of Illegal Acts: An Investigation of
161–180. Fortune 500 Companies.Academy of Management Journal29
(De-Brickley, James A., Coles, Jeffrey L., and Jarrell, Gregg:Corporate cember 1986): 789–799.
Leadership Structure: On the Separation of the Positions of CEO and Kester, W. Carl: An Options Approach to Corporate Finance, in Chairman of the Board, Working paper, University of Rochester, Handbook of Corporate Finance, Edward I, Altman, ed., Wiley,
1994. New York. 1986, pp. 3–35.
Byrd, John W., and Hickman, Kent A.: Do Outside Directors Monitor Lee, Chun I., Rosenstein, Stuart, Rangan, Nanda, and Davidson Wal-Managers? Evidence from Tender Offer Bids.Journal of Financial lace N., III.: Board Composition and Shareholder Wealth: The Economics32 (October 1992): 195–221. Case of Management Buyouts.Financial Management21 (Spring Chow, Chee W.: The Demand for External Auditing: Size, Debt, and 1992): 58–72.
Ownership Influences. The Accounting Review57 (April 1982):
Lipton, Margin, and Lorch, Jay, W.: A Modest Proposal for Improved 272–291.
Corporate Governance.The Business Lawyer48 (November 1992): Chung, Kee H., and Charoenwong, Charlie: Investment Options, 59–72.
Assets in Place, and the Risk of Stocks.Financial Management20
Mallette, Paul, and Fowler, Karen L.: Effects of Board Composition (Autumn 1991): 21–33.
and Stock Ownership on the Adoption of “Poison Pills.”Academy Cotter, James F., Shivdasani, Anil and Zenner, Marc: Do Independent of Management Journal35 (December 1992): 1010–1035.
Directors Enhance Target Shareholder Wealth During Tender
Myers, Stewart C.: Determinants of Corporate Borrowing.Journal of Offers?Journal of Financial Economics43 (1997): 195–218.
Financial Economics5 (November 1977): 147–175. Donaldson, Lex, and Davis, James H.: Stewardship Theory or Agency
Patton, Arch, and Baker, John C.: Why Don’t Directors Rock the Boat? Theory: CEO Governance and Shareholder Returns.Australian
Harvard Business Review (November–December 1987): 10–18. Journal of Management16 (June 1991): 49–64.
Pfeffer, Jeffrey: Size, Composition, and Function of Hospital Boards Fama, Eugene F.: Agency Problems and the Theory of the Firm.
of Directors: A Study of Organization–Environment Linkage. Ad-Journal of Political Economy88 (April 1980): 288–307.
ministrative Science Quarterly18 (September 1973): 349–364. Fama, Eugene F., and Jensen, Michael C.: Separation of Ownership
Pincus, Karen, Rusbarsky, Mark, and Wong, Jilnaught: Voluntary and Control.Journal of Law and Economics27 (June 1983): 301–
Formation of Corporate Audit Committees among NASDAQ 325.
Firms.Journal of Accounting and Public Policy 8 (Winter 1989): Finkelstein, Sydney and D’Aveni, Richard A.: CEO Duality as a 239–265.
Double-Edged Sword: How Boards of Directors Balance
Entrench-Rechner, Paula L.: Corporate Governance: Fact or Fiction?Business ment Avoidance and Unity of Command.Academy of Management
Horizons(July 1989): 11–15. Journal37 (October 1994): 1079–1108.
Rechner, Paula L., and Dalton, Dan R.: CEO Duality and Organiza-Gaver, Jennifer J., and Organiza-Gaver, Kenneth M.: Additional Evidence on
tional Performance: A Longitudinal Study.Strategic Management the Association between the Investment Opportunity Set and
Journal12 (February 1991): 155–160. Corporate Financing, Dividend, and Compensation Policies.
Substitu-tion Effects of Alternative Governance Mechanisms.Strategic Man- Smith, Clifford W., and Watts, Ross L.: The Investment Opportunity Set and Corporate Financing, Dividend and Compensation Poli-agement Journal16 (February 1995): 85–99.
cies.Journal of Financial Economics32 (December 1992): 263–292. Rosenstein, Stuart, and Wyatt, Jeffrey G.: Outside Directors, Board
Independence, and Shareholder Wealth.Journal of Financial Eco- Wall, Francis J.: Statistical Data Analysis Handbook, McGraw Hill, nomics26 (August 1990): 175–191. New York. 1986, p. 145.
Saunders, Anthony: Why are So Many New Stock Issues Un- Weisbach, Michael S.: Outside Directors and CEO Turnover.Journal derpriced?Business Review(Federal Reserve Bank of Philadelphia) of Financial Economics20 (January/March 1988): 431–460. (Mar/Apr 1990): 3–12.
Whittred, Greg, and Zimmer, Ian.:Financial Accounting: Incentive Simunic, Dan A., and Stein, Michael T.: Product Differentiation in Effects and Economic Consequences, 3rd ed., Holt, Rinehart, and
Auditing: Auditor Choice in the Market for Unseasoned New Winston, Sydney. 1992, pp. 29–30. Issues.CGA Research Foundation Monograph No. 13, Vancouver:
Wolfson, Nicholas:The Modern Corporation: Free Markets vs. Regula-Canadian Certified General Accountants’ Research Foundation.
tion, The Free Press, New York. 1984, pp. 94–97. 1987.
Zahra, Shaker A., and Pearce II, John A.: Board of Directors and Singh, Harbir, and Harianto, Farid: Management-Board
Relation-Corporate Financial Performance: A Review and Integrative ships, Takeover Risk, and the Adoption of Golden Parachutes.
