Title: Centralization of DecisionMaking in the Top Management Team
1Centralization of Decision-Makingin the Top
Management Team
- Heitor Almeida
- New Ideas and Open Issues in Corporate Finance
- Amsterdam, June 15, 2003
2Two Notions of Centralization
Top Management Team (CEO)
Division Managers
- Where in organization should decisions be made
- Centralization of decisions at a given level of
the organization - opening up the boxes
3These are Two Different Questions
- Examples
- HP until the 80s decision-making highly
concentrated in the hands of founders, but
company was a loose federation of far flung
divisions each finding its own way to contribute
to the greater good (Anders, 2003) - News Corp.s magic circle (FT May 20, 2003)
office of the chairman is a group of 5 people
which meet every Monday to consider every
acquisition and item of capital expenditure - Augustin will talk about 1), I will talk about 2)
4The Role of Top Managers
- Focus on decision-making at the top level of the
organization - Why is this an important/relevant question for a
presentation on centralization? - Decision-making power could have been sent down
the hierarchy - Isnt the CEO alone at the top of the hierarchy?
5Do Top Managers Matter?
- Question has a long intellectual history in
management - No Lieberson and O'Connor (1972), Hannan and
Freeman (1977) - Yes Weiner and Mahoney (1981), Hambrick and
Mason (1984), Finkelstein and Hambrick (1996) - More recent finance and economics literature
arguing that managers matter - Malmendier and Tate (2001), Bertrand and Schoar
(2002), Adams et. al (2002)
6Is the CEO alone at the top?
- No, or at least not always (Top management team)
- FH (1996) loosely define TMT as top 3-10
executives in the organization such as CEO,
Chairman, President, COO, Senior Vice-President,
Executive Vice-President, etc. - TMT also interacts with the board, and with
markets (at least when company needs funds) - Behavioral theory of senior leadership must
extend beyond considering executives in isolation - Decision-makers are informed, influenced, and
sometimes constrained by others, both inside and
outside the organization (FH 1996)
7A View from the Top
represents an exchange of opinion about decision
TMT
CEO
Other managers
board
markets
proposed decision
final decision
markets opinion carries more weight when company
needs to raise funds - implications for financial
policy
8A Key Ingredient - Differences of Opinion
- Justifications for disagreement
- costly communication
- differences in ability to process information
- differences in priors
- managers are overconfident, or they have vision
- Disagreement especially likely for types of
decisions made by TMT - weak situation (Mischel, 1977)
- for many industries in modern economies there is
uncertainty about logical implications of
information and actions (Allen, 1993, Radner,
1997)
9Disagreement Implies Fallibility
- When people disagree, some (or all) should be
wrong, even when they have common goals (no
agency problem required) - ... biases, egos, aptitudes, experiences...greatl
y affect what happens to companies. This is not
to say that managers are weak and sinister, only
that they are human and limited. (FH, 1996) - With disagreement, issue is not how to give
shareholders control, but who should be given
control (Allen and Gale, 2000)
10Roadmap for Discussion
- Simple model of decision-making with disagreement
based on Sah and Stiglitz (1985, 1986, 1988) - Introduce effect of groups and group interaction
- Applications
- Interactions within TMT and optimal design and
distribution of power in the TMT - Role of corporate boards in decision-making
- Potential implications for behavioral corporate
finance - Managers vs. markets implications for financial
policy - Caveats
- Many more questions than answers
- I will use management and social psychology ideas
that are not formally modeled
11Simple Model Based on Sah and Stiglitz
- CEO and other top managers face decision of
whether or not to implement a given project - Good projects have NPV of G, bad projects have
NPV of -B. Fraction ? of projects are good - Fallible decision-making
- CEO judges a good project to be good with
probability pGCEO lt 1, and he judges a bad
project to be good with probability pBCEO gt 0 - Expected output when CEO is deciding alone
- YCEO ? pGCEO G - (1 - ? ) pBCEO B
- CEO makes two types of mistakes
- Type I reject good project
- Type II accept bad project
12Introducing a Second Member in the TMT
- Second member is subject to similar bias as CEO
(pGM lt 1, pBM gt 0 ) - Disagreement second manager might have a
different opinion about the project (even after
communicating with the CEO) - opinions are not perfectly correlated. Assume
they are uncorrelated - SS never modeled communication explicitly. They
assume managers only observe whether the other
wants to accept or reject, and do not update - How will group decision-making affect the final
decision? - Suppose that both the CEO and the manager have to
accept the project in order for it to be
implemented - This will lead to the following expected output
- YTMT ? pGCEO pGM G - (1 - ? ) pBCEO pBM B
- Group will decrease probability of type II but
increase probability of type I error
13Optimal Distribution of Power in the TMT
- Decentralizing decision-making power pays off
when - CEO is more fallible (low pGCEO, high pBCEO )
- Manager is less fallible (high pGM, low pBM )
- Proportion of good projects is low (low ? )
- Cost of accepting bad projects is high relative
to benefit of accepting good ones (high B, low G) - Notice that when decentralization pays off, it
does so because of disagreement - No benefit when opinions are perfectly correlated
- It may be important to incorporate opinions of
both managers in the final decision - See Sah (JEP, 1991) for a summary of other
implications of the Sah and Stiglitzs framework
14The Role of TMT Heterogeneity
- Management literature has also argued for need to
have diverse top management team in some
circumstances - Focus on demographic heterogeneity, i.e.
dispersion in age, tenure, educational and
functional background - Basic idea
- Firms in complex and unstable environments
require more diverse TMT - Benefits of broader range of skills, diverse
interpretations, more debate and questioning,
i.e. promotion of disagreement - In simpler environments heterogeneity may be
dysfunctional because it promotes conflict and
hinders communication - This idea might already be in the SS model
- Complex environments might be those where CEO
is more likely to make mistakes, and where cost
of type II errors is high - But I think they are talking about real exchange
of information
15The Role of Group Interaction
- Decision-making group in Sah and Stiglitz is not
a truly interacting group - Nothing is really exchanged, or at least the
exchange of information is not modeled - TMT heterogeneity idea suggests that group
members with diverse perspectives can influence
each other and mitigate all types of individual
biases - In this case a complex environment can simply
be one where fallibility is more important - Large social psychology literature characterizing
exchange of information in groups, and processes
by which group members adjust priors following
group discussion - Groupthink (Janis, 1972)
- Group polarization (Moscovici and Zavalloni, 1969)
16Groupthink
- ...dysfunctional mode of group decision-making
characterized by a relentless striving for
unanimity, resulting in a reduction in
independent critical thinking (Forbes and
Milliken, 1999) - Fewer options are considered in the discussion
preceding a decision because members seek
concurrence (they dislike disagreement) - In the model, manager M says that she always
agrees with the CEO even when she does not - Groupthink reduces benefit of decentralization
when it is optimal - Janis (1972) used this notion to try to
understand some famous policy fiascoes such as
invasion of Bay of Pigs by US in 1961 (a perfect
failure) - GT could be important in TMTs, but I am unaware
of any empirical work exploring that
17Group Polarization
- Tendency for group judgment to deviate
significantly from average of the pre-discussion
individual judgements, in the direction of higher
extremity - Extremity of a judgment is a measure of its
deviation from some neutral point in a scale,
such as not agreeing nor disagreeing with a
particular statement - Evidence that group consensus is more distant
from the neutral point of the scale than the
preconsensus mean - Individuals seem to change their priors in the
direction of higher extremity - Unlike GT, it is unclear whether this is a bad
thing - Experimental evidence based on laboratory
experiments with college students in artificial
contexts - unclear whether this is relevant for
top executives
18Economics Literature on Communication and
Exchange of Information
- In my view, what is missing from the SS model is
an explicit modeling of communication and
exchange of information among group members - Effectiveness of communication will affect
optimal structure of decision-making - Model could use some of the management and social
psychology ideas that I presented for motivation - Economics literature on communication
- Radner (1992), Bolton and Dewatripont (1994) -
not very relevant for problems proposed here (I
think). More about how to process a given batch
of information in the fastest and cheapest way,
than about minimizing costs of decision-making
errors - Krishna and Morgan (2001), Dessein (2002), Harris
and Raviv (2002) - focus on quality of decision
and allocation of decision-making authority under
cheap talk
19Cheap Talk Models
- Models based on Crawford and Sobel (1982)
- Strategic communication of information by an
expert to a decision-maker - Decision-maker makes decision that affects
utility of both, but the expert has a taste for
size (empire builder) - This agency problem prevents full communication
and decreases quality of decision - Krishna and Morgan (2001) consider multiple
biased experts - Dessein (2002) considers possibility of
delegating decision to expert (this turns out to
be always better than communicating information
to top level decision-maker) - Harris and Raviv (2002) consider possibility that
top level decision-maker also has private
information
20Cheap Talk Models and Top Management Teams
- These papers are more about structure of
hierarchies, than about distribution of
decision-making power in a given level of the
hierarchy - Usually agent (division manager) has more
information (with the exception of Harris and
Raviv) but is biased towards larger investments - In the TMT
- CEO also has information (perhaps he has the best
information) - If there are agency problems these should be
similar for all top managers - Still, it is possible that managers behave
strategically when they have biased opinions (if
they are aware of other peoples biases) - i.e, they might exaggerate opinions when they
interact in a group (group polarization?)
21The Role of Boards
- The established view in corporate finance
- Boards are an economic institution that in
theory helps to solve the agency problems
inherent in managing an organization (Hermalin
and Weisbachs survey, 2002) - Problems
- Weak link between board structure and performance
(i.e. insiders vs. outsiders, CEO Chairman not
robustly related to performance) - Board composition seems inconsistent with pure
agency view - CEOs make up largest single slice of outside
directors among the SP500 (20) - Very few institutional investors, or other direct
representatives of SH - SH have very little influence over who fills
board seats
22What do Boards Really Do?
- The cynical view - nothing
- Boards dont matter because they are captured by
the CEO - Boards are just a product of regulation
- Outside directors are carefully chosen so as to
be overcommited (thus CEOs) - Is there an alternative view?
- Perhaps boards are an additional tool to help
the CEOs decision-making - a supra top
management team (FH, 1996) - Example of an active board doing more than just
monitoring on behalf of SH - HPs board after Fiorina took over (Anders, 2003)
23Ideas From Management Literature (FH, 1996)
- External and internal roles for board
- external linking organization to critical
resources/information that reside in the outside
environment - internal setting strategy and monitoring
management - Limited empirical evidence that boards are
systematically involved in setting strategy, and
conditions that make this more/less likely - Boards more likely to be advising/evaluating than
forming strategy - Top managers more likely to have influence on
CEO, and thus literature has focused more on them
24Interactions Between Monitoring and Advising
- Key difference between boards and other top
managers is that boards have (formal) authority
over CEO and other managers - Boards can fire CEO/other top managers
- Hiring/firing might be the key monitoring role of
board - Interesting effects to model - complementarities
and trade-offs between monitoring (firing/hiring)
and advising (direct participation in decision
making) by the board - Such a model might help uncover roles that are
particular to corporate boards
25Adams and Ferreira (2003)
- Key idea information between CEO and board flows
both ways - Board needs information from CEO to perform
advising role. Advising is a suggestion to accept
or reject a project (like in SS) - But information provided by the CEO also allows
board to update prior about CEOs ability - CEO cares about being fired and thus may not
provide information if board uses the information
to update - Main result is that board may commit not to
monitor in order to encourage the CEO to share
information - A theory of friendly boards. In equilibrium
boards may participate in decision-making but not
monitor much
26Some Ideas for Behavioral Corporate Finance
- Behavioral corporate finance is about the impact
of managerial biases on firm policies and
performance (i.e., Malmendier and Tate, 2001) - Managerial discretion idea (FH, 1996)
- ...discretion attenuates the relationship
between executive characteristics (values,
experiences and so on) and organizational
outcomes. Namely, if high discretion exists,
executive orientations become reflected in
organizational outcomes if low discretion
exists, they do not. - Discretion is related to
- Distribution of power in the TMT, and between the
TMT and the board - Environmental factors such as industry and
financial slack
27Power Levers Up Overconfidence
- In the SS model, we can represent an
overconfident (or more precisely, an optimist)
CEO as one who holds a stronger belief that the
project is good - Optimist CEO judges a good project to be good
with probability qG gt pGCEO, and he judges a bad
project to be good with probability qB gt pBCEO - The optimist CEO will clearly overinvest relative
to the rational CEO (Malmendier and Tate) - However, tendency to overinvest will also depend
on the distribution of power in the TMT - Optimist deciding alone invests with probability
?qG (1 - ?) qB x - Optimist deciding in group invests with prob. ?qG
pGM (1 - ?)qB pBM lt x
28Optimal Bias and Centralization
- It might be optimal to have an optimist CEO
- Even in basic SS model, rational CEO makes more
type I errors (rejects good projects more often) - Optimality of bias even more likely if we
introduce motivating effects of optimism
(vision), as in Rotemberg and Saloner (2000) and
Van den Steen (2001) - If it is optimal to have a visionary CEO it might
also be optimal to centralize top-level decision
making (within TMT, and perhaps also weak boards) - No dilution of vision with decentralized
decision-making - Example CEO Chairman is required to ensure
existence of illusion of strong leadership in a
firm (Finkelstein and DAveni, 1994)
29Empirical Measures of CEO Power and Discretion
- It is possible to construct proxies for the
distribution of power in the TMT, and between TMT
and boards (Adams et al. 2002, Haleblian and
Finkelstein, 1993) - Formal titles of CEO (President, Chairman, etc.)
- Founder status
- TMT executives membership on the board
- Number of boards where executive is member
- Compensation ratios in the TMT
- It is also possible to construct environmental
measures of managerial discretion based on
industry (Adams et al, 2002, Hambrick and
Abrahamson, 1995) - These measures can help sharpen tests of
behavioral hypotheses!
30Managers and Markets
- Compared to top managers and boards, outside
investors are farther away from corporate
decision-making - And boards do not seem to represent their
interests that much - Still, outside investors will have their own
opinions about optimal actions and consequences
of actions taken by management - Ex HP Compaq merger
31Disagreement Between Managers and Markets
- What is the nature of this disagreement?
- Allen (1993) because many investors are
computing values, on average mistakes cancel out
and markets end up more precise than managers - Allen and Gale (1999) markets may allow for
matching between opinions of investors and
managers - Behavioral finance sometimes assumes that markets
are right and managers are biased (i.e.
Malmendier and Tate), sometimes that managers are
right and markets are wrong (i.e. Baker and
Wurgler) - Given possibility of disagreement, question is
when will outside investors have discretion to
influence firm policies
32Disagreement and Financial Policy
- The most obvious effect
- Outside investors will have more discretion when
firms need to raise funds - One of the primary purposes of corporate
financial policies might be to determine the
extent to which firms (managers) can rely on
their own perception of the world (Allen, 1993) - Theories based on disagreement and discretion
might help explain some of the main facts of
corporate finance - Firms dislike external finance
- Financial slack matters for investment and
financial policies
33Why not Focus on Cash Policy ?
- There are already some papers relating
disagreement and discretion (flexibility) to
financial policy - i.e., Boot and Thakor, 2002
- Focus is on effects of disagreement and
flexibility on optimal security design and
issuance - This might be a secondary concern cash (internal
finance) gives you maximum flexibility. The main
issue might be how to avoid raising external
finance - A framework to think about optimal cash policies
Almeida et al, 2003 - The empirical challenge will be how to
differentiate such a theory from existing
arguments - Financial constraints, agency
34Conclusions
- Behavioral theory of senior leadership should
incorporate role of other constituencies that
might have discretion to influence top-level
decision making - Top management teams, boards and markets
- Questions
- When do they have discretion?
- What is the optimal level of discretion?
- What is their effect on the quality of corporate
decisions and corporate performance? - These research questions might generate insights
about important unsolved issues in corporate
finance, such as the role of boards and optimal
financial policy