Julio Rotemberg.

Leadership style and incentives online

. (page 1 of 3)
Online LibraryJulio RotembergLeadership style and incentives → online text (page 1 of 3)
Font size
QR-code for this ebook





u >

no. 32.55-



Leadership Style and Incentives

Julio J. Rotemberg
Garth Saloner

WP# 3250-91-EFA

September 1990





Leadership Style and Incentives

Julio J. Rotemberg
Garth Saloner

WP# 3250-91-EFA September 1990

© 1991 Massachusetts Institute of Technology

Sloan School of Management

Massachusetts Institute of Technology

50 Memorial Drive

Cambridge, MA 02139

Acknowled gement

The authors are respectively Professor of Applied Economics at MITs Sloan School
of Management and Professor of Strategic Management and Economics and BP
Amenca Faculty Fellow at Stanford University's Graduate School of Business. We
wish to thank Tom Allen for help as well as the National Science Foundation and
the International Financial Services Research Center at MIT for support.


We study the relationship between a firm's environment and its optimal leadership style.
We present conditions under which firms benefit from having autocratic leadership and
conditions under which they benefit from a more participatory style. We use an economic
model in which contracts between the firm (or CEO) and managers are incomplete so
that providing incentives to subordinates is not straightforward. CEO's with different
preferences (or personalities) adopt different leadership styles and these styles in turn
have an effect on the incentive contracts that can be offered to subordinates. We show
that shareholders gain from appointing a CEO who adopts a participatory style when the
firm has the potential for exploiting numerous innovative ideas. By contrast, when the
environment is poor in new ideas, shareholders benefit from hiring a CEO whose style is
more autocratic.

Students of business organizations have long recognized that the heads of different companies
exercise their authority in different ways. Some leaders are quite autocratic; they seek and receive
only minimal advice from their subordinates. Other leaders are more democratic and seek consensus
within their organizations. Some chief executive officers issue directives concerning minute details
of operation. Others suggest only broad principles and give considerable autonomy to those below

In this paper we provide an economic model in which leadership style has an important effect
on firm profitability. We show that senior management's style can alter the incentives that can be
provided for subordinates to ferret out profitable opportunities for the firm. The resulting theory
has predictions for the circumstances where shareholders benefit from having either autocratic or
democratic CEO's.

We consider a setting in which firms undertake investment projects which are sequential in
that the firm has the option of shutting down projects mid-way. This is appropriate to a wide
variety of business settings: Almost any business venture goes through a variety of stages from
conceptualization to eventual adoption and implementation, and a venture can be cut short at any
stage by withdrawing the needed additional financing.

As the literature on sequential investment initiated by Roberts and Weitzman (1981) has
shown, in such a setting the optimal investment strategy for the firm involves experimenting with
the early stages of many projects. Indeed the firm should initiate some projects for which the
expected cost if the project if funded to completion exceeds the corresponding expected revenue.
The reason is that the option to shut down projects is valuable.

The challenge to senior management in such a setting is to provide the appropriate incentives
for their subordinates to think up and undertake these potentially profitable projects. One simple
incentive scheme that the firm can offer is to reward employees whose projects are carried through
to fruition. Here, however, the valuable option that the firm has to close projects down mid-way
interferes with its ability to provide the appropriate incentives to its managers. In particular, the
knowledge that the senior management may abort their projects mid-way reduces the incentive for
the subordinates to generate potentially profitable ventures in the first place.

Since the leadership style of the senior management is likely to affect how decisions are made
about whether to continue projects already underway, it also affects the ability of the firm to provide
incentives for its workers. We investigate the effect that autocratic and democratic leadership styles
have on the ability of the firm to provide incentives and study the circumstances in which different
leadership styles are optimal.

Our main finding is that in environments which are rich in potential new ideas which the firm
could exploit - and where it is correspondingly more important to provide incentives to employees
to ferret those ideas out - the democratic leadership style is most attractive. On the other hand,
when the environment is less rich in new ideas, autocratic management may be more profitable.

The literature on leadership styles has classified these styles in numerous related ways. Robert
Tannenbaum and Warren H. Schmidt (1958), see leadership behavior as falling in a continuum be-
tween extreme "boss- centered" leadership and extreme "subordinate-centered" leadership. While
the former managers make decisions on their own and announce them the latter insist that each
subordinate be allowed to make decisions within broad limits imposed by his/her superior.

Likert (1967) classifies organizations as adopting 4 "systems", labelled 1 through 4. These
systems differ along a variety of characteristics. System 1 is more autocratic in that subordinates
are consulted less than systems 2, 3 and 4 with system 4 being the one with the most subordinate
participation. Managers are increasingly friendlier (and subordinates grow less afraid of their
superiors) as one goes from system 1 to 4.

Without offering as complete a classification, Pascale and Athos (1981) stress similar contrasts.
They compare Harold S. Geneen who managed ITT from 1960 to 1979 to the management of
Matsushita. Geneen emphasized "unshakeable facts" and obtained these "unshakeable facts" by

promoting confrontation between line managers and those in charge of staff functions. Geneen is
famous for his pressure-cooker meetings where managers has to defend their results in the face ot
aggressive questioning by Geneen and other managers. To some extent, these meetings provided
subordinates with the ability to express their opinions. However, the unshakeable facts were often
then used to remove employees whose performance was less than wholly satisfactory.

The management style at Matsushita was much more participatory. Decision making at Mat-
sushita involved the seeking of consensus and not the unilateral decisions from the CEO. In addition,
and perhaps as a result, managerial turnover was much lower since that Japanese company prac-
tices lifetime employment. Finally, the company spent considerably more resources developing and
training its employees.

The early students of leadership style such as Likert and Tannenbaum and Schmidt suggest
that the more participatory leadership style is always better. The alternative view that the optimal
style depends on the environment has gained wide currency since the work of Paul R. Lawrence
and Jay W. Lorsch (1967) and Charles Perrow (1970). These authors focus mostly on the effect of
the predictability of tasks. Lawrence and Lorsch put particular stress on the fact that production
divisions face relatively predictable tasks while research divisions do not with sales divisions falling
some place in the middle. They find that those divisions where tasks are more predictable differ
in three respects. The first is that the structure is more formal, there are more rules. The second
is that the planning horizons are shorter and the final one is that people are more concerned with
getting tasks done than with their interpersonal relations.

Perrow (1970) too focuses on the effect of the extent to which tasks are routine. He argues
that the ability to routinize tasks is desirable for firms (because it stabilizes their earnings) and
that, when such routinization is possible, bureaucratization ensues as well. When it is impossible,
the firm's style is by necessity more participatory.

One issue that is left somewhat open by this research is whether the management style depends
on the personality of the leader or whether any CEO can pick any leadership style. In this paper we
assume the former. This is supported by the observations of Desmond Graves (1986) who reports
the following example (p. 123): "A likable but indecisive leader, who coordinated but did not
interfere with the efforts of able subordinates, provided the culture of expansionism necessary for a

business to make its mark in the marketplace. He was succeeded by one of those subordinates who
consolidated the position and quadrupled profitability of the organization in six years..." These
characteristics of the leader then permeate the whole organizations' culture, i.e., its basic beliefs
about what behavior is appropriate and what behavior is deviant.

Fiedler (1965) also gives great weight to the psychological characteristics of the leader. He
shows that leaders who have a more participatitive style also have higher esteem for their co-
workers. He regards esteem for co-workers as a personality trait. He concludes that, because
different leadership styles are successful in different settings, firms ought to place managers using
information on the managers' tendency to esteem their co-workers. While he also studies the effect
of the groups' task on the ideal personality of the leader, Fiedler emphasizes that the leader's ideal
type depends on the psychological relationship between the leader and his subordinates (i.e., on
whether the subordinates like the leader).

Our paper is also closely related to the work based on Coase (1937) which argues that firms
arise as a partial solution to the intrinsic incompleteness of contracts between parties. 1 Firms are
seen as creating rights of control and those who are given these rights make those decisions that
cannot be contractually stipulated. This solution is generally only partial in that control structures
are generally not sufficient to obtain the outcome that is possible with complete contracts. A better
outcome would be possible if the universe of available contracts were expanded. For this reason
firms do better when, as is proposed by Shleifer and Summers (1987), they hire "fair-minded"
CEO's who can be trusted to follow through on implicit promises. Shleifer and Summers' (1987)
proposal is similar to ours in that personal characteristics of the CEO matter because contracts
are incomplete. The main difference is that "fair-mindedness" is a desirable characteristic for all
CEO's whereas we focus on personal characteristics that are appropriate in some contexts and
inappropriate in others.

The next section presents our basic model of sequential research and implementation. Section
2 analyzes the model and compares the outcome under two extreme types of CEO. Section 3
formulates and solves the problem of maximizing the fit between the CEO personality and the
environment of the enterprise. Section 4 concludes the paper.

1 See Holm»trom and Tirole (1989) for a survey

1. The Model

The situation we examine is one in which a manager can expend effort developing an idea which,
if successfully implemented, might improve firm profitability. This effort might entail the search
for an improvement in either product design or delivery to the final customer, the investigation
of a method to reduce costs, or the development of a new product. Firms obviously differ in
their potential for undertaking such profit-enhancing ventures so that only some managers are in
a position to pursue them. For example, in mature industries with stable markets and established
technologies such opportunities are likely to be rarer than in emerging industries.

We have in mind product enhancements that go through two stages. In the first, the manager
invests a great deal of personal time and effort in researching the profit-enhancing idea and devel-
oping a proposal for its implementation. The second stage consists of the implementation itself,
which may be carried out by the manager or by others. Importantly, however, the final decision as
to whether or not to implement the project is in the hands of a more senior manager whom we call
the CEO.

The fruits of the manager's efforts are assumed to be stochastic. That is, the potential profit
to be reaped by the firm if it eventually implements a manager's project is uncertain at the time
that the manager must decide whether or not to put his effort into the project. While some projects
will look better than others, the exact profitability of the venture is unknown until the project has
been researched by the manager.

1.1. Notation and Timing

It is useful to think of events as unfolding over three periods. Managerial effort to develop the
idea takes place in the first period. In the second period, the idea is adopted (or implemented). This
implementation usually requires that the firm spend additional resources on the project. Finally,
in the third period, the implemented project bears fruit. The random variable G denotes the profit
of an implemented project from the second period on.

Ignoring discounting, G equals the increased revenue (or reduced cost) in the third period
minus the implementation costs incurred in period two. We let the realization G of the random
variable G have a cumulative distribution function F(G) and corresponding density function f{G).
It is important to stress that the realizations of G can be negative. A negative G simply means


that the costs of implementing the idea exceed the benefits. The less "rich" the environment in
terms of the profit opportunities it presents, the higher the likelihood that G is negative, i.e., the
greater is F(0).

For the project with payoff G to become available, the manager must devote effort c to it
during the first period: e = 1 if the manager does explore the idea (devotes effort) and e = if
he does not. At the end of the first period, if the manager has researched the project (c = 1),
the potential gain from the project G becomes known to both the manager and the CEO. It is not
important for our analysis whether the actual value of G becomes known at the end of the first
period or whether the CEO and manager simply have a more accurate estimate of what G will be.
For simplicity we assume G becomes known, but the results can be derived by having manager and
CEO obtain an estimate, say G e , of what G will be.

The CEO must then decide whether or not to implement the project during the second period.
We denote the implementation decision by /: I = 1 if a project is implemented and / = if it is
not. Finally, at the end of the second period, the profit from any project which is researched and
implemented is earned by the firm.

1.2. Preferences and Profits

We assume that the manager's utility depends only on his effort e and on his income. We
break down the manager's compensation into the wage, w, for a manager who does not have the
opportunity of undertaking a profit-enhancing project and who thus performs only his "usual"
tasks, and the "incentive" payment k which is tied to the effort e. As we discuss more fully below,
we do not let k depend on the actual effort made. It will depend instead on whether the project to
which the employee devoted effort is actually implemented.

As is typical in principal-agent models, we suppose that there is some minimum level of w
below which the manager will refuse to work. This minimum level of w for performing "normal"
tasks can be thought of as being determined in a competitive market for managerial talent.

We assume that the manager's utility is linear in k (so that he is risk-neutral over income
in excess of w) and that exerting the effort e gives him disutility d. 3 Since we are interested in

3 At the cost of complicating the analysis, we could also have let the manager obtain some direct utility from the implemen-
tation of hia idea. The improvement in the manager's prospects in the external labor market following implementation of his
idea could generate some direct utility of this sort. As long as the expected value of direct utility from implementation does
not exceed the disutility of effort d this modification has no substantive effect* on the conclusions.

3 Letting the manager's utility be linear in income beyond w implies a form of risk aversion since the worker does not accept

the change in the manager's utility from undertaking projects, we normalize his utility so that his
utility when he receives u; is zero. Then the manager's expected utility as a function of jfc and c
can simply be written as u(k, e) = E(k) — ed where E(k) is the expectation of it.

The focus of our analysis is the utility function of the CEO. In a traditional microeconomic
model in which the CEO maximizes profits, the CEO would seek to maximize I(G — k). However
we posit instead that the CEO is concerned not only about firm profitability but also about the
well-being of the manager. In particular, we suppose that at the time he makes his implementation
decision, the CEO places weight 9 on profits and weight 1 — 9 on the manager's utility. Thus the
CEO seeks to maximize:

I\{l - 9){G - k) + 9k\. (1)

Note that any effort that might have been expended by the manager prior to the implementation
decision is "water under the bridge" and therefore the disutility of that effort doesn't enter the
CEO's preferences.

The variable 9 is the key variable in our analysis. It can be though of as representing the
"personality type" of the CEO. A CEO with a 9 of is a profit-maximizer: he cares only about
the "bottom line" and not at all about the utility of the manager who works for him. A CEO with
a 9 that is equal to 1, in contrast, cares only about the welfare of his manager and is completely
oblivious to the "bottom line" . Values of 9 between and 1 represent intermediate cases: As 9
increases from to 1 the CEO's concern for profit vis-a-vis his concern for his manager's welfare

We term a CEO for whom 9 = an "autocratic" manager, and one for whom 9 = 1 a
CEO with a "participatory" style. The motivation for this is the following. We think of the
implementation decision generally being taken by the CEO with some degree of consultation with
the manager. If the manager's compensation is higher if the project is implemented, as it often will
be in equilibrium, the manager will have an incentive to try to convince the CEO that the project
should be implemented. The higher is 9, the more the CEO takes the manager's preferences into
account. Hence the term "participatory" to describe a CEO with a high 9 By contrast a CEO with

a wage below tu. This form of risk aversion is not, per ae central to our analysis. What is central is that the employee remain an
employee and not become the owner of the enterprise. In other words, we cannot let the employee become the residual claimant
on all the firm's cash flows. It might be thought that a risk neutral employee would be willing to become the owner. However,
this becomes impossible once it is recognired that a company has many interdependent employees and that they cannot each
become the residual claimant of the entire firm.

a 9 = does not take the interests or concerns of the manager into account and need not consult
with him. Hence the label "autocratic".

In general one might expect that the overall efficiency of the firm would also be affected
by the extent to which the CEO is concerned with his manager's welfare. A CEO who cares
about his manager's welfare will generally indulge his manager's desire for perquisites and for
nonessential equipment and personnel. Since shareholders are probably incapable of stopping all
of these payments, we expect autocratic manager to be more efficient. 4 We denote the increase in
costs over what they would be under profit-maximization by C(9) (so that C(0) = 0). 5

In the "background" are the firm's shareholders who have the authority to hire the CEO.
Since they are removed from the day to day operations of the firm, they are assumed to be solely
interested in profit-maximization and to be unconcerned about the manager's utility as long as he
is compensated sufficiently to induce him to do his job. Thus the CEO is the intermediary between
the two sets of stakeholders, the manager and the shareholders, and his preferences play a role in
determining the sharing of the profits between them.

1.3. Informational Assumptions and Contractibility

An important distinction in our model is whether or not the CEO is able to base the manager's
compensation directly on the manager's effort. Consistent with the contracting literature 6 we term
the case where the CEO can do so, the "complete contracts" case. One simple condition that makes
complete contracts possible is that e be verifiable. It turns out, however, that the outcome with
complete contracts based on e can be replicated under somewhat weaker conditions. In particular,
it is possible to obtain this outcome as long as G is verifiable.

"Verifiability" means not only that the relevant information is observable by the CEO, but
also that it can be established by the body responsible for enforcing the compensation agreement
between manager and CEO. In an extreme case this enforcement body might be the Courts in
which case verifiability refers to the ability to establish the facts before a judge or jury. More often,
however, the enforcement is performed by other employees. If the CEO is observed to have reneged

4 A countervailing force exi»t» when manager and CEO do not observe the actual value of G in the second period. Then,
a CEO who maximizes profits will tend to spend too much in finding out the true value of G. He may, for example, have to
appoint separate "task force" to perform this function.

6 Little hinges on this assumption, and indeed the opposite assumption (that the participatory manager is more efficient)
simply involves reinterpreting C(6) as a cost saving, rather than a cost increase.

'See Holmstrom and Tirole (1989).


on an implicit agreement with the manager (for example by withholding payment of a "bonus"
when it is understood by the employees that the circumstances warrant a bonus being paid) this
has deleterious effects on his reputation for "fair dealing" . In this context, verifiability refers to
the manager's ability to convince the other employees in the firm that he did indeed carry out the
required effort.

Typically e is not verifiable because it is difficult to distinguish cases in which the manager is
really putting in the necessary effort and when he is simply going through the motions. Indeed, in
the agency literature it is typically assumed that the effort cannot be precisely observed by anyone
other than the manager himself.

The fact that e is not observable does not, in and of itself, prevent a complete contract from
being written in our setting. Indeed a complete contract will still be possible if G is verifiable. Since
the manager is assumed to be risk-neutral, he is indifferent between receiving a direct payment d
for exerting effort and receiving a larger payment (which depends on G) when G turns out to be
positive i.e., only in those cases where he in fact comes up with a profitable proposal. Formally, the
manager is indifferent between receiving d always and receiving a payment 4>G where = d/E(G).
This is because the expected value of G at the beginning of the first period is simply E(G) =
E(kG/E(G)) = d. On average G is a perfect indicator of whether or not the manager exerted the
effort and therefore he is as happy to have his compensation based indirectly on the outcome of his
effort as directly on the effort itself.

In practice, however, even G is likely to be very difficult to verify. Accounting profit figures are
subject to manipulation through the allocation of overhead and other cost items. This manipulation
is extremely costly to detect. Moreover, this form of manipulation is not easy to reduce by making
partial audits and imposing big fines on firms found to have manipulated their books. The reason
is that many of these manipulations are conceivably justified so that it is hard to decide, even ex

1 3

Online LibraryJulio RotembergLeadership style and incentives → online text (page 1 of 3)