Firm Investment and Stakeholder Choices: A Top‐Down Theory of Capital Budgeting

• Published date: 01 October 2017
• Date: 01 October 2017
• DOI: http://doi.org/10.1111/jofi.12526
• Author: SHERIDAN TITMAN,ANDRES ALMAZAN,ZHAOHUI CHEN

THE JOURNAL OF FINANCE •VOL. LXXII, NO. 5 •OCTOBER 2017

Firm Investment and Stakeholder Choices: A

Top-Down Theory of Capital Budgeting

ANDRES ALMAZAN, ZHAOHUI CHEN, and SHERIDAN TITMAN∗

ABSTRACT

This paper develops a top-down model of capital budgeting in which privately in-

formed executives make investment choices that convey information to the firm’s

stakeholders (e.g., employees). Favorable information in this setting encourages

stakeholders to take actions that positively contribute to the firm’s success (e.g.,

employees work harder). Within this framework we examine how firms may distort

their investment choices to influence the information conveyed to stakeholders and

show that investment rigidities and overinvestment can arise as optimal investment

distortions. We also examine investment distortions in multi-divisional firms and

compare such distortions to those in single-division firms.

THIS PAPER DEVELOPS A top-down model of capital allocation in which privately

informed executives make investment choices that convey information to the

firm’s stakeholders. The information conveyed by the executives’ investment

choices is relevant because it influences the actions of the stakeholders, which

in turn affect the likelihood of the firm’s success. Specifically, since higher in-

vestment expenditures are associated with better firm prospects, stakeholders,

who receive a share of the output, are willing to expend more effort to improve

the probability that the firm will be successful. Within this setting, we examine

the optimal design of capital allocation mechanisms. We show that the process

used to make investment choices affects how the information conveyed by in-

vestment choices is interpreted by stakeholders, which in turn influences how

the capital budgeting process is designed.

∗Almazan is at the University of Texas at Austin, Chen is at the University of Virginia, and

Titman is at the University of Texas at Austin and NBER. For their comments and sugges-

tions we thank Bruno Biais (the Editor), Matthieu Bouvard, Adolfo de Motta, David Dicks, Alex

Edmans, Chitru Fernando, Paolo Fulghieri, John Hatfield, Neal Galpin, and Maxwell Stinchcombe

We are also grateful for comments from referees as well as seminar participants at AFA-San

Diego (2013), Australian National University, Cass Business School, CEMFI, DePaul University,

UCLA, Universidad Carlos III, Chinese University of Hong Kong, University of Essex, Univer-

sity of Melbourne, Nanyang Technological University, National University of Singapore, UNC-

Chapel Hill, McGill University, University of Notre Dame, University of Oklahoma, University of

Pennsylvania, University of Virginia, University of Texas at Austin, University of Texas at San

Antonio, the XVIII Finance Forum at CEU (Elche-Spain), and the Finance Down Under 2011 Con-

ference at the University of Melbourne. Zhaouhi Chen acknowledges financial support from the

McIntire Foundation’s King Fund for Excellence and the Walker Fund. The authors do not have

any potential conflicts of interest to disclose, as identified in Journal of Finance Disclosure Policy.

DOI: 10.1111/jofi.12526

2179

2180 The Journal of Finance R

Toformalize these ideas, our model considers a firm whose production process

requires a capital investment as well as effort exerted by a stakeholder (also

referred to as the worker). In particular, we consider a framework in which

the firm’s owner (also referred to as the executive) obtains private information

about the firm’s prospects, which can be high or low, and chooses both the level

of investment and the worker’s compensation. In our framework, the worker’s

effort is more productive when the future prospects of the firm are better,which,

in combination with the optimal compensation contract, induces the worker to

exert more effort when his beliefs about the firm’s future prospects are more

favorable. As a result, the owner has an incentive to overinvest relative to the

case in which he has no private information, because doing so conveys favorable

information.

In this setting, there is a trade-off between the benefits of using the exec-

utive’s private information to adjust its capital allocation and the inherent

tendency to overinvest that arises from such discretionary capital allocation

choices. As a result of this trade-off, two alternative capital allocation mech-

anisms emerge as potentially optimal: (i) a separating mechanism in which a

high-prospect firm invests more (and offers higher compensation to the worker)

than a low-prospect firm, and (ii) a pooling mechanism in which firms commit

to a fixed level of investment and compensation. Under the separating mecha-

nism, the firm invests more when the marginal productivity of capital is higher

but tends to overinvest because of the potential benefits associated with convey-

ing favorable information. Under the pooling mechanism, investment rigidities

arise, that is, the level of investment is independent of the executive’s informa-

tion, and the firm’s overinvestment tendency is suppressed. The choice between

the pooling and the separating mechanisms is determined by a trade-off be-

tween the efficiency loss of a rigid investment policy that does not incorporate

information and the cost associated with overinvestment.1

The information and incentive issues explored in this paper also have impli-

cations for how corporations are organized. To investigate this issue, we extend

our analysis to consider the case of a two-division firm with top executives that

have superior information about the prospects of each division. The capital

investment choices of each of the divisions are observable and convey informa-

tion. As we show,the optimal investment policy of a two-division firm generates

(weakly) more value than the optimal investment policies of two independent

firms acting independently.Intuitively, the gain from conglomeration is directly

related to the incentives to distort investment that arise when executives have

private information about the divisions’ prospects. Although executives of a

two-division firm have incentives to exaggerate the prospects of each of the

divisions, they do not have an incentive to misrepresent the divisions’ relative

prospects. As a result, instead of overinvesting, firms can use asymmetric cap-

ital allocations to provide credible information about the relative prospects of

1As discussed in Section V, the analysis suggests that investment rigidities observed in practice

tend to arise when there is limited uncertainty about the firm’s prospects as well as a higher

likelihood of more favorable prospects (e.g., in industries in which growth is more likely).

Firm Investment and Stakeholder Choices 2181

the divisions.2Our analysis therefore provides a novel rationale for conglom-

erates, namely, the reduction in the cost of investment distortions that arise

when a firm’s stakeholders use the firm’s investment choices to make inferences

about the firm’s prospects.

This paper addresses questions that have been previously considered in a

number of threads of the literature. For instance, our analysis is related to

papers in the leadership literature that examine how choices made by lead-

ers influence individuals at lower levels in the organization (e.g., Rotemberg

and Saloner (2000)). In particular, our analysis is closely related to Hermalin

(1998) and Komai, Stegeman, and Hermalin (2007), who show that an in-

formed leader signals his favorable information by expending greater effort,

and in doing so motivates his subordinates to work harder (i.e., leadership

effects). There are key differences, however, between our setting and the set-

ting considered in these leadership papers. First, while these studies focus on

how leadership effects can mitigate undereffort in teams, our analysis focuses

on how the optimal design of the capital investment process can sometimes

limit and other times take advantage of these leadership effects. Moreover, we

consider a setting with contractible investment choices, which contrasts with

leadership models that assume that the leader’s effort is not contractible. In-

deed, we find that a firm may find it optimal to commit to a rigid investment

policy (i.e., a pooling mechanism) that does not fully incorporate top executives’

information.

More broadly, our paper belongs to the principal-agent literature and specif-

ically to the thread that considers the case of an informed principal that faces

an agent with a moral hazard problem.3Among other things, this literature

analyzes the effects of the principal’s information on the optimal compensation

contract (e.g., Beaudry (1994) and Inderst (2001)), the value of private infor-

mation to the principal (e.g., Chade and Silvers (2002)), and the incentives to

disclose information (i.e., provide “advice” ) to the agent (e.g., Strausz (2009)).

While most of these papers (Strausz (2009) is an exception) consider the case

of a principal who makes choices after becoming informed (i.e., signaling), we

consider a mechanism design setting in which the principal commits to a self-

screening mechanism prior to acquiring information and then, after becoming

informed, makes choices that convey information that influences the agent’s

actions.4Within this self-screening paradigm we focus on the case in which

2This result is related to the findings in Chakraborty and Harbaugh (2007), which considers the

case of multi-dimensional cheap talk and illustrates how cheap talk about rankings can sometimes

convey information.

3Maskin and Tirole (1990,1992) study the optimal mechanism in the case of a principal who

proposes the contract after being informed and faces an agent without moral hazard. In contrast,

we consider the case of a principal who becomes informed after proposing the contract and faces

an agent subject to moral hazard in her actions.

4See Biais and Mariotti (2005) for a mechanism design analysis with a similar timing of events,

namely,a commitment by the principal to narrow his choices to a pre-specified set of actions before

receiving his private information. In Section II.B, we discuss the realism and implications of our

timing assumption further.