ArticlePDF Available

Delegation Within Hierarchies: How Information Processing and Knowledge Characteristics Influence the Allocation of Formal and Real Decision Authority

March 2015
Organization Science 26(3):687-704

DOI:10.1287/orsc.2014.0954

Authors:

Magdalena Dobrajska

University Carlos III de Madrid

Stephan Billinger

University of Southern Denmark

Samina Karim

Northeastern University

We investigate trade-offs associated with delegating authority over multiple interrelated decisions in a complex task structure. The empirical setting is a business process of a global Fortune 50 firm. The firm decentralized its organization and redefined decision authority across organizational hierarchies between 2008 and 2011. We employ regression analysis of microlevel data on the allocation of decision authority between formal and real authority, and further on the organization design of 761 decision tasks within a hierarchy. Our findings show how the specialization of decision-relevant knowledge, the matching of required knowledge and managers’ expertise, and information processing intensity affect (a) the occurrence of delegation and, (b) if delegation occurs, how far down the organizational hierarchy authority is delegated. We discuss how these findings complement existing theories on delegation by providing insights into when and how interrelated decisions are delegated across multiple levels of an organizational hierarchy.

Overview of Variables

…

Correlations

…

Heckman Regression, First Stage: Predictors of a Decision Authority Split Between the FAH and RAH DV: Authority split between FAH and RAH

…

Figures - uploaded by Samina Karim

Content may be subject to copyright.

Content uploaded by Samina Karim

Content may be subject to copyright.

This article was downloaded by: [96.237.155.126] On: 28 January 2015, At: 12:23

Publisher: Institute for Operations Research and the Management Sciences (INFORMS)

INFORMS is located in Maryland, USA

Organization Science

Publication details, including instructions for authors and subscription information:

http://pubsonline.informs.org

Delegation Within Hierarchies: How Information

Processing and Knowledge Characteristics Influence the

Allocation of Formal and Real Decision Authority

Magdalena Dobrajska, Stephan Billinger, Samina Karim

To cite this article:

Magdalena Dobrajska, Stephan Billinger, Samina Karim (2015) Delegation Within Hierarchies: How Information Processing and

Knowledge Characteristics Influence the Allocation of Formal and Real Decision Authority. Organization Science

Published online in Articles in Advance 26 Jan 2015

. http://dx.doi.org/10.1287/orsc.2014.0954

Full terms and conditions of use: http://pubsonline.informs.org/page/terms-and-conditions

This article may be used only for the purposes of research, teaching, and/or private study. Commercial use

or systematic downloading (by robots or other automatic processes) is prohibited without explicit Publisher

approval, unless otherwise noted. For more information, contact permissions@informs.org.

The Publisher does not warrant or guarantee the article’s accuracy, completeness, merchantability, fitness

for a particular purpose, or non-infringement. Descriptions of, or references to, products or publications, or

inclusion of an advertisement in this article, neither constitutes nor implies a guarantee, endorsement, or

support of claims made of that product, publication, or service.

Please scroll down for article—it is on subsequent pages

INFORMS is the largest professional society in the world for professionals in the fields of operations research, management

science, and analytics.

For more information on INFORMS, its publications, membership, or meetings visit http://www.informs.org

Organization Science

Articles in Advance, pp. 1–18

ISSN 1047-7039 (print) ISSN 1526-5455 (online) http://dx.doi.org/10.1287/orsc.2014.0954

Delegation Within Hierarchies: How Information Processing

and Knowledge Characteristics Inﬂuence the Allocation of

Formal and Real Decision Authority

Magdalena Dobrajska

Department of Strategic Management and Globalization, Copenhagen Business School, 2000 Frederiksberg, Denmark,

md.smg@cbs.dk

Stephan Billinger

Department of Marketing and Management, Strategic Organization Design Unit, University of Southern Denmark, 5230 Odense, Denmark,

sbi@sod.dias.sdu.dk

Samina Karim

Strategy and Innovation Department, Boston University School of Management, Boston, Massachusetts 02215,

samina@bu.edu

We investigate trade-offs associated with delegating authority over multiple interrelated decisions in a complex task

structure. The empirical setting is a business process of a global Fortune 50 ﬁrm. The ﬁrm decentralized its orga-

nization and redeﬁned decision authority across organizational hierarchies between 2008 and 2011. We employ regression

analysis of microlevel data on the allocation of decision authority between formal and real authority, and further on the

organization design of 761 decision tasks within a hierarchy. Our ﬁndings show how the specialization of decision-relevant

knowledge, the matching of required knowledge and managers’ expertise, and information processing intensity affect (a) the

occurrence of delegation and, (b) if delegation occurs, how far down the organizational hierarchy authority is delegated. We

discuss how these ﬁndings complement existing theories on delegation by providing insights into when and how interrelated

decisions are delegated across multiple levels of an organizational hierarchy.

Keywords : delegation; decentralization; decision making; formal and real authority; hierarchy; information processing;

knowledge-based view; organization design

History: Published online in Articles in Advance.

Introduction

Delegation serves as an adaptive organizational mech-

anism that is fundamentally rooted in how the organi-

zation distributes decision-making authority and designs

an effective division of labor across hierarchies. In this

paper we use an in-depth study to examine some of

the underlying mechanisms of delegation. Delegation is

one of the key dimensions of organizational design, and

its roots can be found in many classics in organization

theory (Barnard 1938; Simon 1947, 1955; March and

Simon 1958; Cyert and March 1963; Thompson 1967;

Galbraith 1973; Mintzberg 1979). Scholars have dis-

cussed several reasons that explain why ﬁrms rely on

delegation, including the improvement of decision qual-

ity (Jensen and Meckling 1992, Grant 1996), the time-

liness of decisions (Radner 1992, Patacconi 2009), the

economizing of managerial time and attention (Harris

and Raviv 2002, Wernerfelt 2007), the facilitation of

employee initiative (Bénabou and Tirole 2003, Gagné

and Deci 2005, Rantakari 2012), and the boosting of

responsiveness to business environments (Mendelson

2000, Bloom et al. 2010a). Although most of the exist-

ing studies provide conceptual clarity by explaining why

delegation may occur within an organization, they focus

on single instances of delegation. This approach largely

neglects that medium or large organizations, which typ-

ically have complex task structures, rely on multiple,

interdependent occurrences of delegation across several

levels of the organizational hierarchy. In this context,

two fundamental yet previously unanswered questions

emerge: When does delegation occur, and if it occurs,

how far down the hierarchy is authority delegated?

Literature on delegation, often based in micro-

economics, depicts delegation as a trade-off between

improving decision quality by utilizing an individual’s

speciﬁc knowledge and loss of manager’s control over

delegated decisions that results in agency costs (Hayek

1945; Jensen and Meckling 1976, Jensen and Meckling

1992; Holmström 1979). Firms approach this trade-off

by allocating formal authority, i.e., the right to decide,

and real authority, i.e., an effective control over deci-

sions, over particular decision tasks to selected employ-

ees located within the ﬁrm’s hierarchy (Aghion and

Tirole 1997). Such distinction between formal and real

authority is fundamental to understanding delegation, as

having the formal right to decide does not mean that

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

the organization is best served if all decisions are made

by formal authority holders (FAHs) at the top. On one

hand, there is a limit to how many decisions senior

managers can make, and on the other hand, it is not

guaranteed that formal authority holders have sufﬁcient

(expert) knowledge to make effective decisions. Exercis-

ing real authority typically arises from an individual’s

superior expertise, which is made up of specialized and

often tacit knowledge that is difﬁcult and costly to trans-

fer among people (Grant 1996, Aghion and Tirole 1997,

Dessein 2002). An FAH may efﬁciently draw on that

expertise by delegating effective control over a deci-

sion to a real authority holder (RAH) and simply rub-

berstamp this person’s decision proposals (Aghion and

Tirole 1997, Rivkin and Siggelkow 2003). The split of

authority over one decision task between two individu-

als who exercise formal and real authority respectively

results in a hierarchical distance, or delegation span,

separating the FAH and the RAH. Literature points to

opposing effects of such organizational separation. On

the one hand, a large delegation span separating the

FAH and the RAH reduces the efﬁciency of monitor-

ing, communication, and managerial intervention, which

leads to agency costs (Jensen and Meckling 1976, 1992;

Dessein 2002). On the other hand, since a large dele-

gation span typically means that a manager can over-

see several RAHs, she can increase her span of control

(Simon 1947). In this study we analyze these opposing

forces that are created by the delegation span between

the FAH and the RAH.

There is a lack of empirical research on multiple

interrelated instances of delegation in large complex

task structures. Classic contributions to organization

design outline a general logic of delegation based on

descriptive, qualitative empirical accounts (Lawrence

and Lorsch 1967, Galbraith 1973, Mintzberg 1979).

More recent contributions, mainly in organizational eco-

nomics, provide detailed insights into the delegation

mechanisms through stylized formal models of organi-

zations (Jensen and Meckling 1976; Aghion and Tirole

1997; Siggelkow and Levinthal 2003; Rantakari 2011,

2012). Although large Nempirical studies corroborate

the drivers of delegation at a general, ﬁrm level of anal-

ysis (Colombo and Delmastro 2004, 2008; Bloom et al.

2010a), what is missing to date and what we present in

this study is the detailed examination of how individ-

ual ﬁrms design and redesign interdependent delegations

across several levels of an organizational hierarchy.

We empirically study delegation structures in a global

Fortune 50 ﬁrm that designs and delivers wind farms.

The ﬁrm has both onshore and offshore business units

(BUs) that are responsible for the delivery of wind

farms. Both BUs were centralized in 2008, were decen-

tralized in 2010, and had their decentralization efforts

ﬁne-tuned in 2011. We collected longitudinal microlevel

data that allow for an examination of the organiza-

tional design of decentralization in the centralized state

as well as the decentralized state. In both, we examine

the BUs’ major business process and study all FAH–

RAH relationships at the granular level of the indi-

vidual task. In this setting we explore when and how

delegation occurs and which underlying rationale may

explain the actual implementation of delegation. We

address these questions by drawing on the notion of for-

mal and real authority and investigate how specialization

of decision-relevant knowledge—the organization’s abil-

ity to match a manager’s expertise with the knowledge

requirements of a decision—and managerial information

overload affect the split of formal and real authority

between the FAH and the RAH as well as the hierar-

chical distance separating these two individuals in the

case of a split. Our ﬁndings allow for a discussion of

the mechanisms that organizational designers can strate-

gically use when delegating authority and, more gener-

ally, the implications for the organization of multilevel

adaption.

Literature

The Decision Whether to Delegate: Formal

Authority vs. Real Authority in Decentralization

Both the economic and the project management litera-

tures acknowledge that only one person can ultimately

be accountable for each task. The decision authority of

this person involves veto power and the right to over-

rule proposals of other task participants, as well as bear-

ing consequences related to the task’s overall objectives

(McCann and Gilmore 1983, Smith and Erwin 2005,

Kendrick 2006). This function reﬂects formal decision-

making authority (Aghion and Tirole 1997). We refer to

an individual holding such a function for a task-related

decision as the formal authority holder.

A task’s FAH may exercise only formal authority and

intervene in decision making mainly to handle excep-

tions, whereas task-related decisions are de facto taken

by the individual who actually delivers and executes the

task. In this case, this latter individual is drawing on his

or her own knowledge and skills as well as on infor-

mation and contributions from other task participants,

as typically she has access to decision-relevant informa-

tion. It is a common assumption in the literature that

the agent holds decision-relevant knowledge that is supe-

rior to the knowledge of the principal (Dessein 2002).

Such an individual, whom we refer to as the task’s real

authority holder, may exercise real decision authority

(Aghion and Tirole 1997).

Tapping into superior knowledge is one of the key rea-

sons for delegation (Hayek 1945, Jensen and Meckling

1992). However, the decision whether to delegate real

authority and then rubberstamp the agent’s recommen-

dation depends on the cost and effort of the principal to

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

acquire relevant knowledge herself (Aghion and Tirole

1997, Rivkin and Siggelkow 2003). Hence, it can be

the case that the FAH does not delegate away decision

authority. In such a case, the FAH is also the RAH (e.g.,

Colombo and Delmastro 2004, 2008). In the following

we use the term “task decision authority” to refer to both

roles of task FAH and RAH, regardless of whether it is

split or kept uniﬁed within one actor.

Several prior studies adopt a distinction between var-

ious degrees of delegation.1In this paper we adopt the

distinction between formal and real authority over deci-

sions put forth by Aghion and Tirole (1997). Whereas

studies based on formal methods have identiﬁed key

antecedents of delegating real authority as relating

to knowledge asymmetries and communication costs

(Aghion and Tirole 1997, Dessein 2002), empirical lit-

erature (which is scarce on this topic) addresses dif-

ferent drivers of delegation of real authority such as

technology (Collins et al. 1999), ﬁrm- and country-level

management practices (Lincoln et al. 1986, Colombo

and Delmastro 2004, Bloom et al. 2010a), and indus-

try dynamism (Mendelson 2000). It remains empirically

unexplored how knowledge characteristics, knowledge

asymmetry, and information processing intensity among

decision makers affect delegation and its adaptation,

especially in a complex task structure with a system of

interrelated decisions that are taken by decision makers

within a multilevel hierarchy.

Delegation Span: How Far Down the Hierarchy

Should Authority Be Delegated?

Organizational hierarchy serves as a structure that shapes

information transmission and communication between

actors (Thompson 1967, Galbraith 1973, Radner 1992,

Dessein 2002), and the “informational distance” (Foss

et al. 2006, p. 805) between an FAH and an RAH

is an important organization design variable that has

been recognized in the economic literature (Jensen and

Meckling 1992, Aghion and Tirole 1997, Mookherjee

2006, Rantakari 2012). We deﬁne the number of orga-

nizational levels between the FAH and the RAH as

the delegation span. It determines the length of the

path through which information travels between actors

involved in decision making. Delegation span creates

trade-offs for an organization. First, the principal–agent

literature (Holmström 1979) suggests that informational

distance arising from delegation span constrains an FAH’s

possibilities for observing the RAH’s actions. In other

words, it becomes harder for the FAH to exercise judg-

ment over delegated decisions (Foss et al. 2006), since

a large delegation span reduces the closeness of con-

tact between the FAH and the RAH and the amount of

the FAH’s attention over the RAH’s decision (Ouchi and

Dowling 1974). As a result, an RAH that may act in a

self-interested manner increases agency costs (Jensen and

Meckling 1976, Jensen and Meckling 1992).

Second, delegation and the distribution of authority

have important motivational consequences (Fehr et al.

2013). Limiting selective interventions (Williamson

1996) can signal conﬁdence in decisions taken at lower

levels; this, in turn, enhances employee initiative and

effort (Bénabou and Tirole 2003, Gagné and Deci 2005).

As a result, managers can enhance employee initiative

both through reducing managerial involvement in deci-

sion making and through signaling the intention of not

intervening and instead relying on the RAH’s knowl-

edge (Aghion and Tirole 1997, Foss 2003, Rantakari

2012). At the same time, a manager’s interference in del-

egated decisions can make an employee’s effort redun-

dant, which can demotivate the employee (e.g., Aghion

and Tirole 1997).

Hypotheses

Delegation and Information Processing

Centralization can be deﬁned as organization design

where “all the power for decision making rests at a sin-

gle point in the organization—ultimately in the hands of

a single individual” (Mintzberg 1979, p. 181). The extent

to which decision-making authority is concentrated in a

single manager is constrained by several factors. In the

face of bounded rationality, orchestrating the execution

of a complex set of tasks in a ﬁrm generates information

processing needs that exceed the cognitive capacity of

any individual (March and Simon 1958, Mintzberg 1979,

Rivkin and Siggelkow 2003). As managerial attention

is a scarce organizational resource, “wealth of informa-

tion” may lead to “poverty of attention” (Simon 1982),

thereby resulting in information overload and reducing

decision-making effectiveness (Mendelson 2000; Simon

1982; Ocasio 1997, 2011; Gavetti et al. 2007). The nega-

tive consequences of information overload of centralized

decision making have been well recognized in the lit-

erature. Information overload slows down the speed of

organizational decision making, which may result in low

responsiveness to the business environment and inappro-

priate or obsolete decisions (Lawrence and Lorsch 1967,

Thompson 1967, Galbraith 1973, Khandwalla 1973,

Tushman and Nadler 1978, Mintzberg 1979, Radner

1992). The risk of information overload is greater if a

large number of decisions rest within the authority of

a single individual. Increasing the number of decisions

that need to be taken by one individual leads to less

time and attention devoted to each decision (Ouchi and

Dowling 1974, Ocasio 2011). An FAH can reduce the

cognitive burden of decision overload by sharing the

authority with the RAH without the need to fully dele-

gate and give up his decision-making authority.

Hypothesis 1A (H1A). Increasing the number of de-

cisions that a single individual is responsible for is posi-

tively associated with the likelihood of an authority split

between the FAH and the RAH.

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

At the same time, delegation of decision authority

generates agency costs arising from loss of control

(Jensen and Meckling 1976, Jensen and Meckling 1992;

Aghion and Tirole 1997). If the FAH is responsible

for too many decisions, she has little time to invest in

acquiring information relevant for high-quality decision

making, and therefore her control over decisions is lim-

ited (Ouchi and Dowling 1974, Aghion and Tirole 1997).

The FAH can reduce loss of control over decisions del-

egated to the RAH when the actions of the RAH are

easier to observe and monitor. This can be achieved by

positioning both actors close to each other within the

organizational hierarchy. Such organizational proximity

reﬂects closeness of contact between the FAH and the

RAH and the efﬁciency of the supervisory time spent on

each decision shared with the RAH (Ouchi and Dowling

1974). It ensures a short chain of command, facilitates

direct communication, and makes coordination among

decision tasks easier. In addition, increasing the number

of decisions under the responsibility of a single FAH

increases the opportunity cost of time spent on monitor-

ing and intervening in each decision. The marginal cost

of intervention in each of those decisions increases with

an increasing number of decisions (Foss et al. 2006). So

the more decisions there are to monitor for a single FAH,

the more costly it is to get involved in intervening into

each of them (Rantakari 2012). Should the FAH want

to exert effort into monitoring the decision making of

an RAH, such involvement is more costly and time con-

suming when these two authority holders are located far

from one another in the hierarchy. We therefore expect

that the FAH counteracts loss of control resulting from

the authority split by delegating authority using a short

delegation span.

Hypothesis 1B (H1B). Increasing the number of de-

cisions that a single individual is responsible for is neg-

atively associated with the delegation span between the

FAH and the RAH.

Delegation and Specialized Knowledge

Decisions made in a large organization often require het-

erogeneous knowledge, such as customer preferences,

product design, or local market requirements. As a

result, individuals in the ﬁrm specialize in tasks and

are responsible for a subset of decisions; this typi-

cally results in increasing returns from task specializa-

tion (Becker and Murphy 1992, Bolton and Dewatripont

1994, Grant 1996). Thus, within a complex task struc-

ture, this means that knowledge necessary for decision

making is distributed among multiple individuals who

specialize in their respective tasks and who are located

at various functional or divisional units. Speciﬁcally, this

implies that some employees may hold expert knowl-

edge that is more relevant for certain decisions than

knowledge held by the FAH, as decision authority may

in some cases be assigned to the FAH on merits other

than expert knowledge (e.g., seniority, tenure, general

knowledge, managerial skills).

The FAH can improve the quality of decisions either

by acquiring the necessary knowledge or by delegating

decision authority to an employee that possesses supe-

rior, decision-relevant knowledge (Hayek 1945, Jensen

and Meckling 1992). Individual decisions may rely on

general as well as on speciﬁc knowledge (Jensen and

Meckling 1992). Whereas general knowledge is rela-

tively inexpensive to aggregate and transfer among indi-

viduals in the organization, speciﬁc knowledge is not,

as it involves specialized knowledge that is often tacit

(Hayek 1945, Jensen and Meckling 1992, Kogut and

Zander 1992, Grant 1996). Difﬁculties in transferring

such specialized knowledge from an employee to the

FAH arise from the time and effort necessary to transmit

and absorb that knowledge. These knowledge transfer

costs may outweigh the beneﬁts of knowledge acquisi-

tion by the FAH (Jensen and Meckling 1992). We there-

fore expect the following.

Hypothesis 2A (H2A). Specialization of knowledge

required for decision making is positively associated

with the likelihood of an authority split between the FAH

and the RAH.

As ﬁrms reach out for specialized knowledge resid-

ing in employees, they are likely to employ mechanisms

that promote best utilization of such knowledge. These

mechanisms need to address the trade-off associated

with delegation. Firms can beneﬁt from distributed deci-

sion making when authority is delegated to an RAH

whose decision-relevant knowledge is superior to that of

an FAH (Jensen and Meckling 1992). In that case, it is

likely that the FAH has conﬁdence in the RAH’s spe-

cialized knowledge and therefore may relax the control

over the RAH. Recall that a long delegation span also

signals no intention of intervention and overruling of

decisions that would make an employee’s efforts redun-

dant (Aghion and Tirole 1997). Therefore, managers that

want to motivate RAHs to use their decision-relevant

specialized knowledge may want to purposefully limit

their own intervention and overruling behavior. We

therefore suggest the following.

Hypothesis 2B (H2B). Specialization of knowledge

required for decision making is positively associated

with a delegation span between the FAH and the RAH.

Above, we explore how the knowledge characteristics

of a task (i.e., its degree of knowledge specialization)

may be related to delegation of decision authority. We

now turn our attention to the knowledge characteristics

of the person with authority for the task.

A scenario in which the person with authority also

has the relevant knowledge needed to complete a task

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

is an example of “colocation” of authority and knowl-

edge (Jensen and Meckling 1992). This colocation of

authority and relevant knowledge, on one hand, grows

as personnel’s career paths move them up in a hierar-

chy, and people’s earlier stages of their career accrue

knowledge that may later be relevant and fall into their

area of responsibility. On the other hand, we expect that

cases of colocation diminish as personnel move higher

in the hierarchy and become managers who supervise

large areas that are outside their original area of exper-

tise. Within large organizations that have complex task

structures, this implies that these upper-level managers

are facing a situation where they may have expertise in a

few areas and less (or no) expertise in most other areas.

We therefore expect the following.

Hypothesis 3A (H3A). A match of the FAH’s exper-

tise with the knowledge domain required for decision

making is negatively associated with the likelihood of an

authority split between the FAH and the RAH.

When decision authority is split between the FAH and

the RAH despite the fact that the FAH holds decision-

relevant knowledge, these actors can improve decision

quality by contributing both of their expertise—but there

is a cost to this coordination. A short delegation span

means that as informational distance is lessened (Foss

et al. 2006), direct communication and mutual adjust-

ment between these individuals becomes easier and less

noisy (Dessein 2002). A short delegation span is con-

ducive not only to knowledge coordination but also to

maintaining control over delegated decisions. As the

FAH possesses decision-relevant knowledge, she typi-

cally has an opinion on decisions made by the RAH;

thus a short span makes it easier not only to contribute

to decision making but also to evaluate the quality of the

RAH’s decisions and to exercise some form of monitor-

ing (Rantakari 2012). This suggests the following.

Hypothesis 3B (H3B). A match between the FAH’s

expertise and the knowledge domain required for deci-

sion making is negatively associated with a delegation

span between the FAH and the RAH.

Methods and Measures

Data and Sample

This paper is based on the study of a global Fortune 50

ﬁrm in the wind turbine industry. The ﬁrm manufactures

wind turbines and installs wind farms. In this study we

focus on the BUs of this ﬁrm as it allows us to construct

a unique data set of pooled cross sections. The core of

the data set comes from “responsibility charts” of the

years 2008, 2010, and 2011, which map in detail the core

business process that the ﬁrm uses for the planning and

execution of wind farms. A responsibility chart describes

a project organization, which is designed to plan and

execute wind farm projects. It considers the functional

organization structure of headquarters and BUs, and it

disaggregates a wind farm installation project into indi-

vidual tasks, such as checking completeness of techni-

cal speciﬁcations, clarifying tax-related issues, providing

commercial input to risk assessment, planning transport

equipment, planning site resources, etc., all of which are

also described in detail within an accompanying pro-

cess handbook. It is important to note that this project

planning and execution process does not include the

manufacturing of wind turbines nor the physical erec-

tion of the wind farm. More speciﬁcally, only 3% of

the observed BUs’ tasks involve physical labor-intensive

tasks,2and thus, the great majority of the tasks are

specialized and decision intensive with a high degree

of problem solving. The responsibility charts deﬁne

the task participants’ interdependence structure, thereby

deﬁning knowledge ﬂows and, more generally, the ﬁrm’s

project planning and execution infrastructure. The charts

also reﬂect the nature of the BU’s business model, which

excludes manufacturing, because this is the responsibil-

ity of an upstream division, and also excludes labor-

intensive installation tasks that are largely outsourced to

suppliers.

We obtained ﬁve responsibility charts from three

years. These charts were developed by the headquarters

to ensure that the various geographic units would exe-

cute the same generic business processes while having

the identical jobs and roles within the entire ﬁrm. For

2008, we have a chart of the centralized organization that

was structured as a single BU. For 2010, the ﬁrm decen-

tralized into headquarters and BUs serving offshore and

onshore wind farm markets; thus we have a chart for

offshore BU activities and another chart for onshore

BU activities. For 2011, the ﬁrm ﬁne-tuned its organi-

zation, so we again obtained a chart each for offshore

and onshore activities. For the decentralized structure in

2010 and 2011, both the offshore and onshore BUs have

tasks that are allocated among actors at the BU and the

headquarters. The main trigger for the change of organi-

zational structure was market growth that involved both

a geographical expansion of the ﬁrm and a signiﬁcant

growth in the offshore wind farm market. Since offshore

wind farms are much larger than onshore wind farms,

the management of the ﬁrm decided to facilitate adapta-

tion to the differing market conditions by decentralizing

the organizational structure into separate BUs.

A responsibility chart consists of a list of tasks (aver-

age N=151) and job titles of individual actors (average

N=111). Because there are many cases of multiple peo-

ple assigned to a job title, the 111 job titles in our setting

involve about 600 individuals in total. The responsibil-

ity charts assign particular responsibility types to all job

titles for all given tasks (which include “responsible,”

“approve,” “execute,” “contribute” and “is informed”).

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

Such a matrix is a standard way of charting task respon-

sibilities in a project organization (see, for examples,

Galbraith 1973, p. 147 and Billinger and Stieglitz 2009,

pp. 52–53), and it ensures in large corporations that stan-

dardized operating procedures and best practices are uni-

formly executed throughout the organization.3We use

these assignments of responsibility type to code the for-

mal and real authority holder for each task. The cod-

ing procedure is a straightforward copying of each job

title’s “responsible” and “execute” assignments from the

responsibility charts into a numeric data set. This proce-

dure does not involve interpretation by the authors. Each

task has only one actor with the assignment “execute.”

We code this role as the holder of real authority (i.e.,

RAH) over task decision making, because this role actu-

ally delivers the completed task drawing on one’s own

knowledge and resources of other task participants. We

also conﬁrmed this coding convention by reviewing the

process handbook, reading job descriptions, and inter-

viewing key managers. The process handbook describes

an RAH task using action verbs (emphasized here) such

as the “project manager shall review the technical part of

the contract,” the “technical project manager ensures that

all invoices are sent to the customer,” the “installation

project manager ensures that XX,” and the “installation

project manager also plans YY.”

Even though the RAH makes all task-related deci-

sions, she may be overruled by a formal authority holder

(i.e., FAH). Formal task authority is assigned to an actor

(i.e., the FAH) who has the role “responsible” and who

can overrule the role “execute.” In such cases, the pro-

cess handbook graphically maps assignments “responsi-

ble” and “execute” to respective roles and describes the

responsibilities of each role; for example, “The project

manager [FAH] is overall responsible for this activity

and initiates the mobilization of the site, while the site

manager [RAH] performs the mobilization of the site

according to the project checklist.” For tasks where the

“responsible” role is not assigned to any actor, the for-

mal authority is embedded in the “execute” role, and

thus the FAH and the RAH are a single person. In our

data set, formal and real authority over a task is split

between two separate actors in 38% of the observa-

tions. In total, across the ﬁve pooled responsibility charts

within our sample over the three-year cross sections,

we have a total of 4,329 task assignments, each with a

responsibility-type assignment.

Apart from this straightforward coding, we also

wanted to assess the degree to which the responsi-

bility charts and process handbooks correspond to the

actual allocation of responsibilities in daily life. To do

that, we conducted interviews with several managers and

key project participants, and we reviewed relevant com-

pany reports. The head of project management explained

that the responsibility charts have been widely used

both before and after decentralizing the organizational

structure into headquarters and business units. To further

triangulate an assessment of the actual implementation

of the responsibility charts, we approached the corpo-

rate quality management department, which conducts a

process maturity audit every two years; its purpose is to

evaluate the degree to which processes are codiﬁed and

how the codiﬁcation corresponds to the actual organiza-

tional routines executed in practice. The project planning

and execution process we study received positive scores

from this department for the time period in which we

studied it. Based on this evidence, we are conﬁdent that

the data from the process handbooks and responsibility

charts represent the actual allocation of authority and

information ﬂows sufﬁciently accurately.

Variables and Operationalization

To code and operationalize our variables, we used orga-

nizational charts to identify the hierarchical allocation of

all actors. We reviewed all individual job titles and then

content-analyzed and coded job descriptions to obtain

variables related to all roles’ expertise domains. We

also content-analyzed task descriptions to code the vari-

ables related to knowledge specialization and knowledge

domain required for task decision making. Moreover, we

conducted 36 semistructured interviews to understand

the nature of the process and the rationale behind reas-

signing the responsibilities to roles as a result of the

decentralization of the organization structure in 2010 and

the following incremental changes in 2011. Finally, we

used process handbooks, internal documents, and depart-

mental newsletters as additional secondary data to verify

and triangulate our data.

Dependent Variables. Our ﬁrst dependent variable is

the split of decision authority between the FAH and the

RAH (authority split between FAH and RAH). We oper-

ationalize this measure of delegation by observing the

FAH and the RAH in each task. The FAH maps the actor

from the responsibility chart that has the task assignment

of being responsible, and the RAH maps the actor with

the task assignment to execute. We coded the authority

split between the FAH and the RAH as a binary variable

with a value of 1 when assignments of being responsi-

ble and having to execute the task are assigned to two

different individuals.

The second dependent variable in our study is the del-

egation span, which counts the number of hierarchical

levels separating the FAH and the RAH on the organi-

zational charts. A delegation span of 0 means that the

FAH and RAH are located at the same hierarchy level;

i.e., their hierarchical distance to the chief executive ofﬁ-

cer (CEO) is identical. This is often the case when an

FAH delegates real authority to an RAH located in a

different functional department.4An example includes a

site manager who resides in the planning department (as

the FAH) delegating real authority to a quality inspector

(the RAH) who is residing in the department of quality

assurance.

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

Explanatory Variables. Hypotheses 1A and 1B pre-

dict the effects of information overload of the FAH on

our measures of decentralization. We capture informa-

tion overload by counting the number of individual deci-

sions that the FAH of a task is responsible for (number

of FAH’s decisions), which we obtain by identifying the

job title of a task’s FAH and counting for that job title

the number of tasks in the responsibility chart that carry

the task assignment of being responsible.

Hypotheses 2A and 2B refer to the effect of knowl-

edge specialization required for decision making on del-

egation from the FAH to the RAH. We code the required

knowledge specialization by conducting a content anal-

ysis of the task descriptions. We operationalize special-

ized knowledge as a binary variable that equals 1 when

the task decision making requires specialized knowl-

edge (tasks related to engineering expertise, tax or export

regulations, etc.) and that equals 0 when task decision

making relies on general knowledge (tasks related to

coordination, administrative assignments, etc.).

Finally, H3A and H3B predict the effect of an FAH

having expertise that is relevant or necessary for a par-

ticular task. The match of knowledge and expertise

(knowledge–expertise match) is also a binary variable;

we conduct content analysis of both task descriptions

and job descriptions of job titles. We check whether an

FAH’s knowledge domain required for her task decision

making matches the expertise domain of her job descrip-

tion and set it to 1 when a match takes place and set it

to 0 otherwise.

Control Variables. We include several control vari-

ables to account for changing characteristics. Year2010

and Year2011 are dummy variables coded as 1 for

tasks in the 2010 and 2011 responsibility charts, respec-

tively, with Year2008 being omitted. We also control

for business unit type (offshore), as the BU deliv-

ering offshore wind farms faces greater environmen-

tal complexity than the BU delivering onshore wind

farms.

We include several control variables to control for

agent heterogeneity at the FAH level. We identify the

FAH responsible for the decision task and create vari-

ables that capture the FAH’s (a) role characteristics,

(b) prior experience with delegation, and (c) prior expe-

rience with handling multiple tasks. The variable per-

sonnel responsibility is coded as 1 when the role has

subordinates listed on the organizational chart and 0 oth-

erwise. The intuition is that managers who have subor-

dinates are more likely to delegate tasks to them and

in this way act more as coordinators. The variable FAH

execution phase orientation analyzes the task portfolio

of the FAH and captures whether the role is respon-

sible for relatively more tasks in the project execution

phase as opposed to the project planning phase. To oper-

ationalize this, we count the number of the FAH’s tasks

in the project execution phase, deduct the number of the

FAH’s tasks in project planning phase, and divide that

difference by the total number of the FAH’s task. A pos-

itive sign indicates a greater presence of the FAH in the

project execution phase. This logic is similar to that of

our task-level control of execution phase, which expects

these tasks to be delegated to specialists after project

planning has been performed in a more centralized fash-

ion. To control for the FAH’s prior experience with del-

egation, we measure the percentage of the FAH’s tasks

that were delegated to RAHs in t−1 (share of FAH’s

delegated tasks t−1) as well as the percentage change

in delegated tasks between 2008 and 2010 (% change

in FAH’s delegated tasks 2008 vs. 2010). Including any

change control measures reduces our sample because

they require excluding observations from 2008 and from

2008 to 2010, respectively. To measure the FAH’s expe-

rience with handling multiple tasks, we calculate the per-

centage change in the FAH’s number of tasks in 2008

versus 2010 (% change in FAH’s number of decisions).

At the task level, we control for ease of task moni-

toring. Agency theory predicts that delegation of tasks

whose completion is easier to observe and monitor gen-

erates lower agency costs and brings relatively little

negative consequence of delegation. We proxy such “del-

egatability,” or ease of monitoring, with the variable

count informed colleagues, which counts the actors that

have an assignment “is informed” for a task in the

responsibility chart. We infer that if there are more “eyes

on a task,” it is possible to transmit information about

the task and to better monitor its proper completion. This

is therefore a proxy for informational interdependence,

which reduces the agency cost of delegation, and that

we expect will positively inﬂuence responsibility split.

Although we predict the ease of task monitoring to

be positively associated with authority split, we expect

that tasks with high levels of task interdependence

require signiﬁcant coordination and are therefore less

likely to be delegated. We proxy for such task coordi-

nation requirements using the variable count contribut-

ing colleagues, which counts how many actors actively

contribute to task completion by counting the “con-

tribute” assignments in the responsibility charts. The

process handbooks elaborate on this type of assignment

as follows: “The Installation Project Manager [FAH

and RAH] involves role [X] [‘contribute’], role [Y]

[‘contribute’], and role [Z] [‘contribute’] in task [YY].

Role [V] [‘contribute’] must also participate in order to

advise role [Y] on HSE matters”; also, “the Head of

Quality Management [FAH] must consult all relevant

parties such as role [Y] [‘contribute’] and role [Z] [‘con-

tribute’] in task [YY].”

Table 1 provides an overview and descriptive statistics

of our variables.

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

Table 1 Overview of Variables

Variable Description Obs. Mean Std. dev. Min-max

Dependent variables

Authority split between

FAH and RAH

Coded as 1 when formal and real decision authority over task is

split among two individuals. Formal authority is derived from

code “responsible” on the responsibility chart, and real authority

denotes code “execute” for a task.

751 00382 00486 0/1

Delegation span Count of hierarchical layers separating the FAH and RAH. A code

of 0 means that the FAH and RAH are at the same hierarchy

level or that the RAH is higher in the hierarchy located in

different department (no direct subordinate–superior

relationship). A code of 1 means that the RAH is at least one

level lower than FAH. Increasing values signify greater distance.

751 00175 00493 0–3

Explanatory variables

Number of FAH’s

decisions

The number of decisions under the responsibility of an FAH. 751 550197 410857 1–110

Specialized knowledge Coded as 1 when knowledge required for task decision making is

of specialized nature.

751 00217 00412 0/1

Knowledge–expertise

match

Binary variable coded as 1 when the knowledge domain of an

FAH overlaps with the knowledge domain required for task

decision making.

751 00705 00456 0/1

Control variables

Count informed

colleagues

Ease of monitoring of tasks. Count of agents with assignment “is

informed” about task completion.

751 10399 10999 0–16

Count contributing

colleagues

Proxy for task coordination requirements. Count of agents with

assignment “contribute” to task completion.

751 2091 30518 0–21

Execution phase Coded as 1 when a task belongs to the execution phase of the

business process.

751 00355 00478 0/1

Personnel responsibility Coded as 1 when the role has subordinates listed on the

organizational chart.

751 00761 00426 0/1

FAH execution phase

orientation

Indication of whether the FAH is responsible for relatively more

tasks in the project execution (positive sign) phase as opposed

to project planning (negative sign) phase in her task portfolio.

751 00086 00502 −1–1

FAH planning and

execution phase

Indication of whether the FAH is responsible for tasks in both the

planning and execution phases of the project.

751 00784 00411 0/1

FAH task diversity Variable that captures in how many task groups a task’s FAH has

decision authority.

751 60419 30218 0–9

Share of FAH’s delegated

tasks t−1

An FAH’s prior experience with delegation: percentage of FAHs

where formal and real authority were split between two actors in

t−1.

488 00405 00191 0–1

% change in FAH’s

delegated tasks 2008

vs. 2010

An FAH’s prior experience with delegation, based on a count of

delegated tasks of the FAH in 2008 versus 2010.

331 10225 30963 −0.272–16

% change in FAH’s

number of decisions

An FAH’s experience with handling multiple tasks, based on a

count of the FAH’s decisions in 2008 versus 2010.

331 20433 80018 −1–32

Offshore Coded as 1 for an offshore wind farm business unit, which

operates in more complex environments than an onshore wind

farm BU.

751 00414 00492 0/1

Year2010 Coded as 1 for tasks from the 2010 responsibility charts. 751 00384 00486 0/1

Year2011 Coded as 1 for tasks from the 2011 responsibility charts. 751 00440 00496 0/1

Methods

We use a two-step Heckman selection estimation proce-

dure (Heckman 1979, Greene 2003) since it is appropri-

ate for handling regression models in which there is a

selection bias arising from using a nonrandomly selected

sample. In our study, this is the case for delegation span

that is only observed when the ﬁrm decides to split for-

mal and real authority between two individuals.

The ﬁrst step of the Heckman model estimates the

selection equation that is the random effects probit

equation estimation of the likelihood of an authority

split between the FAH and the RAH. Two-stage selec-

tion models require that the ﬁrst stage include an instru-

mental variable that inﬂuences the selection equation

but not the dependent variable in the second stage. In

our analyses, we use the binary project execution phase

variable as an instrumental variable that predicts the

likelihood of split decision authority between the FAH

and the RAH but does not necessarily affect hierarchi-

cal span.5The logic for this instrumental variable is the

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

following. Individual tasks of the business process are

arranged into three consecutive project stages: project

planning, execution, and completion; each is described

in the ﬁrm’s process handbook. These sequential phases

are important because planning and completion stages

require greater centralization for the purpose of main-

taining an overview (and in particular, cost control) over

an entire project, whereas decisions related to the project

execution may be delegated to various specialists who

then use template procedures. We do not have any the-

oretical expectations for why decisions related to the

project execution phase would be associated with a nar-

row or wide delegation span. The results from the ﬁrst-

stage regression are reported in Table 3.

In the second step, the second stage of the Heckman

model then estimates the outcome equation; i.e., it esti-

mates the delegation span, after including as a control

the inverse Mills ratio from the ﬁrst selection equation.

Since delegation span can only be measured for deci-

sions where authority has been split, the sample size

in the second-stage model is reduced from 751 to 287

decision tasks.6We report results from the second-stage

regression in Table 4. We conduct likelihood ratio (lr)

tests to assess the improvement of overall model ﬁt as

well as Wald 2tests for the ﬁrst-stage probit models

and F-tests for the second-stage ordinary least squares

(OLS) models.

Results

The results section is structured as follows. First, we

tabulate descriptive statistics and correlations of our vari-

ables in Tables 1 and 2.

Next, we report the analyses that test our hypothe-

ses. Table 3 presents the analysis of the likelihood of

authority split. To test whether the inclusion of the main

explanatory variables provides signiﬁcant improvement

in model ﬁt, we used an lr test. The log likelihood

ratio test asks whether constraining parameters to 0, i.e.,

leaving out the predictor variables, signiﬁcantly reduces

the ﬁt of the model. The log likelihood ratio compar-

ison is reported at the bottom of Table 3. Given that

Model 1 represents the baseline containing only control

variables, there is no lr statistic reported for this model.

The comparison of Model 2 versus Model 1 indicates

that adding the three main explanatory variables signiﬁ-

cantly increased the overall model ﬁt (p < 00001).

Table 4 reports the second-stage OLS regression mod-

els that analyze delegation span. The ﬁrst model contains

only control variables. Model 2 adds our three explana-

tory variables and shows that the variance explained

(i.e., R2statistic) increases from 0.096 to 0.146. An

increase in log likelihood and a likelihood ratio com-

parison between Model 1, the restricted (basic) model,

and Model 2 shows a signiﬁcant difference (p < 0001).

Model improvement is also conﬁrmed by an F-test

(p < 0001).

Finally, we interpret the effects of the control variables

and report additional robustness checks.

Main Findings

We turn our attention ﬁrst to H1A. Recall that H1A pre-

dicted that increases in the number of decisions under

the FAH’s responsibility would be positively related to

the likelihood of an authority split between the FAH

and the RAH. Table 3 shows our analysis in which

the dependent variable is the likelihood of an authority

split between the FAH and the RAH. Model 2 shows

a signiﬁcant positive effect of the number of decisions

under the responsibility of the FAH on the likelihood

of an authority split (p < 00001). Thus, managers del-

egate real decision authority to other individuals when

they are responsible for many decisions, thereby reduc-

ing their risk of information processing overload. This

result lends support to H1A and is aligned with the argu-

ment that rational decision makers reduce their cognitive

burden by sharing decision making with one (or several)

RAH(s).

Hypothesis 1B predicted that increases in the num-

ber of decisions under an FAH’s responsibility would be

negatively associated with the delegation span between

the FAH and the RAH. Model 2 in Table 4 shows

a signiﬁcant negative coefﬁcient (p < 0005) and sup-

ports H1B. Information overload of the FAH makes

her inclined to delegate decision authority; however, she

delegates to individuals positioned on adjacent (close)

hierarchy levels. As a short hierarchical distance makes

it easier for the FAH to observe and supervise the

RAH’s actions, it is conducive to maintaining control

over decisions.

Having interpreted how information overload inﬂu-

ences our measures of delegation, we now turn to

analyzing the impact of knowledge specialization.

Hypothesis 2A predicted that the specialized nature of

knowledge required for decision making and task execu-

tion would be positively related with an authority split

between the FAH and the RAH, and H2B predicted its

positive effect on delegation span. A positive and sig-

niﬁcant (p < 00001) effect of knowledge specialization

in Table 3 yields support for H2A. This reveals that

when decision making requires narrow expertise, author-

ity holders seek to increase decision quality by delegat-

ing real decision authority to experts. We do not ﬁnd a

signiﬁcant direct effect of knowledge specialization on

delegation span in Model 2 of Table 4. This initial result

does not support H2B and indicates that in cases where

the decision authority is split, the FAH searches for

experts with specialized knowledge regardless of their

location in the organizational hierarchy. We will further

qualify this relationship by analyzing several interaction

effects below.

We now move to testing the hypotheses predict-

ing the impact on delegation of an existing match

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

Table 2 Correlations

Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

1Authority split between 1

FAH and RAH

2Delegation span 00197∗1

3Number of FAH’s 00132∗−000418 1

decisions

4Specialized 000706 000311 −00253∗1

knowledge

5Knowledge– −00285∗−000783 −00225∗−000949 1

expertise match

6Count informed −000009 −000702 −000258 −00158∗00209∗1

colleagues

7Count contributing −00103∗00028 00088∗−00106∗000151 00075∗1

colleagues

8Execution phase 00152∗000548 −00229∗000349 −00158∗−00219∗00119∗1

9Personnel 00261∗00104 00653∗−00199∗−00142∗000165 00124∗−00103 1

responsibility

10 FAH execution 000675 000327 −00274∗000200 −000171 −00651 000668 00487∗−00143∗1

phase orientation

11 FAH planning and 00142∗−000851 00584∗−00281∗−000994 000223 000292 −00347∗00628∗−00384∗1

execution phase

12 FAH task diversity 00203∗000123 00908∗−00298∗−00226∗000086 00108∗−00286∗00737∗−00372∗00759∗1

13 Share of FAH’s 00425∗00314∗00450∗−00121∗−00233∗−00337 00129∗−000655 00590∗−000453 00427∗00588∗1

delegated

tasks t−1

14 % change in FAH’s 00127∗−00102 −00326∗000919 000926 −00940 000079 00101 000652 00286∗000788 −00238∗000303 1

no. of decisions

15 % change in FAH’s 00152∗−000844 −00303∗000850 000767 −000946 000227 00111 000991 00324∗00128∗−00207∗000841 00991∗1

delegated tasks

2008 vs. 2010

16 Offshore 00138∗00180∗−00450∗000196 000491 00118∗00141∗000400 −00105 000751 000124 −00118∗00156∗00288∗00301∗1

17 Year2010 000722∗000307 −00087∗000105 −000111 00101∗000516 000293 0000380 −000345 000670 000356 −001796 · · 00140∗1

18 Year2011 −000212 −000525 −000096 −000279 000662 00089∗00073∗−000139 −000548 000583 −00073∗−00083∗001796 · · 00159∗−00702∗1

∗p < 0005.

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

Table 3 Heckman Regression, First Stage: Predictors of a Decision Authority Split Between the FAH and RAH

DV: Authority split between FAH and RAH

Variable 1 2 3 4 5 6 7

Number of FAH’s decisions 00013∗∗∗ 00013∗∗∗ 00008∗∗ 00013∗∗∗ 00013∗∗∗ 00012∗∗∗

4000025 4000025 4000035 4000025 4000025 4000025

Specialized knowledge 00464∗∗∗ 00469∗∗∗ 00485∗∗∗ 00382∗00495∗∗∗ 00483∗∗∗

4001295 4001325 4001305 4001955 4001305 4001305

Knowledge–expertise −00464∗∗∗ −00461∗∗∗ −00544∗∗∗ −00501∗∗∗ −00600∗∗∗ −00432∗∗∗

match 4001155 4001175 4001195 4001325 4001285 4001165

Execution phase 00465∗∗∗ 00806∗∗∗ 00807∗∗∗ 00809∗∗∗ 00796∗∗∗ 00844∗∗∗ 00827∗∗∗

4001025 4001205 4001205 4001215 4001215 4001215 4001215

Offshore 00208∗00731∗∗∗ 00729∗∗∗ 00731∗∗∗ 00734∗∗∗ 00744∗∗∗ 00718∗∗∗

4001045 4001285 4001285 4001295 4001285 4001285 4001295

Year2010 00257†00251 00251 00262 00245 00210 00293†

4001525 4001625 4001625 4001625 4001625 4001635 4001645

Year2011 00130 00076 00078 00100 00069 00038 00105

4001505 4001605 4001605 4001605 4001605 4001615 4001625

Count informed colleagues 00013 00065∗00065∗00063∗00066∗00298∗∗∗ 00062∗

4000265 4000295 4000295 4000295 4000295 4000885 4000295

Count contributing colleagues −00065∗∗∗ −00099∗∗∗ −00099∗∗∗ −00106∗∗∗ −00099∗∗∗ −00103∗∗∗ −00101∗∗∗

4000175 4000205 4000205 4000205 4000205 4000205 4000215

Number of FAH’s decisions ×00001

Specialized knowledge 4000035

Number of FAH’s decisions ×00007∗

Knowledge–expertise match 4000035

Knowledge–expertise match ×00142

Specialized knowledge 4002515

Knowledge–expertise match × −00257∗∗

Count informed colleagues 4000915

Number of FAH’s decisions × −00001∗∗

Count contributing colleagues 4000015

Constant −00556∗∗∗ −10375∗∗∗ −10370∗∗∗ −10014∗∗∗ −10349∗∗∗ −10566∗∗∗ −10386∗∗∗

4001245 4002195 4002215 4002625 4002245 4002315 4002235

Observations 751 751 751 751 751 751 751

Log likelihood −4780981 −4190964 −4190949 −4160848 −4190804 −4150698 −4150447

Likelihood ratio comparison 118003∗∗∗ 0003 6023∗00320 80531∗∗ 90033∗

Wald 200029 60217∗00320 70947∗∗ 80690∗∗

2410033∗∗∗ 1590068∗∗∗ 1590097∗∗∗ 1650300∗∗∗ 1590387∗∗∗ 1670599∗∗∗ 1680101∗∗∗

Pseudo R200041 00159 00159 00165 00160 00168 00168

Notes. Standard errors are in parentheses. DV, dependent variable.

∗p < 0005; ∗∗p < 0001; ∗∗∗ p < 00001; †p < 001.

between required knowledge and the FAH’s expertise.

We hypothesize that an overlap of the FAH’s knowl-

edge with the knowledge domain required for decision

making is negatively related with the likelihood of an

authority split (H3A) and negatively related with the del-

egation span between the FAH and the RAH (H3B).

The negative and signiﬁcant (p < 00001) effect (for

a knowledge–expertise match) in Model 2 of Table 3

yields support for H3A, conﬁrming that the FAH is

indeed less likely to delegate if she already possesses the

necessary knowledge resources. The analysis presented

in Model 2 of Table 4 shows a nonsigniﬁcant direct

effect of the knowledge–expertise match on the delega-

tion span, thereby showing no initial support for H3B.

Later, we explore these relationships further through

analysis of interaction effects to qualify our result.

Interpreting Controls

We now turn our attention to interpreting the control

variables. Controlling for phases of the business process

reveals that decisions made during the project execution

phase, as opposed to project planning and completion,

tend to involve an authority split between the FAH and

the RAH. This is intuitive, as tasks related to execu-

tion may involve more template procedures (or standard

operating procedures) with individual milestones than

decisions related to other stages, which require more

overview and monitoring of results and project perfor-

mance. The decision authority in the offshore business

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

Table 4 Heckman Regression, Second Stage: Predictors of Delegation Span

DV: Delegation span

Variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7

Number of FAH’s decisions −00003∗−00001 −00006∗∗∗ −00003∗−00003∗−00004∗∗

4000015 4000015 4000025 4000015 4000015 4000015

Specialized knowledge 00133 00155†00151†00278∗00137 00118

4000855 4000845 4000835 4001095 4000855 4000855

Knowledge–expertise −00121 −00134 −00209∗−00026 −00117 −00106

match 4000825 4000815 4000845 4000935 4000825 4000825

Offshore 00102 00051 00077 00061 00036 00060 00038

4000785 4000935 4000935 4000915 4000935 4000935 4000935

Year2010 −00386∗∗ −00373∗∗ −00367∗∗ −00360∗∗ −00362∗∗ −00358∗∗ −00375∗∗∗

4001205 4001135 4001125 4001115 4001125 4001145 4001135

Year2011 −00450∗∗∗ −00433∗∗∗ −00448∗∗∗ −00397∗∗∗ −00413∗∗∗ −00424∗∗∗ −00422∗∗∗

4001145 4001115 4001105 4001095 4001115 4001115 4001105

Count informed colleagues −00004 00010 00012 00012 00006 −00028 00006

4000165 4000165 4000165 4000165 4000165 4000385 4000175

Count contributing colleagues 00068∗∗∗ 00064∗∗∗ 00065∗∗∗ 00061∗∗∗ 00065∗∗∗ 00063∗∗∗ 00067∗∗∗

4000185 4000185 4000185 4000185 4000185 4000185 4000185

Number of FAH’s decisions × −00005∗∗

Specialized knowledge 4000025

Number of FAH’s decisions ×00006∗∗∗

Knowledge–expertise match 4000025

Knowledge–expertise match −00313∗

Specialized knowledge 4001485

Knowledge–expertise match ×00047

Count informed colleagues 4000425

Number of FAH’s decisions × −00001†

Count contributing colleagues 4000005

Inverse Mills ratio −00187 00042 00060 00090 00009 00083 −00022

4002215 4001485 4001475 4001465 4001485 4001535 4001525

Constant 00691∗∗ 00693∗∗ 00529∗00895∗∗∗ 00685∗∗ 00674∗∗ 00786∗∗∗

4002515 4002275 4002325 4002305 4002265 4002285 4002325

Observations 287 287 287 287 287 287 287

Log likelihood −2360021 −2270971 −2240039 −2210710 −2250678 −2270331 −2260274

Likelihood ratio comparison 160101∗∗ 70864∗∗ 120523∗∗∗ 40586∗10280 30394∗∗∗

F-statistic 50474∗∗ 70667∗∗ 120309∗∗∗ 40446∗10234 30283†

R200096 00146 00169 00182 00159 00149 00156

Notes. Standard errors are in parentheses. DV, dependent variable.

∗p < 0005; ∗∗p < 0001; ∗∗∗ p < 00001; †p < 001.

unit, representing a more complex environment, tends to

be split more frequently than authority in the onshore

business unit (see Table 3). Offshore business does not

have a signiﬁcant effect on the extent of delegation span

(see Table 4). Year dummies are not signiﬁcant predic-

tors of an authority split between the FAH and the RAH

(see Table 3); however, their signiﬁcant negative effect

in Table 4 reveals that the ﬁrm adopted a shorter delega-

tion span in later years compared with the initial 2008

organization design.

We also control for the ease with which tasks can

be monitored, which we proxy with a count of indi-

viduals who are informed about task completion (count

informed colleagues), as the number of people that are

informed reﬂects information transfer. The full speciﬁ-

cation Model 2 of Table 3 shows a positive effect on

the likelihood of an authority split (p < 0005), conﬁrm-

ing our expectation that tasks that are easier to monitor

are more likely to be delegated. This ease of monitoring

does not display a signiﬁcant effect on the delegation

span (see Table 4). On the other hand, task coordination

requirements, which we proxy with a count of individu-

als who contribute to task completion (count contribut-

ing colleagues), is negatively related to the likelihood of

an authority split (all models in Table 3) and positively

related to a delegation span (all models in Table 4).7

This ﬁnding shows that if a task requires coordination of

inputs from multiple individuals (or collaboration), this

coordination is handled in a centralized manner by the

FAH. If, however, authority over such decision tasks is

split between the FAH and the RAH, these two roles

maintain a close proximity.

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

Further Analysis of Interaction Effects

As we did not ﬁnd direct support for H2B and H3B,

we probe our data further to check the conditions under

which these hypotheses may hold. Model 4 in Table 3

shows a positive main effect of the number of an FAH’s

decisions on the likelihood of an authority split (p <

0001), a negative main effect of match between required

knowledge and the expertise of FAH (p < 00001), and

a positive interaction effect (p < 0005). A Wald 2

statistic of 6.217 is signiﬁcant (p < 0005) and reveals

that adding an interaction effect is signiﬁcantly differ-

ent from Model 2. Additionally, the lr test shows a sig-

niﬁcant (p < 0005) improvement in ﬁt over Model 2.

This result reveals that the likelihood of authority split

is greater both when the FAH is responsible for many

decisions and when she holds decision-relevant knowl-

edge, demonstrating that decision overload is a stronger

predictor of delegation than the relevance of the FAH’s

knowledge. Interaction effects in Models 3 and 5 do

not show signiﬁcant effects on the likelihood of author-

ity split, nor do they offer a signiﬁcant improvement in

model ﬁt.

Next, we analyze interaction effects between our

explanatory variables in their inﬂuence on delegation

span. Model 3 in Table 4 offers a signiﬁcant improve-

ment over Model 2 (F-statistic of 7.667; p < 0001). This

model shows a nonsigniﬁcant main effect of the number

of an FAH’s decisions, a positive main effect for knowl-

edge specialization (p < 001), and a negative interaction

effect (p < 0001). Thus, although knowledge specializa-

tion has a positive effect on delegation span, its effect

decreases with an increasing number of decisions under

the responsibility of the FAH. This result qualiﬁes the

relationship that we propose in H2B. Recall that H2B

suggests that with a wide delegation span, the FAH pro-

motes an RAH’s initiative by signaling an intention not

to intervene in decisions that have been delegated to a

specialist RAH. In Model 3 we see that this only holds

when the cognitive capacity of the FAH is not over-

loaded by the number of decisions to manage. With spe-

cialization of knowledge and an increase in the number

of decisions, the effect reverses and becomes negatively

associated with delegation span when the number of the

FAH’s decisions reaches 84. This result indicates that

when the FAH faces risk of information overload and the

task requires specialized knowledge, she tends to focus

on control over the decision (instead of signaling non-

intervention) and seeks to facilitate communication with

experts through shortening the hierarchical distance.

We also investigate the analogous interaction between

the match in knowledge expertise and the number of

decisions under responsibility of the FAH. Both of these

variables are associated with a short delegation span,

as shown by the main effects in Model 4 of Table 4.

There is, however, a positive interaction effect between

them (p < 00001), indicating a substitution effect. The

Wald 2and the lr tests show that adding this interaction

term offers a signiﬁcant improvement in model ﬁt over

Model 2 (p < 00001). Although a match of knowledge–

expertise lowers the delegation span, this relationship is

reversed as the FAH has more decisions to manage. We

observe the change in the direction of this relationship

when the FAH is responsible for 86 decisions. Stated

differently, although an increase in the number of deci-

sions is associated with closer hierarchical distance and

oversight, a greater match of knowledge–expertise miti-

gates this and thus the span widens. When the FAH is at

a high risk of information overload, she faces substantial

demands and does not necessarily use her own matched

knowledge to execute a decision or keep oversight of

the decision; instead, she depends on others who also

have the relevant knowledge to focus on selected deci-

sions (Simon 1982, Ocasio 1997), and these others may

be further away in the organization.

Finally, in Model 5 of Table 4, we analyze the inter-

action effect between the match of knowledge–expertise

and knowledge specialization. Model 5 offers a signif-

icant improvement in model ﬁt over Model 2. When

decision requires specialized knowledge and the FAH’s

knowledge domain contains that expertise, then a short

delegation span is used. Matched knowledge and exper-

tise has a greater inﬂuence over the delegation span

than specialized knowledge. When the FAH has rele-

vant matched knowledge and expertise, she delegates

to a nearby hierarchy level to contribute with her own

knowledge, which is aligned with our arguments. This

demonstrates that when two experts jointly participate in

a decision, they are positioned closely to each other in

the organizational hierarchy so that vertical communica-

tion and coordination of specialized knowledge is facil-

itated because the informational distance is short. Since

expertise refers to having the ability to evaluate and con-

trol delegated decisions, not having relevant knowledge

implies a limited capacity to assess delegated decisions

or to contribute to joint decision making with one’s own

knowledge. When these two conditions meet, the FAH

employs a long delegation span. This additional analy-

sis of interaction effects reveals that we do ﬁnd some

nuanced support for H2B and H3B that is consistent

with our arguments.

We also further explore how the extent of monitor-

ing and coordination roles may inﬂuence the delegation

span. Recall that we found (in Table 3) that our con-

trols for the number of individuals who are informed

or contribute have an association with the likelihood to

split decision authority. We examine several interaction

effects that further shed light on these constructs.

Model 6 in Table 3 displays a negative main effect

of the match between the FAH’s knowledge domain and

expertise required for task decision making on the like-

lihood of an authority split, a positive main effect of the

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

number of individuals who are informed about task com-

pletion, and a negative interaction effect. This means that

when many people are monitoring task completion (i.e.,

informational interdependence), then having the match

of required knowledge and available expertise makes

authority split even less likely.

This result shows that having many eyes on a task pro-

vides a strong monitoring mechanism that prevents del-

egation even though available knowledge resides within

an FAH. Second, we examine the impact of individ-

uals who contribute to a task. Model 7 of Table 3

shows a positive main effect of the number of the FAH’s

decisions on the likelihood of authority split, a nega-

tive main effect of the count of individuals who con-

tribute to the task, and a negative interaction effect of

these two. This result shows that when the FAH has

many decisions to manage, the likelihood of authority

split decreases when task decision making requires col-

laboration (i.e., task interdependence). In other words,

even though increasing the number of the FAH’s deci-

sions generally increases the likelihood of an author-

ity split, coordination requirements within tasks trump

this effect, and overall, the likelihood of an authority

split decreases both when the FAH has many decisions

under her own responsibility and when the task needs

collaborative efforts from many individuals. Both Mod-

els 6 and 7 offer a signiﬁcant improvement in model

ﬁt over Model 2. The ﬁndings set interesting boundary

conditions to the likelihood of delegation when cogni-

tive capacity of an individual manager is strained by

coordination requirements within a task. It also shows

that different types of interdependence, i.e., informa-

tional interdependence versus task interdependence, have

different effects on delegation (split).

Finally, we examine a similar interaction effect and its

inﬂuence on delegation span; this is reported in Model 7

of Table 4 and reveals another interesting ﬁnding. When

the FAH has a high decision load, then increasing the

number of task contributors has a negative effect on

delegation span, which is consistent with our reasoning

motivating H1B. This ﬁnding reveals that task overload

along with the desire to keep tasks close trumps coor-

dination requirements that usually move work further

down in the hierarchy.8

Discussion and Conclusions

This paper sets out to empirically investigate the con-

ditions under which delegation takes place in a large

and complex task structure that is characterized by dis-

tributed, interdependent decision authority across multi-

ple levels of an organizational hierarchy. Our baseline

results show that task decision authority is likely to be

split between an FAH and an RAH when (a) special-

ized knowledge is required for decision making; (b) the

FAH’s domain of expertise does not match with the

knowledge required for decision making, and therefore

the FAH searches for an RAH having relevant knowl-

edge; or (c) the FAH’s cognitive capacity is overloaded

because she is managing a large number of decisions.

Whenever authority is split, the delegation span between

the FAH and the RAH (a) decreases as the FAH faces

a task overload and (b) increases as more task contribu-

tors get involved. Broadly speaking, these baseline ﬁnd-

ings conﬁrm existing literature on delegation, and our

research serves to empirically corroborate the theoreti-

cal arguments proposed by Aghion and Tirole (1997).

Novel insights are derived from our investigation con-

cerning the speciﬁc conditions that shape delegation. In

the following, we further elaborate on these conditions

and discuss implications.

First, we discover that task monitoring and both inter-

task and intratask coordination requirements can prevent

delegation. We ﬁnd that the split of authority between

the FAH and the RAH is less likely when there exists

a match of required knowledge and the FAH’s expertise

and when many people are being kept informed about

a task completion. In other words, if a well-informed

FAH’s task is being monitored by many people, the

likelihood that she delegates to an RAH decreases.

This ﬁnding is counterintuitive, as agency literature pre-

dicts that when a task is easy to monitor, then the

agency cost of delegation is less, and therefore delega-

tion should take place under these conditions. Our ﬁnd-

ing demonstrates that delegation can become redundant

when formal decision makers possess decision-relevant

knowledge. In such cases, knowledge relevance argu-

ments are more important than those regarding the ease

of monitoring. We further ﬁnd that the coordination

requirements within tasks can also decrease the likeli-

hood of an authority split. This is the case when an FAH

is facing responsibility for many decisions as well as

coordination obligations with many task contributors. In

these situations the FAH becomes a coordination hub

that pools information by having a central position in

the overall task structure. Both results point to moni-

toring necessities and coordination requirements being

important for deﬁning speciﬁc conditions under which

delegation does not occur. In general, this is in align-

ment with contributions that discuss the possible danger

of having RAHs that may act in a self-interested manner

(Jensen and Meckling 1976, Jensen and Meckling 1992)

and more recent observations that point to the preva-

lence and persistence of hierarchal, rather than the often-

discussed ﬂat, organizational structures (Diefenbach and

Sillince 2011). Our ﬁndings partially conﬁrm the latter,

as they point to the need for nondelegated, top-down,

decision making as an important characteristic that we

expect in large and complex task structures.

Second, our analysis has shown that different types

of interdependence, i.e., informational interdependence

versus task interdependence, have different effects on

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

delegation. For managers this creates the need to acquire

a realistic, ﬁne-grained understanding of who can and

who should be accountable for what. In large, com-

plex task structures with interdependent decisions, this

understanding requires knowledge concerning (a) how a

task can be disaggregated into subtasks, (b) how deci-

sion rights can be distributed between individual actors,

and (c) how both the disaggregation of tasks and distri-

bution of decision rights can interdependently be man-

aged within the organization. Prior studies of modularity

(e.g., Baldwin and Clark 2000) have put an empha-

sis on task disaggregation by systematically examin-

ing process and product structures. Our study shows

that in addition to these types of structure, the distribu-

tion of authority is particularly indispensable for large

task structures to function. In particular, responsibility

charts, which outline in detail delegation by assigning

unambiguous, distinct responsibilities for all individu-

als involved, seem to be useful tools for designing a

large and complex organization’s division of labor across

several hierarchical levels. It is also apparent that such

a design entails extensive effort, such as the signiﬁ-

cant time managers need to ﬁrst accumulate and then

reﬁne the required understanding of what works how

with whom. As a result, revisions of the responsibility

charts do not occur very frequently (e.g., in our case

they only occur every one to two years), which indi-

cates that organizational learning concerning distributed,

interdependent delegation structures necessitates long

time durations. This is important because it shows that

effective decision structures, which have attracted atten-

tion by several researchers more recently (Christensen

and Knudsen 2010, Csaszar 2012), become demand-

ing to design and adapt, especially across multiple lev-

els of the organizational hierarchy. Our study suggests

that in large, complex task structures, the most difﬁ-

cult challenge in adapting decision structures may be the

nuances associated with the division of ﬁne-grained task

responsibilities.

Third, middle management perspectives have received

much attention (Wooldridge et al. 2008), examining, for

instance, management roles (Bartlett and Ghoshal 1997),

leadership styles (Beatty and Lee 1992), and manage-

ment actions (Mantere 2008). This stream of literature

studies what middle managers do and what functions

they can possess in a given environment. Our study pro-

vides a complimentary perspective by outlining the spe-

ciﬁc ways in which an organization assigns decision

rights to middle managers. This perspective is important

because it offers concrete ways to observe and study

the discretion that an organization allocates to personnel;

organizations deﬁne this discretion either by restrict-

ing authority (e.g., to achieve an organization’s over-

all goals) or by granting authority (e.g., to motivate

or empower lower-level employees). Our results sug-

gest that delegation is an important design parameter

with which organizations can deﬁne the actions in which

managers and their subordinates engage. As such, a care-

fully devised delegation strategy offers organizations a

way to proactively manipulate the roles and styles that

managers are able to adopt within the organization.

Fourth, another important aspect of delegation is the

role it plays with knowledge and information ﬂows

within large organizations. Since the emergence of orga-

nizational knowledge creation theory (Nonaka 1994),

signiﬁcant attention was given to how upper-echelon

managers can facilitate the process of knowledge cre-

ation. More recent contributions (e.g., von Krogh et al.

2012) propose the need to distinguish centralized and

distributed leadership in order to be able to adequately

and accurately comprehend existing resource assets. Our

analytical framework assesses resource assets at the

granular level of individuals, thereby offering insights

into where and how knowledge assets are created and

accumulated. It shows, on a micro level, how an organi-

zation can orchestrate resources (Sirmon et al. 2011) by

granting lower-level decision makers sufﬁcient decision

rights—not only to implement their expert knowledge

for a given task but also to invest, for instance, in orga-

nizational search that may eventually result in resource

assets.

Fifth, delegation is arguably a key determinant that

deﬁnes the division of labor across hierarchies. As our

literature review shows, delegation has mainly received

attention in contributions that either conceptually or for-

mally approach the topic (e.g., Jensen and Meckling

1992, Aghion and Tirole 1997, Rivkin and Siggelkow

2003). Empirical microstudies are rare, and the research

community still seems to lack appropriate empirical

measures. We extend this existing research and intro-

duce the delegation span as a novel empirical measure

that captures the distribution of authority between differ-

ent organizational levels. Although this measure cannot

capture all aspects of delegation, and alternative mea-

sures are conceivable, it adds an important empirical

component that operationalizes the hierarchical distance

between organizational decision makers. Our results sug-

gest that this distance measure not only is important for

describing “informational distance” (Foss et al. 2006,

p. 805) but may also offer ways for organizations to

measure, assess, and shape accountability in an attempt

to reduce self-serving behavior (Rus et al. 2012). This is

important because it brings us closer to answering sev-

eral prevailing issues, including questions such as the

following: What is the “right” number of organizational

levels within a ﬁrm, what are the limits of ﬂat hierar-

chies, and what is sufﬁcient formal authority in a given

context? All of these questions require ways to charac-

terize the division of labor and decision rights across

hierarchies; our construct of delegation span does this in

a ﬁrst approximation.

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

Our study has several limitations, many of which are

rooted in the nature of our data. In this paper, all obser-

vations stem from a division that is part of a large cor-

poration, but we cannot compare with other divisions in

the ﬁrm or across ﬁrms. Future studies that may be able

to replicate some of our constructs in multiple parts of

an organization would be able to test for within-ﬁrm het-

erogeneity in delegation patterns. Also, because of the

structure of the data, we are unable to track tasks over

periods of time, and so it is not possible for us to track

the evolution of tasks in greater detail. An interesting

project would be one that examines the reconﬁguration

of tasks; this would involve tracing the evolution of tasks

and examining which are split into multiple tasks, are

combined into larger tasks, or stay unchanged. A rel-

evant follow-up study to this may be to examine how

different task evolutions affect decision authority split

and delegation span. Finally, our data sample derives

from a ﬁrm that did not change its strategy during the

observation period; this means that we cannot examine

how delegation structures change when a ﬁrm’s under-

lying strategic paradigm shifts. The study of delegation

in such contexts is potential future research.

This paper contributes to the literature on delegation

and, more broadly, on organization design and ﬁrm adap-

tation during decentralization. A more general insight

from this study is that delegation seems remarkably

understudied despite its undeniable important role in the

organization of multilevel hierarchies. Our results reveal

that only a thorough understanding of the underlying

mechanisms can help develop robust theory that is useful

for both scientiﬁc progress and real-world application.

We hope that this study will motivate future delegation

studies that investigate this central theme of organiza-

tional adaptation.

Supplemental Material

Supplemental material to this paper is available at http://dx.doi

.org/10.1287/orsc.2014.0954.

Acknowledgments

The authors thank, Markus Becker, Jörg Claussen, Nicolai

Foss, Martin Goossen, Aseem Kaul, Jacob Lyngsie, Solon

Moreira, Andrea Prencipe, Phanish Puranam, Tim Simcoe,

Nils Stieglitz, the Strategic Organization Design research unit

at the University of Southern Denmark, and the Department of

Strategic Management and Globalization at Copenhagen Busi-

ness School. The authors are also grateful for the constructive

feedback from the anonymous reviewers and the senior edi-

tor, Giovanni Gavetti, as well as conference participants at

the 2012 Academy of Management annual meeting, the 2013

DRUID Conference, and the 2014 Strategic Management Soci-

ety annual meeting. Finally, the authors thank Viento for pro-

viding access to the data.

Endnotes

1Research that employs formal methods has modeled delega-

tion as centralized or coordinating hierarchies (Mihm et al.

2010); department heads making decisions with or without

seeking approval from superiors (Siggelkow and Rivkin 2005);

or managers rejecting, accepting, or rubberstamping propos-

als (Rivkin and Siggelkow 2003). Large Nempirical stud-

ies have measured “degrees of delegation” as the extent to

which employees have the authority to make speciﬁc decisions

(Vázquez 2004, Chowdhury 2011), the degree to which a plant

manager can take certain decisions without prior authoriza-

tion from the central headquarters (Bloom et al. 2010a, b),

or distinguished between partial and full delegation by asking

whether certain decisions require formal authorization from a

superior (Colombo and Delmastro 2004, 2008). Other stud-

ies investigated the hierarchical locus of formal authority and

decision inﬂuence (Collins et al. 1999) and the locus of formal

authority versus the hierarchical level at which decisions are

usually made in practice (Lincoln et al. 1986).

2We exclude these tasks from the data set because the theoret-

ical arguments of the study require an empirical setting with

decision-intensive tasks.

3Although situation-speciﬁc adjustments may be made to dele-

gation structures, in our case, such adjustments were negligible

because the ﬁrm’s headquarters insisted on uniform structures

across the corporation.

4The organizational structure of the ﬁrm is divisional, with a

distinction between the headquarters and business units; fur-

thermore, the business units are internally conﬁgured by func-

tion. When referring to “departments,” we mean functional

departments within headquarters or business units.

5To check the validity of our instrumental variable project

phase execution, we regressed the residuals from the second-

stage model with the dependent variable delegation span on

the instrument. The coefﬁcient on phase execution is not sig-

niﬁcant in this test, which gives us conﬁdence in this variable

as an instrument for the ﬁrst-stage selection equation.

6As a robustness check, we also conducted the two stages of

analysis as separate regressions. We conducted a ﬁrst-stage

probit regression with all regression predictors in one model

and compared the results to a model that includes our explana-

tory and control variables excluding the instrument. The addi-

tion of the instrumental variable phase execution improves

the goodness of ﬁt of the model from 2=81757 to 11,349,

which suggests that it may serve as a good instrument. In

addition, we compared the results from the second-stage Heck-

man model for the dependent variable delegation span with

the results of an ordered logit regression (since our span vari-

able is incremental and ranges from 0 to 3). In this step we

reduced the sample to 287 observations, corresponding to the

cases when authority is split between the FAH and RAH. The

results of these additional analyses are qualitatively equivalent

to the results that we report in Tables 3 and 4.

7Recall that we expect that with more people collaborating on

a task that needs to be coordinated, there is less likely to be an

authority split. Interestingly, we ﬁnd that for the sample that

does split (see Table 4), when there is an increase in the num-

ber of contributing colleagues, this increases (has a positive

association with) delegation span. Our conjecture is that the

FAH is more comfortable with having a greater span distance

because there are also usually many others (as informed task

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

members) monitoring the task. To conﬁrm our speculation, we

tested the interaction of count informed colleagues and count

contributing colleagues, and although the coefﬁcient estimate

was positive, it was not signiﬁcant.

8In addition, we conducted a robustness analysis controlling

for alternative explanations that may affect current delegation

decisions. These alternatives include the FAH’s prior experi-

ence with delegation and handling multiple tasks, the absolute

hierarchy level of the FAH (the number of hierarchy levels

separating the FAH from the CEO), personnel responsibility,

ﬁrm growth, and task diversity (task diversity captures in how

many task groups the task’s FAH has decision authority), and

the FAH’s presence in the planning and execution phases (an

indication of whether the FAH is responsible for tasks in both

the planning and execution phases of the project). These tests

conﬁrm the intuition that the FAH learns to delegate and is

more likely to delegate because she has been exposed to del-

egation in the past. Our main explanatory variables still dis-

play statistically signiﬁcant coefﬁcients, which demonstrates

that our key results hold after controlling for these alterna-

tive explanations. The most prominent exception is that the

number of the FAH’s decisions loses signiﬁcance when we

control for the FAH’s prior delegation experience, which con-

ﬁrms the above result that FAHs learn to delegate over time.

See the online appendix (available as supplemental material

at http://dx.doi.org/10.1287/orsc.2014.0954) for a regression

table.

References

Aghion P, Tirole J (1997) Formal and real authority in organizations.

J. Political Econom. 105(1):1–29.

Baldwin CY, Clark KB (2000) Design Rules: The Power of Modular-

ity, Vol. 1 (MIT Press, Cambridge, MA).

Barnard C (1938) The Functions of the Executive (Harvard University

Press, Cambridge, MA).

Bartlett CA, Ghoshal S (1997) The myth of the generic manager:

New personal competencies for new management roles. Calif.

Management Rev. 40(1):92–116.

Beatty CA, Lee GL (1992) Leadership among middle managers: An

exploration in the context of technological change. Human Rela-

tions 45(9):657–990.

Becker G, Murphy K (1992) The division-of-labor, coordination costs

and knowledge. Quart. J. Econom. 107(4):1137–1160.

Bénabou R, Tirole J (2003) Intrinsic and extrinsic motivation. Rev.

Econom. Stud. 70(3):489–520.

Billinger S, Stieglitz N (2009) Organizational tradeoffs and the dynam-

ics of adaptation in permeable structures. Bøllingtoft A, Håkons-

son DD, Nielsen JF, Snow CC, Ulhøi J, eds. New Approaches to

Designing and Organizing (Springer, Berlin), 43–60.

Bloom N, Sadun R, Van Reenen J (2010a) Does product market

competition lead ﬁrms to decentralize? Amer. Econom. Rev.

100(2):434–438.

Bloom N, Sadun R, Van Reenen J (2010b) Recent advances in

the empirics of organizational economics. Annual Rev. Econom.

2:105–137.

Bolton P, Dewatripont M (1994) The ﬁrm as a communication net-

work. Quart. J. Econom. 109(4):809–839.

Chowdhury S (2011) The moderating effects of customer driven com-

plexity on the structure and growth relationship in young ﬁrms.

J. Bus. Venturing 26(3):306–320.

Christensen M, Knudsen T (2010) Design of decision-making orga-

nizations. Management Sci. 56(1):71–89.

Collins PD, Ryan LV, Matusik SF (1999) Programmable automa-

tion and the locus of decision-making power. J. Management

25(1):29–53.

Colombo MG, Delmastro M (2004) Delegation of authority in

business organizations: An empirical test. J. Indust. Econom.

52(1):53–80.

Colombo MG, Delmastro M (2008) The Economics of Organizational

Design: Theoretical Insights and Empirical Evidence (Palgrave

Macmillan, Basingstoke, UK).

Csaszar FA (2012) Organizational structure as a determinant of per-

formance: Evidence from mutual funds. Strategic Management

J. 33(6):611–632.

Cyert R, March J (1963) A Behavioral Theory of the Firm (Prentice

Hall, Englewood Cliffs, NJ).

Dessein W (2002) Authority and communication in organizations.

Rev. Econom. Stud. 69(4):811–838.

Diefenbach T, Sillince JAA (2011) Formal and informal hierarchy in

different types of organization. Organ. Stud. 32(11):1515–1537.

Fehr E, Herz H, Wilkening T (2013) The lure of authority: Motiva-

tion and incentive effects of power. Amer. Econom. Rev. 103(4):

1325–1359.

Foss NJ (2003) Selective intervention and internal hybrids: Interpret-

ing and learning from the rise and decline of the Oticon spaghetti

organization. Organ. Sci. 14(3):331–349.

Foss K, Foss NJ, Vázquez XH (2006) “Tying the manager”s hands”:

Constraining opportunistic managerial intervention. Cambridge

J. Econom. 30(5):797–818.

Gagné M, Deci EL (2005) Self-determination theory and work moti-

vation. J. Organ. Behav. 26(4):331–362.

Galbraith JR (1973) Designing Complex Organizations (Addison

Wesley, Reading, MA).

Gavetti G, Levinthal D, Ocasio W (2007) Neo-Carnegie: The

Carnegie School’s past, present, and reconstructing for the future.

Organ. Sci. 18(3):523–536.

Grant RM (1996) Toward a knowledge-based theory of the ﬁrm.

Strategic Management J. 17(S2):109–122.

Greene W (2003) Econometric Analysis, 5th ed. (Prentice Hall, Upper

Saddle River, NJ).

Harris M, Raviv A (2002) Organization design. Management Sci.

48(7):852–865.

Hayek FA (1945) The use of knowledge in society. Amer. Econom.

Rev. 35(4):519–530.

Heckman JJ (1979) Sample selection bias as a speciﬁcation error.

Econometrica 47(1):153–161.

Holmström B (1979) Moral hazard and observability. Bell J. Econom.

10(1):74–91.

Jensen M, Meckling W (1976) Theory of the ﬁrm: Managerial behav-

ior, agency costs and ownership structure. J. Financial Econom.

3(4):305–360.

Jensen M, Meckling W (1992) Speciﬁc and general knowledge, and

organizational structure. Werin L, Hijkander H, eds. Contract

Economics (Basil Blackwell, Cambridge, MA), 251–274.

Kendrick T (2006) Results Without Authority: Controlling a Project

When the Team Doesn’t Report to You—A Project Manager’s

Guide, 1st ed. (AMACOM, New York).

Dobrajska, Billinger, and Karim: Delegation Within Hierarchies

Khandwalla PN (1973) Viable and effective organizational designs of

ﬁrms. Acad. Management J. 16(6):481–495.

Kogut B, Zander U (1992) Knowledge of the ﬁrm, combinative

capabilities, and the replication of technology. Organ. Sci. 3(3):

383–397.

Lawrence PR, Lorsch JW (1967) Organization and Environment

(Harvard Business Press, Boston).

Lincoln JR, Hanada M, McBride K (1986) Organizational struc-

tures in Japanese and U.S. manufacturing. Admin. Sci. Quart.

31(3):338–364.

Mantere S (2008) Role expectations and middle manager strategic

agency. J. Management Stud. 45(2):294–316.

March J, Simon H (1958) Organizations (John Wiley & Sons,

New York).

McCann JE, Gilmore TN (1983) Diagnosing organizational decision

making through responsibility charting. Sloan Management Rev.

24(2):3–15.

MendelsonH(2000)Organizationalarchitecture and success in the infor-

mation technology industry. Management Sci. 46(4):513–529.

Mihm J, Loch CH, Wilkinson D, Huberman BA (2010) Hierarchical

structure and search in complex organizations. Management Sci.

56(5):831–848.

Mintzberg H (1979) The Structuring of Organizations: A Synthesis of

Research, 1st ed. (Prentice Hall, Englewood Cliffs, NJ).

Mookherjee D (2006) Decentralization, hierarchies, and incen-

tives: A mechanism design perspective. J. Econom. Literature

44(2):367–390.

Nonaka I (1994) A dynamic theory of knowledge creation. Organ.

Sci. 5(1):14–37.

Ocasio W (1997) Towards an attention-based view of the ﬁrm. Strate-

gic Management J. 18(S1):187–206.

Ocasio W (2011) Attention to attention. Organ. Sci. 22(5):1286–1296.

Ouchi WG, Dowling JB (1974) Deﬁning the span of control. Admin.

Sci. Quart. 19(3):357–365.

Patacconi A (2009) Coordination and delay in hierarchies. RAND J.

Econom. 40(1):190–208.

Radner R (1992) Hierarchy: The economics of managing. J. Econom.

Literature 30(3):1382–1415.

Rantakari H (2011) Organizational design and environmental volatil-

ity. J. Law, Econom., Organ. 29(3):569–607.

Rantakari H (2012) Employee initiative and managerial control. Amer.

Econom. J.: Microeconomics 4(3):171–211.

Rivkin JW, Siggelkow N (2003) Balancing search and stability: Inter-

dependencies among elements organizational design. Manage-

ment Sci. 49(3):290–311.

Rus D, van Knippenberg D, Wisse B (2012) Leader power and self-

serving behavior: The moderating role of accountability. Lead-

ership Quart. 23(1):13–26.

Siggelkow N, Levinthal DA (2003) Temporarily divide to con-

quer: Centralized, decentralized, and reintegrated organizational

approaches to exploration and adaptation. Organ. Sci. 14(6):

650–669.

Siggelkow N, Rivkin JW (2005) Speed and search: Designing orga-

nizations for turbulence and complexity. Organ. Sci. 16(2):

101–122.

Simon HA (1947) Administrative Behavior: A Study of the Decision-

Making Process in Administrative Organization (Macmillan,

New York).

Simon HA (1955) A behavioral model of rational choice. Quart. J.

Econom. 69(1):99–118.

Simon HA (1982) Designing organizations for an information-rich

world. Models of Bounded Rationality: Economic Analysis and

Public Policy, Vol. 2 (MIT Press, Cambridge, MA), 171–185.

Sirmon DG, Hitt MA, Ireland RD, Gilbert BA (2011) Resource

orchestration to create competitive advantage: breadth, depth,

and life cycle effects. J. Management 37(5):1390–1412.

Smith M, Erwin J (2005) Role and responsibility charting (RACI).

Project Management Forum 1–14. http://pmicie.org/images/

downloads/raci_r_web3_1.pdf.

Thompson J (1967) Organizations in Action: Social Science Bases of

Administrative Theory (McGraw Hill, New York).

Tushman ML, Nadler DA (1978) Information processing as an inte-

grating concept in organizational design. Acad. Management Rev.

3(3):613–624.

Vázquez XH (2004) Allocating decision rights on the shop ﬂoor:

A perspective from transaction cost economics and organization

theory. Organ. Sci. 15(4):463–480.

von Krogh G, Nonaka I, Rechsteiner L (2012) Leadership in organi-

zational knowledge creation: A review and framework. J. Man-

agement Stud. 49(1):241–277.

Wernerfelt B (2007) Delegation, committees, and managers. J.

Econom. Management Strategy 16(1):35–51.

Williamson OE (1996) The Mechanisms of Governance (Oxford Uni-

versity Press, Oxford, UK).

Wooldridge B, Schmid T, Floyd SW (2008) The middle manage-

ment perspective on strategy process: Contributions, synthesis,

and future research. J. Management 34(6):1190–1221.

Magdalena Dobrajska is an assistant professor in strate-

gic and international management at the Copenhagen Business

School. She received her Ph.D. in business administration from

the University of Southern Denmark. Her research interests

include strategic design of organizations, the boundaries of the

ﬁrm, and organizational decision making.

Stephan Billinger is an associate professor of strategic

organization design at the University of Southern Denmark.

His research interests include individual and group-based

search processes, delegation and decision making, hierarchy

and goal structures, ﬁrm boundary design, and social norms.

Samina Karim is an associate professor of strategy and

innovation at Boston University’s School of Management.

She received her Ph.D. from the University of Michigan

Ross School of Business. Her research interests include

the reconﬁguration of ﬁrms’ resources and market activities,

organization design, and management of interorganizational

relationships including acquisitions and alliances.

Impact of decentralized management on sickness absence in hospitals: a two-wave cohort study of frontline managers in Danish hospital wards

Article

Full-text available

Jul 2024
BMC HEALTH SERV RES

Background This study explores the impact of decentralized management on the sickness absence among healthcare professionals. Sickness absence is a reliable indicator of employees’ wellbeing and it is linked to management quality. However, the influence of decentralized management on sickness absence has not been adequately studied. Methods The research design combined a two-wave, web-survey of frontline managers in two Danish university hospitals with administrative data on sickness absence at the ward-level. The first and second wave included data from 163165 and 137 frontline managers linked to 121 wards and 108 wards. Data was analysed using an ordinal logistic regression model. Results Wards where frontline managers had the highest level of decentralised decision authority compared to none showed lower odds of ward-level sickness absence (ORcrude: 0.20, 95% CI: 0.05–0.87). A very high extent of cross-functional decision authority showed lower odds of sickness absence (ORcrude: 0.08, 95% CI: 0.01–0.49). Overall, the results showed a clear data trend, although not all results were statistically significant. Conclusion Higher levels of decentralized management in wards were positively associated with lower risks of sickness absence in hospital wards. The study supports future research on how to empower decision autonomy at the frontline level of management.

The Influence of Authority Delegation on Leadership Performance during COVID19 Pandemic

Research

Full-text available

Jul 2022

In 2019, there appeared a disease within china under the name of Covid-19. It is known as an infectious disease caused by the last detected virus of the Coronavirus series. There was no knowledge of this new virus and its disease before it began spreading in Wuhan, China, in December 2019. Covid-19 has now turned into a pandemic affecting many countries of the world. With the appearance of this virus, all countries started delegating authorities to its lower departments in order to increase its readiness to face the pandemic and manage the daily lives of people in the best way possible (Al-Jammal et al, 2015). Current study aimed at examining the influence of authority delegation in increasing the efficiency of leadership performance during COVID 19 pandemic within the Kuwaiti ministry of health in 2020. The Study adopted the quantitative approach through exposing total of (372) officials within Kuwaiti ministry of health to an internal questionnaire related to authority delegation and leadership performance[1]. Results of study indicated that there excites a positive influence of authority delegation variables including (Unity of Commands, Degree of Absoluteness, Leadership Empowerment and Function of Responsibility) in increasing the efficiency of leadership performance within the ministry. This influence – as according to results - appeared to be stringer in fields of leadership empowerment and functions of responsibilities referring to the areas of delegation and the expected results of such delegation, in addition to that, results indicated that authority delegation helped the state of Kuwait in facing and containing COVID 19 pandemic in the …

Environmental vertical management reform and data manipulation in the public sector: evidence from China

Article

Full-text available

Dec 2024

Based on monthly panel data from 2014 to 2020 and employing the staggered difference-in-differences (staggered DID) method, we examine the impact of environmental vertical management reform on data manipulation in the public sector. We reveal that environmental vertical management reform significantly reduces data manipulation in the public sector. Moderating effect analysis shows that economic growth targets weaken the inhibitory impact of this reform. Conversely, public environmental concerns could enhance the inhibitory impact of this reform on data manipulation. Mechanism analysis reveals that environmental vertical management reform works through strengthening grassroots environmental law enforcement. The increased independence of law-enforcing departments has reduced the tendency of local governments to engage in data manipulation.

Inclusion Management Practices as a Pathway to Enhance Organizational Resilience in Response to a Crisis: An Empirical Test in the Context of the COVID‐19 Pandemic

Article

Full-text available

Nov 2024
PERS PSYCHOL

Organizations inevitably face various forms of disruptive events (e.g., external crises), and sustaining long‐term prosperity requires them to stay resilient when encountering unexpected adversity. Prior crisis management research predominantly relied on qualitative case studies to examine efforts after a crisis had occurred, treating the crisis as a “given” rather than a variable. The exceptionality of crisis situations and the ad hoc nature of crisis countermeasures largely limit current knowledge about how organizations may manage employees to remain in a preparative stance for disruptive events. Integrating the inclusion literature, crisis management research, and event system theory, we propose inclusion management practices as a viable pathway for organizations to develop resilience resources and capabilities prior to a crisis, allowing them to exhibit greater robustness and agility when a crisis arises. Such robustness and agility, in turn, enhance organizational performance thereafter. We further pinpoint the strength of a crisis event as an important contingency shaping the effects of pre‐crisis inclusion management practices on organizations’ resilient responses and thereby performance. We tested our hypotheses in the context of the COVID‐19 pandemic crisis using longitudinal manager‐report survey data (N = 884 workplaces). We found that workplaces that implemented more inclusion management practices before COVID‐19 were more robust and agile in response to the pandemic crisis. Agility (but not robustness), in turn, was positively related to organizational performance. In addition, the effect of inclusion management practices on agility was stronger for workplaces with greater COVID‐19 event strength.

How does digital transformation affect corporate governance paradigms? A synthesis of the literature

Article

Full-text available

Nov 2024

At present, digital transformation has become a vital way for companies to achieve sustainable growth. This paper reviews the literature on the correlation between digital transformation and corporate governance paradigm, analyzes the specific impact mechanism of AI technology and big data technology in the field of corporate governance, and explores the influence effect of the most popular ChatGPT technology on corporate governance from the perspective of business practice. It is found that digital transformation has an important impact on stakeholder management, information disclosure, green governance and other aspects of corporate governance. The purpose of the study is to provide a new reference for the construction of corporate governance paradigm and help companies achieve long-term development in the digital wave.

Organization Design: Current Insights and Future Research Directions

Article

Sep 2024

We review the research on organization design from 2000 to 2023, inclusive. We identify four major approaches to organization design in the contemporary literature: configuration, control, channelization, and coordination. We discuss the key streams of research that characterize each of these approaches, as well as three emerging areas of research: AI and organizational decision-making, flat organizations, and multiple goals. Beyond the specific contributions of individual papers and streams of work, our review makes a number of high-level observations across approaches. We identify patterns that characterize this body of work, the methods used, open questions for future research, and a discussion of organization design as a theory. Collectively, these observations define the state of organization design research and may provide scholars with a foundation for future research.

Categorizing the Complexity: A Scoping Review of Structures Within Organizations

Article

Sep 2024

Structures involve a patterned regularity of interactions and frameworks that guide what individuals work on, with whom, and who influences those decisions. A deeper understanding of structures that exist within organizations has begun to emerge and illuminate new forms of structures (over 100 of them) that drive behavior in organizations. In this scoping review, we organize the fragmented insights on structure within organizations into a unifying framework that provides a coherent foundation for the domain by identifying nine topic domains and offering a summary of each (i.e., authority structures, cognitive structures, communication structures, coordination structures, leadership structures, motivational structures, social structures, task structures, and temporal structures). Next, as multiple structures co-occur within organizations, we explore the connections across topic domains, including their combinations. Understanding the separate topic domains and their combinations enables researchers and practitioners to understand why employee behaviors are inconsistent with the behaviors endorsed by a particular structure and better navigate the inherent complexity of structures within organizations. Finally, we outline implications for future work featuring structure combinations as well as emergent areas from the topic domains, such as the potential for change. Given the ubiquity of structures in organizations and their links with a variety of theoretical domains, this article’s implications have the potential to benefit a wide range of scholars and managers.

Keeping The Iron Cage from Closing: How Clashes Over Shared Principles Elicit Collective Authoring

Article

Jun 2024

When less hierarchical organizations attempt to avoid positional authority and rely more on the authority of shared principles, they are challenged to balance a relative open-endedness and sufficient consensus of those principles. However, situations inevitably arise when assertions about the principles clash, disturbing the sense of collective sharing of principles and so triggering tendencies to make the principles more rigid or default to positional rule. This paper explores how internal collective dialogue unfolds in such situations in the central online channels of three mid-sized Nordic software consultancies aspiring for less hierarchical, principles-based organizing. Our research reveals the emergence of ‘collective authoring’— peripheral participation that reacts to the distress stirred by the clash. The collective responds with tension-mediating remarks and, most markedly, with a polyphony of inquiries and musings that are attentive to what is unfolding but not stubbornly fortified with either side of the clash. We discover how this collective dialogue comes to embed the initial clash with the iteratively maturing and saturating soundscape of deliberations, gradually invoking a tacit sense of the collective understanding of the issue on the channels. The dialogues persist until the most acute sense of distress over the sharing of principles seems to dissolve. We discuss how our findings, by illuminating the role of ‘the collective,’ contribute to the extant conceptions of relational authority and less hierarchical organizing.

Digital transformation and the allocation of decision-making rights within business groups – Empirical evidence from China

Article

Jun 2024
J BUS RES

Information processing style and institutional trust as factors of COVID vaccine hesitancy

Article

Full-text available

May 2024

This study investigates the factors contributing to COVID vaccine hesitancy. Vaccine hesitancy has commonly been attributed to susceptibility to misinformation and linked to particular socio-demographic factors and personality traits. We present a new perspective, emphasizing the interplay between individual cognitive styles and perceptions of public health institutions. In January 2020, before the COVID-19 pandemic, 318 participants underwent a comprehensive assessment, including self-report measures of personality and clinical characteristics, as well as a behavioral task that assessed information processing styles. During 2021, attitudes towards vaccines, scientists, and the CDC were measured at three time points (February–October). Panel data analysis and structural equation modeling revealed nuanced relationships between these measures and information processing styles over time. Trust in public health institutions, authoritarian submission, and lower information processing capabilities together contribute to vaccine acceptance. Information processing capacities influenced vaccination decisions independently from the trust level, but their impact was partially mediated by authoritarian tendencies. These findings underscore the multifactorial nature of vaccine hesitancy, which emerges as a product of interactions between individual cognitive styles and perceptions of public health institutions. This novel perspective provides valuable insights into the underlying mechanisms that drive this complex phenomenon.

Allocating decision rights on the shop floor: A perspective from transaction cost economics and organization theory

Article

Full-text available

Jul 2004

Xose H. Vazquez

This paper explores the links that the allocation of decision rights on the shop floor maintains with labor transaction attributes and several structural traits of the firm. The approach is based on the transaction cost apparatus and harnesses the theoretical and empirical background provided by organization theory. Data are presented from a wide field survey in the Spanish food and electronics industries (Standard Industry Classification (SIC) 20 and 36). Evidence not only verifies the influence of firm size, property, age, and unionism, but also shows that the allocation of decision rights is related to a particular mix of labor transaction traits. Specifically, one of the most important results is that employer opportunism offers greater explanatory power than employee opportunism.

Design Rules: The Power of Modularity

Book

Jan 2000

Specific and general knowledge, and organizational structure

Article

Jan 1992

The economics of organizational design: Theoretical insights and empirical evidence

Book

Jan 2008

This book synthesizes the existing quantitative evidence on organizational design, bringing together the growing stream of theoretical work on the economics of firms' organization and available empirical findings produced by qualitative work in management, business history and organizational sociology. © Massimo G. Colombo and Marco Delmastro 2008. All rights reserved.

Programmable Automation and the Locus of Decision-Making Power

Article

Feb 1999

Computer-based technology is often credited with making decentralized decision-making possible, helping firms to respond rapidly to changing marker conditions. Research on this subject, however, shows contradictory effects: some studies support decentralization and others support centralization. This longitudinal study examines how one form of computer-based technology, programmable automation (PA), affects centralization. Unlike previous studies, it attempts to clear up some of the confusion surrounding technology's effect on centralization by distinguishing between strategic and operating decisions, and between decision-making authority and influence. As expected, PA flattened the hierarchy. It had no effect on strategic decision-making power, but surprisingly, did influence centralized line-operating authority and influence. This finding is particularly striking because firms with decentralized line-operating decision-making are more likely to adopt PA.

Selective intervention and internal hybrids: Interpreting and learning from the rise and decline of the Oticon spaghetti organization

Article

May 2003

Nicolai J. Foss

Infusing hierarchies with elements of market control has become a much-used way of simultaneously increasing entrepreneurialism and motivation in firms. However, this paper argues that such "internal hybrids," particularly in their radical forms, are inherently hard to successfully design and implement because of a fundamental incentive problem of establishing credible managerial commitments to not intervene in delegated decision making. This theme is developed and illustrated, using the case of the world-leading hearing aids producer, Oticon. In the beginning of the 1990s, Oticon became famous for its radical internal hybrid, the "spaghetti organization." Recent work has interpreted the spaghetti organization as a radical attempt to foster dynamic capabilities by organizational means, neglecting, however, that about a decade later the spaghetti organization has given way to a more traditional matrix organization. In contrast, an organizational economics interpretation of Oticon organizational changes is developed. This lens suggests that a strong liability of the spaghetti organization was the above incentive problem: Frequent managerial meddling with delegated rights led to a severe loss of motivation, and arguably caused the change to a more structured organization. Refutable implications are developed, and the discussion is broadened to more general issues of economic organization.

Econometric Analysis

Article