Complexity theory and public management a ‘becoming’ field

Complexity theory and public management: a ‘becoming’ field
Since the special edition of Public Management Review on
‘Complexity Theory and Public Management’
in 2008 (Volume 10 (3)), co-edited by Geert Teisman and Erik-
Hans Klijn, academic interest in complexity
theory, and how it might be used to understand the world and
inform design and intervention in the
public policy/public management field, has grown and matured.
The inspiration for this special issue
arose out of intensive interactions among interested scholars in
conference panels (at American Society
for Public Administration, International Research Society for
Public Management, and the Challenges of
Making Public Administration and Complexity Theory group)
over the past few years and the realization
that a ‘stock-taking’ was required. While many public
management scholars knew a little bit about
complexity – and some knew a lot – there was still no consensus
about the contribution complexity
theory could or could not make to theory and practice. While we
did not achieve consensus this time

around, the papers selected for this edition provide a picture of
where we are and where scholars in this
field think we should go, and some examples of the most
promising routes to get there. Before
summarizing these findings, we provide a brief overview of
where we have come from and why we are
still a ‘becoming’ field.
Challenging fundamental assumptions
Nineteenth- and twentieth-century sciences which developed
beneath the umbra of Newtonian
theories, embedded some pervasive assumptions which might be
crudely summarized as (1)
relationships between individual components of any system can
be understood by isolating the
interacting parts, (2) there is a predictability to the relationship
among the parts, and (3) the result of
interactions and the working whole might eventually be
understood by simply summing the parts. So in
much the same way as the expert clockmaker might be able to
design, build, disassemble, and modify a
clock, understanding the individual parts and how they fit
together leads to understanding the

functioning whole and the capability to replicate it precisely as
required. This paradigm is dominated by
mechanical metaphors and leads to an assumption that the sum
of the parts equals the whole.
Dissatisfaction with the limitations of mechanical explanations
led to more sophisticated models which
were better at explaining the observed behaviour, initially of the
physical world, and then increasingly
the biological, ecological, and social worlds (e.g. Byrne 1998;
Cilliers 1998; Holland 1995; Kauffman
1993; Prigogine 1978; Prigogine and Stengers 1984; Stacey
1993; Waldrop 1992). Such modelling offered
new ontological insights about the nature of our world and the
way it behaves. This is summed up
briefly by saying that there are recursive, ongoing non-linear
interactions between the elements that
make up the whole and these elements adapt to each other in
non-linear ways. Their interactions create
contingency and uncertainty about what the future will become.
As a result, the whole lacks the
predictability of the machine model. Boulton (2010) refers to a
complex world view as ‘becoming’
because individual components in these worlds are
interdependent and in processes of ongoing

interaction with each other with the result that the world is not
static and fixed, but dynamic, ever-
changing, and becoming something different from what it was in
the past. Recognition of such inherent
uncertainty leads to a conclusion that Newtonian-like
mechanical models are inadequate for these types
of systems because the sum of the parts does not equal the
whole. Understanding of the whole cannot
be based only on an understanding of the disaggregated parts
because of the ongoing non-linear change
caused by the interactions between the parts. This shift in
understanding brings us to a complexity
world view: ‘sandwiched between a view that the world works
like a machine and a view that the world
is chaotic, unpredictable and without structure’ (Boulton, Allen,
and Bowman 2015, 29).
In this complexity-informed world view, ongoing non-linear
interactions result in macro patterns
becoming established. Complexity theory explains the way
many, repeated non-linear interactions
among elements within a whole result in macro forms and
patterns which emerge without design or

direction. Further, an initial pattern might be disrupted by
external events or internal processes and
reform into some new pattern. Boulton and colleagues sum up
what they call the ‘central tenet of
complexity theory’ and its contribution to understanding change
as ‘the detail and the variation’ of each
action – the effect of a regulation on various actors for example
– ‘coupled with the interconnection’ of
action and environment that ‘provide the fuel for innovation,
evolution and learning’ (Boulton, Allen,
and Bowman 2015, 29). That is, the future is a contingent,
emergent, systemic, and potentially path-
dependent product of reflexive non-linear interactions between
existing patterns and events. Its variety,
diversity, variation, and fluctuations can give rise to resili ence
and adaptability; is path dependent,
contingent on local context and on the sequence of what
happens; subject to episodic changes that can
tip into new regimes; has more than one future; can self-
organize, self-regulate; and have new features
emerge.
Introducing a complexity frame to public management

As an alternative to Newtonian mechanics, this last observation
about the contribution of complexity
theory for understanding unpredictability and change in human
systems leads us to its relevance for the
study of public policy and public management. Scholars and
practitioners of public policy and public
management are concerned with how to create or change
particular patterns of interaction between
actors to get a particular result: for example, how might
governments design a set of institutions to
bring about certain behaviours; or given a set of institutions,
how might the interactions between actors
and the institutions be governed to achieve a particular
outcome; and how might unintended negative
effects be avoided or positive ones enhanced? Furthermore,
complexity theory facilitates a focus on
multiple levels of scale simultaneously. Thus the individual
actors, and multiple layers of institutions of
varying complexity which interact, can all be brought into view
through the multi-scalar complexity lens.
We note, within the diverse scientific traditions of public policy
and public management theories,
attempts to explain dynamism and non-linear contingency in
how change takes place have become an

increasingly pertinent concern (Eppel 2017). In the last 20 years
– and rising sharply from around 2008
(Gerrits and Marks 2015) – we see increasingly explicit use of
complexity theory concepts for explaining
the way the public policy/management worlds behave and how
we might better design and manage
change in these worlds. David Byrne has also deepened our
understanding of the methodological
implications of complexity for the social sciences generally
(Byrne, 1011, Byrne and Callaghan 2014).
Scholars such as Sanderson (2009), Room (2011), and Morcol
(2012) have all argued for complexity
theory for understanding of how the social world of policy
processes work. Cairney (2012, 2013; Cairney
and Geyer, 2017) caution us that the looseness with which
complexity concepts are sometimes applied
could be an impediment but they also see a place for complexity
theory as a bridge between academic
and policymaker perspectives in support of pragmatism and
insights about how to influence emergent
behaviour. Sanderson (2009) advocates that the ambiguity and
uncertainty arising from a complex

adaptive world can be mitigated through the use of an
epistemology based on pragmatism and
complexity theory. Room (2011) suggests a blending of extant
theories such as institutionalism with
complexity theory for better understanding the micro/macro
dynamics of public policy. He suggests that
there is a complementarity in which complexity theory supplies
the micro mechanisms lacking in
institutional theory and institutional theory supplies a macro
framing specific to public policy which
complexity theory lacks. Morcol (2012) argues further that
complexity theory provides mechanisms and
concepts for understanding the macro/micro problems at the
heart of public policy process. That is,
complexity theory provides a micro mechanism for explaining
the macro patterns of interest to public
policy scholars. Growing interest in complexity and policy is
evidenced in the establishment of a new
Journal on Policy and Complex Systems in 2014.
In a parallel and consistent vein, Teisman and colleagues in the
Netherlands (Teisman, van Buuren, and
Gerrits 2009), Rhodes and colleagues in Ireland (Rhodes et al.
2011), Koliba and colleagues in the United

States (Koliba, Meek, and Zia 2011), and Eppel and colleagues
in New Zealand (Eppel, Turner, and Wolf
2011) have each employed complexity theory concepts to better
understand the core processes of
public management such as agenda setting, policy formation,
decision-making, and implementation.
These authors have more or less independently come to the
conclusion that complexity theory and
network theory are required and should be linked together to
provide an adequate basis on which to
develop governance theory and practice guidelines in modern
public management contexts. The extent
of complementarity between complexity theory and network
governance (Klijn and Koppenjan 2014;
Koppenjan and Klijn 2014) and new public management
theories is reflected in the establishment of the
journal Complexity, Governance and Networks in 2014.
Others have taken aim at how public sector change might be
better managed generally by enlisting
complexity thinking and concepts to inform processes of
designing and generating change (Boulton,
Allen, and Bowman 2015; Geyer and Rihani 2010; Innes and
Booher 2010). These authors identify

common themes such as the impossibility of prediction and
therefore the need to adopt more
experimental approaches to intervention based on the
assumption that there will be new phenomena
(unknown unknowns) likely to emerge endogenously. What has
occurred previously will continue to
affect the present (and the future). As a result, any externally
applied change will have uncertain effects,
some of which will lead to a helpful change and some not so.
Doing public policy and public
management in such a world requires cognisance of the above
characteristics – and particularly the
dynamics of self-organization, path-dependency, adaptation, and
emergence – in how we approach
policy and change (Rhodes et al. 2011). We also need
complexity’s lens to see the whole while taking
into account the relationships between the elements at different
levels of scale. Koliba and Zia (2012)
talk about the need for complexity friendly methods for
modelling the complex governance system.
Innes and Booher (2010) built their theory of collaborative
rationality for public policy on analysis of the

ongoing dialectic interaction between collaboration and praxis
as a means for understanding complex
change. Cairney and Geyer (2015) have made a substantial
contribution to thinking about the
contribution of complexity theory to policy studies and how it
might add to understanding of particular
policy fields, such as health (Tenbensel 2013) or concepts such
as power (Room, 2015) as well as
complexity friendly methods for research and practice.
Overview of papers in this edition
This plethora of contributions and theoretical explorations cries
out for framing and assessment to help
guide scholars engaging with complexity in the public
management/policy domains. To that end, our call
for contributions asked authors to consider how complexity
contributes to public management theory
and practice using one (or more) of three lenses: (1) complexity
theory-informed alternative
perspectives on the framing of problems and design of processes
of public administration to be
considered, (2) insights into alternative institutions that are
shaping public administration and
management processes, and (3) alternative practices to match

the complexity of the environment and
the challenges faced by public management scholars
administrators.
Furthermore, we note the need for a distinction to be made
between the use of complexity theory to
create and test concepts and theories to describe the world as it
is (which is often the domain of the
natural sciences), and the use of these concepts and theories to
design and bring about change (this
latter often the domain of social sciences). While these
perspectives inform each other, they often rely
on different ontological and epistemological foundations, and
this is apparent in the papers in this
special edition where we see both describe and design features
in the way authors have used
complexity theory.
Alternative perspectives
Alternative perspectives provided by complexity theory have
evolved markedly in the intervening years
between this issue and the last special issue of PMR addressing
complexity. We have already mentioned
the application of complexity concepts to understanding multi -

actor decision-making and institutional
change for instance. The authors in this issue further explore
models which attempt to incorporate the
specific use of complexity concepts such as feedback loops,
adaptation, attractors, and emergence to
reframe understanding of common phenomena experienced in
public administration such as policy
processes, implementation, natural resources management, and
public-sector reform.
In all of the papers in this issue, there is the explicit recognition
that a complexity perspective entails the
rejection of assumptions of predictability and control in public
management, and the adoption of
assumptions of multiple, interacting self-organizing entities that
learn and change over time. While
there are periods of stable behaviour and features of the system
that function as constraints on
elements of the system, the diversity and adaptation of entities
creates the possibility for both
evolutionary and unpredictable, sudden change.

An example of two inter-country independent decision-making
processes that became coupled over
time is used by Marks and Gerrits to illustrate the contribution
of game theoretic models to
understanding complex public administration processes. Their
game theory model is tested through an
experiment aimed at explaining how representatives of the two
governments involved who met each
other in two presumed independent decision-making arenas took
the history of their interactions from
one to the other, thereby influencing the overall outcome. Thus
they demonstrate the interdependency
and connectedness between systems that otherwise might be
assumed independent. Further, the
authors provide a testable formalized model that describes the
interaction and co-evolution of
independent agents over time for future scholars to build upon.
Haynes makes use of complexity theory to focus on multiple
levels of public administration systems. He
extends the conceptualization of the public administration
complex system to include the behaviour
disposition of the individual in relation to their public and
personal values, to conclude that the multi-

level capacity in complexity theory is, in part, bounded by
public service values. Further, he uses the
complexity concept of attractors to explain how public service
values at different levels (individual,
family/community, professional, and political) can play a role
in constraining (or indeed enabling) system
change over time. Both Haynes and Marks & Gerrits extend the
understanding of complex adaptive
systems (CAS) theory and public management by taking their
analysis of participating actors below the
level of description of the organization and the institutions.
They consider the largely unconscious
psychological dispositions of individual actors and their history
with other actors and its influence on
patterns of institutional and organizational decision-making
which are relevant to the design.
Rather than develop new models, Rhodes and Dowling assess to
what extent fitness landscape models
(Wright 1932; Kaufmann and Levin 1987) have been used
effectively by public management scholars to
date through a systematic review. Fitness landscapes are
evolutionary models that capture how the
behaviour and characteristics of independent agents operating in
a shared context result in individual

and system-wide outcomes. The authors remark on their
frequent use at the level of metaphor and the
limited attention paid to mapping the concepts of the model to
the features of the empirical
phenomenon being described. This conclusion might easily be
applied to a number of other complexity
concepts (Cairney and Geyer 2017), which, after several
decades of scholarly effort, raises concerns
about the translation of these concepts into the public
management domain. Nevertheless, Rhodes and
Dowling conclude that in combination with network theory,
fitness landscape models are ‘more aligned
with the actual features of complex governance systems than
game theory models which rely on highly
stylized assumptions about how agents behave and equally
fuzzy definitions of performance’ (Rhodes
and Dowling, this issue). We return to these ‘fuzzy definitions
of performance’ in our conclusion.
Alternative institutions
Alternative institutions are those that can influence the actions
of interdependent, autonomous agents

as they iteratively explore alternative solutions to wicked
problems, such as distributed authority
arrangements, multi-sector for a for decision-making and multi-
channel feedback arising from new
communication technologies. For example, in Haynes’
contribution, the notions of public service values
and public value are explored through the lens of CAS theory.
The paper offers a concrete and practical
example for understanding the dynamic influence of values on
complex policy systems. Haynes argues
for recognition of ‘soft’ patterns of values such as belief
systems and their dynamic influence on
organizational behaviours as well as ‘hard’ patterns such as
rules and structures and shows how the CAS
lens enables this.
Castlenovo and colleagues attend to the issues raised by the
federal–state–local governance structures
and how these might be re-imagined/understood using
complexity theory. For them, their complexity-
based lens acts as a heuristic device to understand the
misalignment of locally implemented outcomes
with the centrally defined objectives of a nationwide public
programme in Italy where the ‘Napoleonic’

administrative traditions dominate – arguing for a rethinking of
these traditions.
Tenbensel, rather than arguing for a particular type of
institutional change, builds on the approach taken
by Room (2011) and advocated by Cairney and Geyer (2017) in
bringing institutional theory together
with complexity theory using Crouch’s concept of recombinant
governance. Through an examination of
the fitness of various governance hybrids in the health sector in
New Zealand he demonstrates the
usefulness of being able to distinguish among various versions
of hybridity and to argue for a more
evolutionary perspective on institutional design and change.
Alternative practices
Complexity offers alternative ways of framing intervention and
bringing about successful change that
navigates the traps of unexpected changes and opens up
different ways of achieving innovation. Gear
and colleagues take us into the conceptual framing and research
methodology needed to examine the
complex problem of intimate partner violence (IPV). They
identify the limitations faced in developing

healthcare interventions in the absence of a complex adaptive
systems view. Existing efforts to
understand sustainable approaches in primary healthcare
settings have been dominated by the direct
cause–effect thinking reflected in randomized control trials and
like methodologies that have been so
prevalent in health research. Reframing the person entrapped by
IPV and their world, and the world of a
primary healthcare setting as two interacting complex adaptive
systems, shifts the research focus to the
reflexive interactions that occur between the person
experiencing IPV and the primary healthcare
setting. According to CAS theory, we would expect these
interactions to lead to mutual adaptations
within each of these complex systems, and therefore
intervention sustainability will occur when the
interaction and mutual adaptation generate outcomes that
stimulate ongoing engagement by both
systems. Without the CAS perspective, the self-organization,
coevolution, and emergence that leads to
sustainability cannot be studied. The conceptualization and
research design developed to study
healthcare responses to IPV might also be more widely

applicable to other complex social interventions.
Sustainability of the collaborative governance network is also
the focus of Scott and colleagues.
Complexity theory concepts are used to both describe how
sustainability is linked to the adaptability
and flexibility of the collaborative project but also to offer
insights into how the collaborative process
might be designed to encourage the development of
sustainability. Like many other papers in this
edition, their use of complexity theory is combined with other
theories – collaborative governance, in
this instance.
Meek and Marshall use a CAS lens to understand how the multi -
actor institutional governance of a
complex Southern Californian metropolitan water system
contributes to an adaptive resilience able to
respond effectively to the external stressors of severe and
sustained drought. Ongoing self-organization
and adaptation within and among the governance actors and
other stakeholders are characteristics of
the governance system which lead to emergent features which
help maintain resilience.

In the Castelnovo paper referred to already, we encounter the
empirical descriptions needed to
interpret the complexity factors that shaped an implementation
trajectory. They offer self-organization,
co-evolution, and emergence as mechanisms for understanding
the peculiar implementation path which
might otherwise be assumed to be the cumulative effect of a
series of legislative interventions not
always coherent in and among themselves. In so doing they
pave the way for the design of alternative
implementation practices.
Finally, scholar-teachers have also begun to incorporate
complexity theory into teaching practice. It has
proved useful for both integrating theories and for helping
students and practitioners to better frame
and understand the challenges of public management. In schools
of government, planning, and business
we are starting to see individual modules, components of
programmes, and indeed entire master’s
degrees being developed to introduce students to a complexity
‘perspective’ and to be exposed to the
tools and techniques to understand and intervene in complex
systems. Due to constraints of space, this

issue does not include any articles on this topic, but instead the
editors are working on a separate
special issue in ‘Complexity, Governance & Networks’
dedicated to the ways complexity is being taught
to public management/policy students around the world.
Whither complexity in public management?
The relevance of complexity theory for circumventing the
weaknesses of a mechanistic approach to
understanding public policy and management has been well -
trodden ground for decades. That this
continues to be pursued as complexity theory spreads across
policy domains suggests that it is this
fundamental capacity that is at the core of the attraction for
many scholars and practitioners. As
highlighted above, the use of complexity theory in public
management has developed both in relation to
the description of phenomena and design of institutions and
interventions to effect change.
From a theoretical perspective, the scholarship of the last
decade and the papers in this volume

demonstrate that complexity theory sits alongside, and in many
cases augments existing theories of
public policy and public management. Public policy and public
management draw on a variety of parent
disciplines such as politics, organization science, economics,
management, sociology, and psychology
(Raadschelders 2011) and bridging or integrating this plurality
continues to be an implicit – and in some
cases explicit – objective of scholars applying complexity
theory to this domain. A complexity
perspective can describe how interdependent agents interact
over time – within the constraints of
history, institutional forms, and/or values – to increase or
decrease overall (or individual) fitness,
sustainability, or resilience. It does this without the need to fall
back on predictable cause and effect
relationships among agents or contexts while still leaving room
for the identification of patterns and
likely pathways.
Furthermore, the ‘positive role for complexity theory as a way
to bridge academic and policy maker
discussions’ (Cairney and Geyer 2017, 1) – and we would add
‘practitioners’ – is evident in many of the

papers. Complexity acts as a challenge to the quest for certainty
in policymaking and also prompts
discussion about the role of pragmatism in policymaking. In this
issue, authors have argued for linking
complexity frameworks with institutional theory, network
theory, public value theory, and game theory
to better understand the dynamics of processes, outcomes, and
change in public policy/management
systems over time. Its strengths lie in its facilitation of a focus
on multiple levels of scale and its
provision of micro-level mechanisms for macro-level theories
such as institutional theory and
punctuated equilibrium theory (Eppel 2017). The key
mechanisms explored in this issue are based on
game theoretic interactions, search processes on fitness
landscapes, evolution arising from recombinant
novelty, and information exchange in networks – building on
the core complexity dynamics of self-
organization, adaptability, and emergence. In respect of
institutions, the conclusion one may draw from
these papers is that it is unlikely that current institutional forms
– whether they be hierarchical, market,
network, or values based – exhaust the range of potential
institutional forms that could be designed or

evolve in the public policy and administration space.
Experimenting with new forms would appear to be
an important complexity-friendly policymaking practice that
would lead to more sustainable public
systems.
The concepts of ‘sustainability’ and ‘resilience’ make an
appearance in several of the articles in this issue
as objectives of research and practice that are facilitated by a
complexity approach. However, there is
little agreement or indeed clear definition about what either of
these outcomes represent in the context
of public administration. Survival – or the ongoing existence of
agents, institutions, or systems if not of
the individual humans that make these up – is, of course, one
option, but this is not clarified or
challenged either in the papers in this issue or in the wider
academic community. It is incumbent upon
those scholars working in this area and using these concepts to
clearly define and debate what they
mean if the policy or practice recommendations arising from
their research are to be seriously
considered.

In addition to this definitional lacuna around sustainability and
resilience, the incorporation of
performance management research, theory and practice, has
been largely absent in the public
administration complexity literature. The fitness landscape
literature would appear to provide an
obvious link to performance, as evidenced by the use of the
phrase ‘performance landscapes’ to
describe this approach in organizational theory (Siggelkow and
Levinthal 2003; Rhodes and Donnelly-Cox
2008). This leads us to speculate about the compatibility of
complexity theory with our basic
understanding of the nature of performance management. The
issue may partially be due to the
multidimensionality of performance management (Bouckaert
and Halligan 2008) and the limitations of
how performance management has been conceived and practised
in the new public management
environment (Moynihan et al. 2011). Moynihan and colleagues
(2011) point to the limitations of current
research on performance management to take adequate
cognisance of governance complexity. So there
appears to be some room for each scholarly trajectory to learn

from the other.
But perhaps more important is that fact that we are still quite
far from developing complexity-based
models of agent interactions, behaviour, and change over time
that demonstrably produce/predict real-
world outcomes of any kind, not just performance. However, the
kind of direct cause and effect theories
we have come to believe represent the pinnacle of scholarly
achievement and the reliance on
experiments or random control trials to prove same are unlikely
to address the sorts of ‘wicked’
problems (Rittel and Webber 1973) that lie at the heart of public
policy and management. The need to
continue to adopt and refine complexity-informed theory,
institutions, and practice in a domain of
human endeavour as rich and varied as public administration is
as vital now as it was a decade ago.
Additional information
Notes on contributors
Elizabeth Anne Eppel
Elizabeth Anne Eppel is a Senior Research and Teaching Fellow
in the School of Government, Victoria

University of Wellington, New Zealand. Her research interests
are complexity in public policy processes,
governance networks, and collaborative governance.
Mary Lee Rhodes
Mary Lee Rhodes (B.A., M.Sc., MBA, Ph.D.) is an Associate
Professor of Public Management at Trinity
College, Dublin. Her research is focused on complex public
service systems and the dynamics of
performance. Prof. Rhodes has published numerous articles on
housing as a complex adaptive system
and her most recent book on Complexity and Public
Management was published by Routledge in 2011.
Her current research is on the nature and dynamics of social
impact and she is developing research on
social innovation, social finance, and well-being.
Previous article
View issue table of contents
Next article
References
Bouckaert, G., and J. Halligan. 2008. Managing Performance:

International Comparisons. Abingdon:
Routledge. [Google Scholar]
Boulton, J. G. 2010. “Complexity Theory and Implications for
Policy Development.” Emergence:
Complexity and Organisation 12 (2): 31–40. [Google Scholar]
Boulton, J. G., P. M. Allen, and C. Bowman. 2015. Embracing
Complexity: Strategic Perspectives for an
Age of Turbulence. Oxford: Oxford University Press.
[Crossref], [Google Scholar]
Byrne, D. S. 1998. Complexity Theory and the Social Sciences:
An Introduction. London: Routledge.
[Google Scholar]
Byrne, D. S. 2011. Applying Social Science: The Role of Social
Science Research in Politics, Policy and
Practice. Bristol: Policy Press. [Crossref], [Google Scholar]
Byrne, D. S., and G. Callaghan. 2014. Complexity Theory and
the Social Sciences: The State of the Art.
Abingdon: Routledge. [Google Scholar]
Cairney, P. 2012. “Complexity Theory in Political Science and
Public Policy.” Political Studies Review 10
(3): 346–358. doi:10.1111/j.1478-9302.2012.00270.x.
[Crossref], [Web of Science ®], [Google Scholar]

Cairney, P. 2013. “Standing on the Shoulders of Giants: How
Do We Combine the Insights of Multiple
Theories in Public Policy Studies.” Policy Studies Journal 41
(1): 1–21. doi:10.1111/psj.12000. [Crossref],
[Web of Science ®], [Google Scholar]
Cairney, P., and R. Geyer, eds. 2015. Handbook on Complexity
and Public Policy. Cheltenham: Edward
Elgar. [Google Scholar]
Cairney, P., and R. Geyer. 2017. “A Critical Discussion of
Complexity Theory: How Does ‘Complexity
Thinking’ Improve Our Understanding of Politics and Policy
Making?” Complexity, Governance and
Networks 3 (2): 1–11. doi:10.20377/cgn-56. [Crossref], [Google
Scholar]
Cilliers, P. 1998. Complexity and Postmodernism:
Understanding Complex Systems. London: Routledge.
[Google Scholar]
Eppel, E. 2017. “Complexity Thinking in Public
Administration’s Theories-in-Use.” Public Management
Review 19 (6): 845–861. doi:10.1080/14719037.2016.1235721.
[Taylor & Francis Online], [Web of
Science ®], [Google Scholar]

Eppel, E., D. Turner, and A. Wolf. 2011. “Complex Policy
Implementation: The Role of Experimentation
and Learning.” In Future State: Directions for Public
Management in New Zealand, edited by B. Ryan and
D. Gill, 182–212. Wellington: Victoria University Press.
[Google Scholar]
Gerrits, L., and P. Marks. 2015. “How the Complexity Sciences
Can Inform Public Administration: An
Assessment.” Public Administration 93 (2): 539–546.
doi:10.1111/padm.12168. [Crossref], [Google
Scholar]
Geyer, R., and S. Rihani. 2010. Complexity and Public Policy:
A New Approach to 21st Century Politics,
Policy and Society. Abingdon: Routledge. [Google Scholar]
Holland, J. H. 1995. Hidden Order: How Adaptation Builds
Complexity. Reading, MA: Addison-Wesley.
[Google Scholar]
Innes, J. E., and D. E. Booher. 2010. Planning with Complexity.
Abingdon: Routledge. [Crossref], [Google
Scholar]
Kauffman, S. A. 1993. The Origins of Order. Self Organisation
and Selection in Evolution. New York:

Oxford University Press. [Google Scholar]
Kauffman, S. A., and S. Levin. 1987. “Towards a General
Theory of Adaptive Walks on Rugged
Landscapes.” Journal of Theoretical Biology 128 (1): 11-45.
[Crossref], [PubMed], [Web of Science ®],
[Google Scholar]
Klijn, E.-H., and J. F. M. Koppenjan. 2014. “Complexity in
Governance Network Theory.” Complexity,
Governance and Networks 1: 61–70. doi:10.7564/14-CGN8.
[Crossref], [Google Scholar]
Koliba, C., and A. Zia. 2012. “'Complexity Friendly' Meso-level
Frameworks for Modelling Complex
Governance Systems.” In COMPACT 1: Public Administration
in Complexity, edited by L. Gerrits and P.
Marks, 119–139. Litchfield Park, AZ: Emergent. [Google
Scholar]
Koliba, C., J. W. Meek, and A. Zia. 2011. Governance Networks
in Public Administration and Public Policy.
Boca Raton, FL: CRC Press. [Google Scholar]
Koppenjan, J. F. M., and E.-H. Klijn. 2014. “What Can
Governance Network Theory Learn from
Complexity Theory? Mirroring Two Perspectives on
Complexity.” In Network Theory in the Public Sector:

Building New Theoretical Frameworks, edited by R. Keast, M.
Mandell, and R. Agranoff, 157–173.
London: Routledge. [Google Scholar]
Morcol, G. 2012. A Complexity Theory for Public Policy. New
York: Routledge. [Google Scholar]
Moynihan, D. P., S. Fernandez, S. Kim, K. M. LeRoux, S. J.
Piotrowski, B. E. Wright, and K. Yang. 2011.
“Performance Regimes Amidst Governance Complexity.”
Journal of Public Administration Research and
Theory 41 (Supplement 1): i114–i155.
doi:10.1093/jopart/muq059. [Crossref], [Web of Science ®],
[Google Scholar]
Prigogine, I. 1978. “Time, Structure and Fluctuations.” Science
201 (4358): 777–785.
doi:10.1126/science.201.4358.777. [Crossref], [PubMed], [Web
of Science ®], [Google Scholar]
Prigogine, I., and I. Stengers. 1984. Order Out of Chaos: Man’s
New Dialogue with Nature. Toronto:
Bantam Books. [Crossref], [Google Scholar]
Raadschelders, J. C. N. 2011. Public Administration: The
Interdisciplinary Study of Government. New
York: Oxford University Press. [Crossref], [Google Scholar]

Rhodes, M. L., and G. Donnelly-Cox. 2008. “Social
Entrepreneurship as a Performance Landscape.”
Emergence: Complexity and Organizaton 10 (3): 35–50.
[Google Scholar]
Rhodes, M. L., J. Murphy, J. Muir, and J. A. Murray. 2011.
Public Management and Complexity Theory:
Richer Decision-Making in Public Services. New York:
Routledge. [Google Scholar]
Rittel, H., and M. Webber. 1973. “Dilemmas in a General
Theory of Planning.” Policy Sciences 4: 155–
169. doi:10.1007/BF01405730. [Crossref], [Web of Science ®],
[Google Scholar]
Room, D. G. 2015. “Complexity, Power and Policy.” In Elgar
Handbook on Complexity and Public Policy,
edited by R. Geyer and P. Cairney, 19–31. Cheltenham: Edward
Elgar. [Google Scholar]
Room, G. 2011. Complexity, Institutions and Public Policy.
Cheltenham: Edward Elgar. [Crossref], [Google
Scholar]
Sanderson, I. 2009. “Intelligent Policy Making for a Complex
World: Pragmatism, Evidence and
Learning.” Political Studies 57: 699–719. doi:10.1111/j.1467-
9248.2009.00791.x. [Crossref], [Web of

Siggelkow, N., and D. Levinthal. 2003. “Temporarily Divide to
Conquer: Centralized, Decentralized, and
Reintegrated Organizational Approaches to Exploration and
Adaptation.” Organizational Science 14 (6):
650–669. doi:10.1287/orsc.14.6.650.24840. [Crossref], [Web of
Stacey, R. D. 1993. Strategic Management and Organisational
Dynamics: The Challenge of Complexity.
London: Prentice Hall. [Google Scholar]
Teisman, G., A. van Buuren, and L. Gerrits, eds. 2009.
Managing Complex Governance Systems:
Dynamics, Self-Organization, and Coevolution in Public
Investments. New York: Routledge. [Crossref],
[Google Scholar]
Tenbensel, T. 2013. “Complexity in Health and Health Care
Systems.” Social Science and Medicine 91:
181–184. doi:10.1016/j.socscimed.2013.06.017. [Crossref],
[Web of Science ®], [Google Scholar]
Waldrop, M. M. 1992. The Emerging Science at the Edge of
Order and Chaos. New York: Simon and
Schuster. [Google Scholar]
Wright, S. 1932. “The Roles of Mutation, Inbreeding,

Crossbreeding and Selection in Evolution”.
Proceedings of the Sixth International Congress of Genetics (1):
356-366. [Google Scholar]
Week7-References
1. Crash Course (Producer). (2016). Social policy: Crash course
government and politics #49 [Video file]. (Google)
2. Eppel, E. A., & Rhodes, M. L. (2018). Complexity theory and
public management: A ‘becoming’ field. Public Management
Review, 20(7), 949-959
3. European Geosciences Union. (2018). The policy cycle [Web
resource]. https://www.egu.eu/policy/basics/cycle/
7/19/2021 Roadrunner Search Discovery Service
https://eds.a.ebscohost.com/eds/delivery?sid=4ecb0120-79db-
47cc-ae30-f2627ec8909b%40sdc-v-
sessmgr03&vid=2&ReturnUrl=https%3a%2f%2fe… 1/21
Title:
Authors:
Source:
Publisher Copyright:
ISSN:
DOI:
Document Type:

Language:
Accession Number:
Copyright:
Publisher Logo:
Database:
Record: 1
Performance and Management in the Public Sector: Testing a
Model of
Relative Risk Aversion
Nicholson-Crotty, Sean
Nicholson-Crotty, Jill
Fernandez, Sergio
PUBLIC ADMINISTRATION REVIEW; JUL-AUG 2017, 77 4,
p603 12p.
WILEY
00333352
15406210
10.1111/puar.12619
Journal
English
000404376300018
Copyright (c) Clarivate Analytics Web of Science

Social Sciences Citation Index
Performance and Management in the
Public Sector: Testing a Model of
Relative Risk Aversion.
Research has demonstrated that management influences the
performance of public organizations, but almost
no research has explored how the success or failure of a public
organization influences the decisions of those
who manage it. Arguing that many decisions by public managers
are analogous to risky choice, the authors
use a well‐ validated model of relative risk aversion to
understand how such choices are influenced by
managers’ perceptions of organizational performance. They
theorize that managers will be less likely to
encourage innovation or give discretion to employees when they
are just reaching their goals relative to other
performance conditions. Analyses of responses to the 2011 and
2013 Federal Employee Viewpoint Surveys
provide considerable support for these assertions. The findings
have significant implications for our
understanding of the relationship between management and
performance in public organizations.
Related Content: Stanton (PAR July/August 2017)
Practitioner Points
Perceptions of performance influence the likelihood that public
managers will engage in risk‐ taking behavior.
Public managers are less likely to take risks as performance
begins to meet expectations than when

performance falls short of or exceeds expectations.
The willingness of public managers to embrace organizational
change, such as process innovations or
employee empowerment practices, is likely a function of current
organizational performance.
Over the past several decades, a large and growing literature has
demonstrated quite convincingly that
management matters for the performance of public
organizations. It has shown that the decisions public
https://eds.a.ebscohost.com/eds/delivery?sid= 4ecb0120-79db-
47cc-ae30-f2627ec8909b%40sdc-v-
managers make—including whether to collaborate with
stakeholders in the environment, create green rather
than red tape within their organizations, and empower
employees to innovate, among others—have a
significant impact on the success of public organizations and
programs (see, e.g., DeHart‐ Davis [ 18] ;
Fernandez and Moldogaziev [ 22] ; Meier and O'Toole [ 48] ).
In many ways, the relationship between
management and performance undergirds the modern
administrative state and the systems that define it
(Moynihan and Pandey [ 57] ).
Scholars typically treat the relationship between management
and performance as nonreciprocal, assuming
that one influences the other but that the inverse is not true. As
a result, almost no work has explored how the

success or failure of a public organization influences the
decisions of those who manage it. This is a potentially
significant oversight because if performance is in fact
endogenous to management, this could cause us to
substantially overestimate the importance of the latter. In other
words, if public managers in high‐ performing
organizations make systematically different decisions than those
in poor‐ performing organizations, we might
attribute organizational success to those choices when causality
is actually running in the other direction.
Fortunately, very recent research has begun to address this
shortcoming by theorizing about the impact of
performance on managerial behavior in public organizations
(Meier, Favero, and Zhu [ 46] ). While
acknowledging the importance of that work, we will, for a
variety of reasons, suggest an alternative approach to
understanding the relationship between performance and
management. Specifically, we use a relative risk
model from the private management literature in order to
understand how several major types of decisions
made by public managers are influenced by the performance of
their organizations. This is a well‐ validated
approach to decision making that suggests that risk tolerance is
a function of performance relative to the
decision maker's goals or aspirations. Many managerial
behaviors advocated by modern management theories
—including the encouragement of innovativeness and
entrepreneurial behavior, employee empowerment, and
decentralization of decision making authority—impose potential
costs while having uncertain outcomes and
therefore can be conceived of as risky choice. Drawing on
relative risk theory, we develop specific hypotheses
about the performance conditions under which managers are
most likely to make these types of choices.

We test these hypotheses in analyses of responses to the 2011
and 2013 Federal Employee Viewpoint
Surveys. In order to deal with endogeneity and help alleviate
common source bias, we predict an individual
employee's reports of managerial decisions regarding the
encouragement of innovation, empowerment
practices, and delegation of decision‐ making authority in 2013
with average assessments of performance by
managers within that employee's department in 2011. In an
additional analysis, we test directly for the impact
of a manager's performance assessment on his or her own
reported innovativeness. Across six different
dependent variables measuring managerial decisions, and in the
presence of numerous control variables and
fixed effects, the results are remarkably consistent with the
theoretical prediction that key choices made by
public managers are a function of organizational performance
and relative risk tolerance. We conclude with a
discussion of the implications for public management research.
Performance and Decision Making in the Public Management
Literature
As noted earlier, there is really not much to review when it
comes to studies that have used organizational
performance to predict how public managers will make
decisions. Work in the private sector has frequently
explored the ways in which performance feedback influences
firms’ willingness to adopt behaviors either to
address deficiency or to further leverage success (see, e.g.,
Chen [ 14] ; Iyer and Miller [ 32] ; Miller and Chen [
51] ). However, application of this behavioral theory (see Cyert
and March [ 17] ) to public organizations has
been very rare. A notable exception is Salge ([ 62] ), who finds
that negative performance feedback leads
organizations to search for solutions while slack resources
encourage innovativeness in a sample of English

https://eds.a.ebscohost.com/eds/delivery?sid=4ecb01 20-79db-
47cc-ae30-f2627ec8909b%40sdc-v-
public hospitals. In a related article, Nielsen ([ 58] ) finds that
negative performance information induces Danish
school principals to reorder the multiple goals that their
organizations are asked to pursue, emphasizing areas
in which they are doing particularly poorly.
Despite these exceptions, attention to the impact of performance
on managerial decisions in public
organizations has been limited. Recognizing this significant
oversight, Meier, Favero, and Zhu ([ 46] ) build a
theory that imagines performance as a key driver of decision
making by public managers. Specifically, the
authors take a Bayesian approach in which the distance between
current performance and the manager's prior
regarding acceptable performance shapes the choices they make.
They are clear that priors could be formed
in a variety of ways but offer the intuitive expectation that
information about previous performance and the
performance of similar organizations is likely used by managers
to update their beliefs about acceptable levels
of current performance.
The theory suggests that “positive performance gaps,” where
current performance falls below acceptable
levels, should induce managers to be more innovative, seek
opportunities by networking with those in the
organizational environment, and centralize operations. They

hypothesize that the functional form of these
relationships will likely be quadratic, with managerial behaviors
of the kind just described becoming more
aggressive as the performance gap grows. Finally, the authors
suggest that the theory can be extended to
accommodate multiple goals, different levels of hierarchy, and
other realities faced by public organizations.
Meier, Favero, and Zhu's approach is promising, particularly in
its careful consideration of how managers
decide what level of performance they expect from their
organizations, and its attention to the impact of
performance on management is long overdue. It is also,
however, untested empirically. Additionally, the theory
does not deal adequately with conditions when current
performance exceeds the prior regarding an acceptable
level. The authors acknowledge that this type of “negative
performance gap” is likely to have an important
impact on managerial behavior and speculate that it might “be
translated into both additional resources and
autonomy” (Meier, Favero, and Zhu [ 46] , 1232). They do not,
however, generate precise expectations about
the impact of exceeding performance expectations on
managerial decision making. This is particularly
important if, as noted earlier, we are concerned about
overestimating the impact of management in high‐
performing organizations.
As a final challenge, there are some potential inconsistencies in
the expectations offered by the theory. For
example, the authors rely heavily on Miles et al.'s ([ 50] )
concept of prospector versus defender strategies to
identify the decisions managers may make under different
performance conditions. They suggest that poor
performance should motivate managers to encourage innovation
and expand contacts with the environment,

which Miles et al. classify as prospector strategies. Meier,
Favero, and Zhu also suggest that poor performance
will cause managers to centralize authority, but Miles et al.
argue that seeking “strict control of the
organization” is a defender strategy. More importantly, they
suggest that defenders and prospectors are
essentially opposites of one another and that each has a “high
degree of consistency among its solutions” to
organizational problems. Meier, Favero, and Zhu do not offer
sufficient explanation as to why they expect poor
performance to cause such inconsistent reactions among
managers. These issues will likely be worked out in
future iterations of the theory and subsequent empirical tests,
but for now, they suggest that it might be
profitable to explore other lenses for examining the relationship
between performance and management.
An Alternative Approach to Modeling Performance and
Management
We suggest that models of relative risk aversion can be used to
understand decisions that increase uncertainty
for public managers. We will return in the next section to which
tenets of modern management theory we
47cc-ae30-f2627ec8909b%40sdc-v-
believe fit in this category. Before that, however, it is important
to acknowledge the advancements made in the
study of risky choice in the public sector. Although some
research continues to assert that public organizations

have characteristics that can impede risk‐ taking or
entrepreneurial behaviors (see, e.g., Morris and Jones [ 54]
; Townsend [ 66] ), a growing number of studies have
demonstrated that risk‐ taking behavior is present and
predictable in public organizations. In an early contribution,
Bozeman and Kingsley ([ 10] ) test and refute the
assertion that public managers are inherently more risk averse
than their private sector counterparts.
Subsequent research has demonstrated that hierarchy and red
tape are negatively correlated with risk taking
among public managers and employee–supervisor trust is
positively associated (Nyhan [ 59] ; Turaga and
Bozeman 2005).
One key difference between these studies on public management
risk taking and the model we use is that they
rely on an absolute rather than a relative risk‐ aversion
perspective. Absolute and relative risk approaches both
assume that the utility of an outcome is a function of perceived
risk. Each suggests that decision makers are
risk averse when utility for an outcome diminishes as
uncertainty about obtaining it increases and risk seeking
when utility increases with uncertainty. However, the key
feature of absolute risk tolerance is that the functional
form of the relationship between utility and risk is fixed for
each individual. In other words, each person is either
risk averse or risk seeking. Under that assumption, performance
cannot have any impact on propensity for risk
taking.
Work on individual risk taking in other contexts, including
private firms, has long rejected the idea of fixed risk
tolerance in favor of relative risk aversion (see, e.g., Bromiley
and Curley [ 12] ). In these models, the level of
utility for an outcome is still a function of perceived risk, but
the same person can be risk loving and risk averse

depending on his or her performance relative to expectations.
Perhaps the most well known of these
approaches is prospect theory. Kahneman and Tversky ([ 34] )
offer a model of risky choice, which suggests
that the weights assigned to potential gains and losses change
depending on where decision makers feel they
are relative to their desired goal. Specifically, the theory
predicts that decision makers will be risk averse when
in a domain of gain but risk seeking in a domain of loss. In
other words, they will be more risk averse when
they are “winning” rather than “losing” relative to some
preestablished goal. Numerous studies have confirmed
that individual risk preferences vary based on the reference
point between gains and losses (see Tversky and
Kahneman [ 68] for a review).
Research has extended these findings from individuals to the
behavior of organizations and managers within
them (e.g., Bowman [ 7] ; Bromiley [ 11] ; Fiegenbaum and
Thomas [ 24] ). These studies argue that
organizations have variable risk preferences based on their
proximity to a preestablished reference point. More
specifically, they expect that managers will engage in riskier
behavior when performance and resources are
below aspirational levels but become more risk averse as they
begin to realize aspirations. Key insight into the
matter emerged from early efforts to develop a behavioral
theory of the firm. March and Simon ([ 43] ) argue
that the rate of innovation in organizations increases as existing
organizational structures and practices prove
to be inadequate and actual performance lags behind
expectations (see also Levitt and March [ 39] ).
Elaborating on the notion that necessity is the mother of
invention, Cyert and March ([ 17] ) argue that failure
induces search, which often leads to the adoption of innovative
solutions. They call these innovations

“problem‐ oriented innovations,” which are directly linked to a
problem, in contrast to slack innovations, which
are remotely related to a problem and are much easier to justify
when resources are abundant and rules for
allocating resources are relaxed.
March and Simon's ([ 43] ) and Cyert and March's ([ 17] )
research underscores the influence of performance
and aspirations on managerial behavior, particularly exposing
the organization to risk through innovation. In a
47cc-ae30-f2627ec8909b%40sdc-v-
similar vein, research on organizational learning has
emphasized the impact of performance feedback on
managerial decisions to engage in organizational change. Manns
and March ([ 40] ) predict and find that poor
financial performance, or financial adversity, leads university
departments to adopt changes in their curriculum
in order to improve their financial standing. Lant, Milliken, and
Batra ([ 36] ) find that past failures increase the
likelihood that firms will change their strategic orientation,
although external attributions of failure weaken this
relationship. Greve ([ 25] ) finds that the probability of major
organizational change decreases as performance
increases. Importantly, as performance meets and begins to
exceed historical and social aspiration levels, the
probability of major organizational change declines even more
sharply, highlighting the behavioral
consequences of performance relative to aspirations. Ironically,

transformation efforts in response to
performance shortfalls and other pressures from the external
environment expose organizations to “liability of
newness,” increasing the probability of failure in the future
(Amburgey, Kelly, and Barnett [ 2] ). Even when
change does not result in the demise of the organization, it has a
disruptive effect that imposes significant
costs on the organization (Barnett and Carroll [ 3] for a review
of the evidence).
In a synthesis of existing work, March and Shapira ([ 41] , [ 42]
) offer a model of relative risk preference that
suggests that a firm not facing bankruptcy but performing below
its goals will begin to take greater risks in an
attempt to rise to the aspirational or target level. As the firm
approaches or rises just above the level of
performance or resources it hopes to achieve, however,
managers will once again become risk averse,
overweighting the probability that risk taking could drop the
organization back below an acceptable level of
performance. Finally, the model hypothesizes that organizations
will become less risk averse if they find
themselves doing better than they hoped and may even become
risk seeking at very high levels of success.[ 1]
The model's assertion that organizations will become more
tolerant of risk once aspirations are far exceeded
build on the large literature on slack resources and innovation.
That research, generally speaking, asserts that
slack resources permit firms to more safely experiment with
new strategies (Hambrick and Snow [ 27] ; Moses
[ 55] ) and allows slack search, or the pursuit of projects that
may not be immediately justifiable but have high
potential (Levinthal and March [ 38] ).
March and Shapira use an empirically validated approach to
understanding the relationship between

performance and risk in the private sector (see Miller and Chen
[ 51] ). Authors have also validated very similar
aspirational models of the relationship between relative
performance and risk taking in studies of private firms
(see, e.g., Greve [ 25] , [ 26] ).
Applying the Model to Public Management Decisions
A relative risk model can help us understand the relationship
between performance and managerial decisions
under conditions of risk. The foundational assumption of this
approach is, of course, that the decisions of public
managers are related in some way to performance. We believe
that this should be a relatively uncontroversial
proposition, however, both because of the ubiquity of
performance measurement and management regimes in
the public sector (Poister [ 60] ; Sanger [ 63] ) and the
consistent findings that public managers go to great
lengths to avoid poor performance assessments (see, e.g.,
Heinrich [ 30] ; Van der Waldt [ 69] ).
If we accept the premise that public managers are concerned
about performance, then the next step in an
application of the relative risk approach to the public sector is
demonstrating that managerial decisions in that
context are analogous to risky choice. Obviously, not every
decision public managers make fits this criterion,
but we believe that many do. More importantly, we believe that
many of the managerial behaviors championed
by the New Public Management (NPM) and other modern public
management theories are best conceived of
as risky choice.

47cc-ae30-f2627ec8909b%40sdc-v-
At its core, NPM involves an effort to infuse public
management with ideas and practices from the private
sector (Haynes [ 29] ). When contrasted with “traditional”
public administration, these emphasize increased
innovativeness and entrepreneurship on the part of managers,
along with structures to incentivize such
behavior. They encourage increased discretion throughout the
organization and the empowerment of
employees to make changes that improve services. Finally,
modern management prescriptions suggest that
managers need to focus some energy outward, collaborating and
networking with those in the organizational
environment, in order to improve the performance of their
organizations and programs (see, e.g., Meier and
O'Toole [ 47] ; Milward and Provan [ 52] ; Moore [ 53] ).
Risky choice is typically defined as behavior requiring
investment or imposing potential costs when outcomes
are uncertain. Thus, anything that may increase costs and has
uncertainty surrounding benefits is a risk, and
we believe that many modern prescriptions for public managers
fit this definition. Indeed, scholars argue
explicitly that innovating is risk taking because it involves a
novel way of doing something that may or may not
work (Cohen and Eimicke [ 15] ; DiIulio et al. [ 19] ; Feeney
and DeHart‐ Davis [ 21] ).[ 2] It is a well‐ supported
assumption that adopting a new way of doing something
introduces uncertainty regarding outcomes and
therefore is inherently risky (see, e.g., Massa and Testa [ 44] ;
Mellahi and Wilkinson [ 49] ; Thomas and
Mueller [ 65] ; Vargas‐ Hernández [ 70] ; Vargas‐ Hernández,
Noruzi, and Sariolghalam [ 71] ).

Similarly, giving authority to another party creates both adverse
selection and moral hazard problems because
the true preferences of the agent are difficult for the principal to
observe and information asymmetries make it
hard for the latter to know the quality or efficiency of
production by the former (see Bawn [ 4] ; Bendor, Glazer,
and Hammond [ 5] ; Epstein and O'Halloran [ 20] ). These
issues significantly increase uncertainty regarding
outcomes. Finally, choosing to create or engage service delivery
networks or collaborating with those in the
organizational environment can create risk. Collaboration and
interorganizational networks require significant
resources, make management more challenging, introduce their
own agency problems when contractors are
used, and are often unsuccessful at producing desired outcomes
(see Huxham and Vangen [ 31] ; Romzek
and Johnston [ 61] ).[ 3]
Obviously, this is not an exhaustive list of managerial activities
that are analogous to risky choice. It does
demonstrate, however, that several decisions central to modern
public management are likely correlated with
risk tolerance. Because of that correlation, the model of relative
risk outlined earlier should be able to generate
accurate hypotheses about the relationship between
organizational performance and those decisions. The
theory predicts a quadratic relationship like the one presented in
figure [NaN] , where managers are more risk
averse when their organizations are just reaching performance
goals relative to conditions where they are
performing considerably better or worse than that level. In
terms of the decisions mentioned above, this
suggests the following:
Hypothesis 1: Public managers will promote more innovative

activity when they believe their organizations are
failing to meet or exceeding performance goals relative to when
they are just meeting those goals.
Hypothesis 2: Public managers will empower employees with
greater discretion when they believe their
organizations are failing to meet or exceeding performance
goals relative to when they are just meeting those
goals.
The theory also suggests that managers will network and
collaborate less when they are just achieving their
goals relative to other performance conditions, but we do not
have data to explicitly test this. Thus, we do not
offer it as a formal hypothesis.
47cc-ae30-f2627ec8909b%40sdc-v-
The expectations outlined here may seem counterintuitive
because they suggest that managers may act
similarly at different levels of performance, but they are
consistent with some observations of behavior in public
organizations. For example, some research finds that managers
may become more innovative when their
organizations are under stress and when they are performing
very well. The literature on organizational
turnaround in the public sector suggests that organizational
decline can often stimulate managerial innovation
(see, e.g., Boyne [ 8] ; Jas and Skelcher [ 33] ; McKinley,
Latham, and Braun [ 45] ), at least when there is a

recognition of poor performance and sufficient managerial
capacity.[ 4] These assertions are consistent with
Miles et al.'s ([ 50] ) expectation that private sector managers
would seek to be more innovative, and to more
effectively exploit the environment, when their organizations
are performing poorly.
Alternatively, many authors suggest that managers whose
organizations are performing better than expected
are the ones that are most likely to innovate and take risks.
Berry ([ 6] ) finds a positive relationship between
organizational slack and strategic innovation. Similarly, Boyne
and Walker ([ 9] ) argue that a prospector
strategy, which they define including “innovation and rapid
organizational responses to new circumstances,” is
most likely to be undertaken by public managers in
organizations with slack resources. Finally, Carpenter ([ 13]
) demonstrates that agencies can undertake behaviors that are
politically risky when they have used high
performance in other activities to build support among clients.
A relative risk model can help us understand the relationship
between performance and managerial decisions
under conditions of risk.
Data, Variables, and Methods
Testing these hypotheses requires data on both managerial
perceptions of performance and the decisions that
managers make regarding the promotion of innovation and
awarding of discretion to employees. We find these
data in responses to the 2011 and 2013 Federal Employee
Viewpoint Surveys (FEVS). The U.S. Office of
Personnel Management administered the 2013 FEVS to 781,047
employees in 81 federal government
agencies, including cabinet‐ level departments and independent
agencies of all sizes. A total of 376,577

employees completed the survey, for a government‐ wide
response rate of approximately 48 percent in 2013.
The 2011 FEVS was administered to a more limited sample and
elicited more than 266,000 responses, for a
response rate of 49.5 percent. The 81 agencies that participated
in the two surveys make up approximately 97
percent of the federal executive branch workforce. The Office
of Personnel Management uses a stratified
sampling approach that generates representative samples for the
federal government as a whole, for echelons
within the federal government (nonsupervisor, supervisor, and
executive), and for each of the individual
departments and agencies participating in the survey. The
surveys were administered electronically to
employees who were notified by e‐ mail; multiple follow‐ up
e‐ mails were sent to increase response rates.
We are interested in examining the degree to which perceived
performance influences managerial decisions,
but these variables are likely endogenous to one another. We
use time ordering in order to deal with the
potential for reciprocal causation. Specifically, we use the
average performance assessment of managers
within a unit in 2011 to predict individual nonmanagers’
responses in 2013 regarding the degree to which their
unit encourages innovation and creativity and empowers them to
make important decisions regarding work
processes.[ 5]
In order to have some confidence that we are matching
employees and managers in a meaningful way, we
need the smallest unit of aggregation possible, which in the
2011 FEVS is one tier below the agency level.
There are 284 such units identified in the data; they include
organizations such as the Employee Benefits
Security Administration within the Department of Labor and the

Agricultural Marketing Service within the
47cc-ae30-f2627ec8909b%40sdc-v-
Department of Agriculture. Every response in the FEVS is
associated with an agency, such as the Department
of Transportation or the Department of Health and Human
Services. In 2013, about 82 percent of respondents
identified a unit that the FEVS designates as one level below
the agency, leaving approximately 310,000 total
usable responses. When we match with those suborganizations
identified by respondents in the 2011 data, we
are left with approximately 229,000 responses. As noted earlier,
we average the responses of supervisors
within each unit and predict responses of individual employees
in each. Taking managers out of the sample
leaves us with 167,392 employees, which is reduced to an
analysis sample of between 111,000 and 115,000
because of missing data. Nonetheless, there are not significant
differences between this sample and the full
sample of nonsupervisors in the 2013 FEVS.
Dependent Variables
Our dependent variables in subsequent analyses measure the
degree to which employees suggest that their
organizations or managers create a culture of innovativeness
and empower employees with adequate
discretion. We take a multiple measures approach, modeling two
distinct variables for the first concept and
three for the second. For the concept of innovativeness culture,

we first model encouragement to innovate,
which is measured using the FEVS indicator “I feel encouraged
to come up with new and better ways of doing
things.” This measure represents the affective state or
experience of feeling on the part of the respondent that
makes him or her more inclined to innovate (Fernandez and
Moldogaziev [ 22] ). The second dependent
variable, rewarding innovation, is measured using the 2013
FEVS indicator “Creativity and innovation are
rewarded.” This measure represents the degree to which the
respondent feels his or her superiors in the
agency reward efforts to generate innovative ideas and/or
implement them. Available response categories
range from 1 for “strongly disagree” to 5 for “strongly agree”
for these measures.
We use three variables to capture the concept of employee
empowerment. The first is personal empowerment,
measured using the FEVS 2013 indicator “Employees have a
feeling of personal empowerment with respect to
work processes.” This variable captures management's
propensity to share power to shape work processes
with subordinates. The second variable in this set, leadership
opportunities, is measured with the FEVS
indicator “My supervisor/team leader provides me with
opportunities to demonstrate my leadership skills.” This
measure represents the extent to which employees exercise
power or the authority to act. The final variable
measuring empowerment is involvement in decisions, which we
capture with the question “How satisfied are
you with your involvement in decisions that affect your work?”
This last indicator indicates the extent to which
employees are allowed to influence decisions that affect them
and their work. Combined, the three indicators
represent elements of employee empowerment as a managerial
approach (Fernandez and Moldogaziev [ 22] ).

Again, the response categories for each of these range from 1
for “strongly disagree” to 5 for “strongly agree.”
Independent Variables
Our key independent variable captures managers’ perceptions of
performance. Specifically, we use the 2011
FEVS indicator “My agency is successful at accomplishing its
mission.” Although there are obviously multiple
goals that public organizations and the people within them
might pursue, we believe that the accomplishment
of mission is likely to be prominent among them. Available
answers again range from 1 for “strongly disagree”
to 5 for “strongly agree.” The reference or aspiration point is
described in the literature as a point that is
“psychologically neutral” between winning and losing (Kameda
and Davis [ 35] ) or “the smallest outcome that
would be deemed satisfactory by the decision maker”
(Schneider [ 64] ). We assume, therefore, that each
respondent is at their aspirational threshold somewhere between
the answers “neither agree or disagree” and
“agree.”
47cc-ae30-f2627ec8909b%40sdc-v-
Previous research using relative performance to predict
behavior has devoted significant attention to the ways
in which managers might set goals or aspirations for
performance. Much of this work suggests that the
aspirational threshold is determined through a comparison of
current performance with previous performance,

the performance of peers, or some combination of historical and
social factors (Cyert and March [ 17] ;
Haveman [ 28] ; Levinthal and March [ 38] ; Meier, Favero, and
Zhu [ 46] ; Salge [ 62] ). The assumption is that
managers can use these data as a decision heuristic to generate
some prediction of future performance, which
is assumed to become the aspiration or goal (see Greve [ 25] ).
Although we cannot observe these factors directly, our measure
of perceived performance, which asks about
the accomplishment of mission, likely captures both of these
historical and social factors. In other words,
managers’ responses are likely determined in part by how their
agency did in the past and how peer agencies
are doing. Additionally, because we directly measure responses
regarding perceived performance, we do not
have to infer managers’ perceptions of performance. We believe
this is an improvement over previous studies
because the theoretical model suggests that these perceptions,
rather than actual performance, influence risk
tolerance.
We average the responses to the question regarding mission
accomplishment across all supervisors within a
unit (one level below the agency level) and use the mean value
to predict individual nonsupervisor responses
within a unit to the questions discussed earlier. As noted in
figure [NaN] , relative risk theory predicts a concave
quadratic function, where managers are less willing to tolerate
the risks associated with innovativeness and
employee discretion when just achieving goals relative to
conditions when their organization performing worse
or better than that aspiration. In order to model this function,
we also include the average managerial
performance rating squared in each model. In order to provide
support for our hypotheses, the linear term

should be negatively signed, while the squared term should be
positive.
Control Variables
The models discussed here also include a variety of variables
that control for alternative explanations for our
dependent variables. The first set of these reflect individual
characteristics that may influence the degree to
which someone perceives themselves as innovative or believes
that their organization is supportive of such
activities. These individual‐ level controls include respondent
gender, age, minority status, pay category, and
tenure in the federal service. Studies suggest that all of these
characteristics may influence perceptions of
innovativeness, entrepreneurial behavior, and autonomy,
although the consistency and direction of the impact
for these measures are mixed.
We also include a measure of job satisfaction as a control
variable, assuming that this is likely to be correlated
with an employee's attitudes about the degree to which they are
encouraged to innovate or their feelings about
empowerment. Specifically, we include responses to the
question “Considering everything, how satisfied are
you with your job?” This measure should correlate positively
with the dependent variables.
Estimator
All models discussed here are weighted least squares
regressions using sample weights provided in the
FEVS. Each model also includes fixed effects at the agency
level to account for unmeasured organizational
characteristics, such as policies related to employee
empowerment, that might influence our dependent
variables. Finally, standard errors in our primary models are
clustered at one level below the agency level to

account for the fact that employees within units may be more
likely to answer similarly to one another. We use
weighted ordinary least squares rather than an ordered probit
estimator in our primary analyses because this
allows for a more intuitive plot of predicted responses for the
purpose of hypothesis testing.[ 6] However, to
47cc-ae30-f2627ec8909b%40sdc-v-
sessmgr03&vid=2&ReturnUrl=https%3a%2f%2f… 10/21
ensure that the findings are not biased by this choice of
estimator, we present ordered probit models for each
dependent variable in the appendix (table [NaN] ).
Findings
Primary Models
The findings from our primary analyses are presented in tables
[NaN] and [NaN] . The first contains the models
of encouragement to innovate (column 1) and rewarding
innovation (column 2), which capture the degree to
which employees feel that managers incentivize innovation.
First, it is important to point out that the models
perform quite well, explaining between 38 percent and 40
percent of the variation in more than 150,000
individual responses.
Relationship between Managers’ Perceptions of Performance
and Employee Reports of Innovative Culture
Encouraged to InnovateRewarded for Innovation
Accomplish mission (2011) −4.769 −6.657

(−2.18) (−2.34)
Accomplish mission (2011) squared 0.630 0.888
(2.28) (2.47)
Female 0.0644 0.0154
(5.20) (1.02)
Minority 0.0143 0.0343
(1.41) (3.39)
Pay category −0.0311 −0.0508
(−2.30) (−3.03)
Tenure −0.0130 −0.0327
(−1.90) (−4.39)
Job satisfaction 0.636 0.585
(101.07) (95.33)
Intercept 10.10 13.27
(2.34) (2.36)
N 111,774 108,773
R .33 .31
1 Notes: Models include sample weights and agency fixed
effects; standard errors are clustered at the
subagency level. T‐ statistics in parentheses.
2 * p < .05;
3 p < .01;
4 p < .001.
Relationship between Managerial Perceptions of Performance

and Employee Reports of
Empowerment/Discretion
Personal EmpowermentLeadership OpportunitiesInvolvement in
Decisions
Accomplish mission (2011) −6.600 (−3.39) −6.932 (−2.47)
−4.426 (−2.42)
Accomplish mission (2011) squared 0.872 (3.56) 0.886 (2.48)
0.570 (2.47)
Female −0.0480 −0.0331 −0.0158
(−4.07) (−2.66) (−1.56)
Minority 0.109 −0.0378 0.0130
2
47cc-ae30-f2627ec8909b%40sdc-v-
(9.61) (−3.00) (1.29)
Pay category −0.0181 −0.0473 −0.0159
(−1.38) (−3.47) (−1.16)
Tenure −0.0312 −0.0418 −0.0308
(−3.48) (−5.71) (−6.26)
Job satisfaction 0.606 0.572 0.658
(95.74) (97.62) (113.10)
Intercept 13.36 15.21 9.528

(3.44) (2.75) (2.62)
N 110,477 112,546 113,235
R .35 .29 .42
6 p < .05;
7 p < .01;
8 p < .001.
Before turning to the key independent variables, we can quickly
note the impact of the controls. The measure
of satisfaction is positively correlated with both dependent
variables. The results also suggest that
nonsupervisors in a higher pay category and those with longer
tenure both feel less encouraged to innovate
and less certain that creativity and innovation will be rewarded.
Identifying as a minority is positively and
significantly related to both dependent variables, although it
fails to reach traditional levels of statistical
significance in the model of rewarding innovation. Female
employees feel that managers are more likely to
reward innovation and creativity, but they are not significantly
more likely to feel that they are encouraged to
innovate.
Female employees feel that managers are more likely to reward
innovation and creativity, but they are not
significantly more likely to feel that they are encouraged to
innovate.
The real variables of interest are the measure of average

managers’ assessments of performance within a unit
and the squared term. Both are highly statistically significant in
both models. The negative coefficients on the
first terms coupled with the positive coefficients on the squared
terms suggest a concave quadratic function in
which values decrease until an inflection point and then
increase after that point. These results are easier to
conceptualize graphically. Figure [NaN] shows the predi cted
quadratic form of the relationship between
performance and the two innovation variables, with the
associated 95 percent confidence intervals.
We turn now to table [NaN] , which presents the models of
personal empowerment, leadership opportunities,
and involvement in decisions, which capture the degree of
empowerment and decision‐ making discretion that
employees believe managers of their organizations provide.
These models also perform well, explaining 35
percent, 29 percent, and 42 percent of the variation in employee
perceptions. As with the models of innovation,
employee satisfaction is strongly and positively correlated with
all three dependent variables. Female
respondents are consistently less likely to report that they feel
empowered, have sufficient leadership
opportunities, or are adequately involved in decisions affecting
their work. The other control variables perform
relatively inconsistently across the three models.
2
47cc-ae30-f2627ec8909b%40sdc-v-

The key variables of interest, including managers’ assessment
of performance and assessment of
performance squared, are again significant and in the expected
direction in all models. As in the case of the
innovativeness culture models, signs on the two terms suggest a
concave quadratic function. Again the
relationship between performance and employee assessments is
easier to present graphically, which we do in
figure [NaN] .
An Additional Analysis
The analyses presented here provide considerable evidence in
support of a relative risk approach to the
relationship between organizational performance and
managerial decision making. This section will provide an
additional analysis focusing on individual managers that is
designed to increase confidence in those results.
As noted earlier, we believe that using aggregate manager
assessments of performance calculated in 2011 to
predict individual employee assessments in 2013 is a good
approach because it establishes time order
between the independent and dependent variables and helps
overcome the common source bias problem.
However, the findings discussed earlier are more convincing if
we can show a correlation between an
individual manager's perceptions of performance and statements
about his or her own behavior. We restrict the
sample for this analysis to the approximately 62,000 FEVS
respondents who identified themselves as
supervisors in 2013. We acknowledge that such a design cannot
deal with issues of endogeneity and common
method bias, and thus we suggest that these results only be
interpreted as supplementary to the findings

discussed earlier.
There is only one item in the FEVS that reflects a response by
managers about their own behavior, and we use
that question as the dependent variable in this analysis.
Specifically, we model responses to the item “I am
constantly looking for ways to do my job better.” This measure
captures the behavioral aspect of
innovativeness. Such behavior may include searching for ideas
generated and implemented elsewhere,
developing new ideas through experimentation, vicarious
learning and other behavior, and refashioning the
ideas of others to achieve a better fit with extant conditions
(Altshuler and Zegans [ 1] ; Fernandez and Wise [
23] ).
We again use responses to the item “My agency is successful at
accomplishing its mission” to create the
independent variable. In this analysis, however, we create
individual indicators for different response
categories. We create a single measure titled below threshold,
from responses of “strongly disagree” and
“disagree,” because there are relatively few responses in the
former category.[ 7] We use “neither agree or
disagree” responses as the threshold where managers do not feel
they are doing overly well or overly poorly.
Finally, we create an indicator titled above threshold using the
“agree” response category. We use “strongly
agree” as the excluded category. Based on the relative risk
approach, we expect that the coefficients on these
indicators will form a U‐ shaped pattern that matches figure
[NaN] , where managers are least likely to innovate
when they are just reaching performance goals, relative to other
conditions.
The model includes controls for the manager's gender, minority

status, pay category, federal tenure, and
satisfaction with his or her job. It also includes sample weights
and fixed effects at the agency level. This
reduces the analyzed sample to 51,970, but is important to
control for unmeasured characteristics that may
influence innovativeness. Standard errors are again clustered at
one level below the agency level.
The results from the analysis of individual managers are
presented in table [NaN] . The model is highly
significant and explains a reasonable amount of the variation in
individual manager response. Female and
minority bureaucrats, along with those who were more satisfied
with their job and in a higher pay category,
47cc-ae30-f2627ec8909b%40sdc-v-
report more personal innovation. After controlling for those
factors, the amount of time in the federal service is
negatively correlated with the dependent variable.
Relationship between Individual Managers’ Perceptions of
Performance and Their Reports of Personal
Innovativeness
Personal Innovation
Below threshold −0.275
(−12.75)
Aspirational threshold −0.344

(−23.58)
Above threshold −0.260
(−34.92)
Female 0.0665
(9.60)
Minority 0.0486
(4.77)
Pay category 0.00738
(0.91)
Tenure −0.0554
(−10.13)
Job satisfaction 0.138
(15.83)
Intercept 4.173
(118.24)
N 51,970
10 * p < .05;
11 p < .01;
12 p < .001.
The key independent variables are the indicators of managerial

performance assessment. The impacts of
those dummy variables relative to our theoretical expectations
are most easily assessed visually. Figure [NaN]
graphs those coefficients, with associated 95 percent confidence
intervals. Beginning at the right side of the
figure, the plot suggests that managers who just agree, rather
than strongly agree, that their organization is
meeting its goals report significantly less personal innovation.
Moving left across the x‐ axis, reports of personal
innovativeness drop significantly for those managers responding
“neither agree nor disagree” to the statement
about the agency accomplishing its mission. This represents the
lowest point in reported innovativeness. When
we move to managers who disagree or strongly disagree with
the statement about agency performance, the
coefficient becomes significantly less negative, indicating that
managers are more likely to report personal
innovativeness at that performance condition.
Discussion
The results presented here strongly support our expectations
that managers in public organizations become
more risk averse when they are just accomplishing their goals
relative to higher and lower performance
47cc-ae30-f2627ec8909b%40sdc-v-
conditions. The findings presented in table and figure [NaN]
confirm that managers are less likely to foster
innovation and entrepreneurial behavior in their organizations

when they are just meeting performance
aspirations. Relative risk approaches suggest that this is what
we should expect because managers are more
likely to overweight losses, and the probability of backsliding,
when they are just realizing acceptable levels of
performance. These results suggest that a relative risk approach
might offer some theoretical clarity to
previous, somewhat counterintuitive observations that public
managers are more likely to seek out innovative
solutions both when their organizations are in decline (Jas and
Skelcher [ 33] ) and when they are flush with
high performance (Boyne and Walker [ 9] ).
The findings presented in table and figure [NaN] support our
expectations that managers will be more risk
averse, and thus less likely to cede discretion to employees,
when they are just meeting performance goals.
They also appear less likely to give leadership opportunities to
employees or to create an environment in which
subordinates feel they have an adequate voice when their
agencies are just meeting expectations relative to
performing at either a higher or a lower level. Again, this is
what we would expect given the predictions of
relative risk theory.
Finally, the findings from the analysis of individual ‐ level
managers, presented in table and figure [NaN] ,
strongly support our theoretical expectations regarding
managerial behavior at the aspirational threshold, when
they feel their organizations are just reaching or just about to
reach their goals. The correlation between
individual managers’ assessments of performance and their
reported innovativeness form a U‐ shaped pattern,
similar to the concave quadratic function revealed in analyses
using average manager attitudes about
performance to predict employee reports about managerial

behavior. Moreover, the inflection points, where
relative risk theory suggests that managers are most risk averse
and least willing to make risky decisions, are
quite similar in both analyses. This provides confirmation of the
results reported here in a different sample and
at a different unit of analysis.
Conclusion
The relationship between performance and managerial decision
making in public organizations has gone
essentially unexplored until very recently. In order to address
this gap, we propose that relative risk aversion
theories can help us understand when managers make decisions
that impose potential costs but have
uncertain payoffs—in other words, when they make decisions
that are analogous to risky choice. Empirical
analyses of responses from the 2011 and 2013 Federal
Employee Viewpoint Surveys illustrate the value of
such theories for understanding and predicting when public
managers will take risks.
These results have significant implications for our
understanding of public management. A large and growing
literature suggests that differences in performance across public
organizations can be attributed to the actions
of public managers. Our findings suggest the degree to which
“management matters” could be misestimated in
this work. If public managers do more to encourage innovati on
or empower employees when they are in
organizations that are already performing highly, then
researchers could find artificially large effects for these
behaviors when they use them to predict performance in a
sample of such organizations. Our results suggest
that managers in low‐ performing organizations will also be
more willing to delegate discretion to employees
and to encourage innovation. In a cross‐ sectional study, this

could lead to the conclusion that these
management activities reduce performance, when in fact the low
levels of performance are causing the
observed management behavior.
Our results also imply that the effectiveness of recent public
sector reforms designed to incentivize innovation
and entrepreneurial behavior, such as performance pay and
employee empowerment, likely depend in part on
47cc-ae30-f2627ec8909b%40sdc-v-
the existing level of organizational performance. These reforms
are typically designed to increase the benefits
or decrease the costs of innovation, creativity, and risk taking.
Our results suggest that such reforms may be
most effective in organizations that are far exceeding or falling
far short of their performance goals because
those are the organizational contexts in which public managers
will be most willing to take risks and, therefore,
most likely to embrace such reforms.
The effectiveness of recent public sector reforms designed to
incentivize innovation and entrepreneurial
behavior, such as performance pay and employee empowerment,
likely depend in part on the existing level of
organizational performance.
Obviously, these conclusions are tentative and need to be
confirmed in subsequent research. At the very least,

however, our results suggest that studies of public
entrepreneurship and the organizational characteristics that
contribute to it must take account of the relationship between
goal accomplishment and risk tolerance. More
generally, they suggest that performance matters for
management.
Appendix Table A1 Results from Models Estimated Using
Ordered Probit Regression
EncourageRewardedEmpowermentLeadershipInvolvement
Accomplish mission (2011) −5.440 −7.528 −8.069 −7.980
−6.323
(2.294) (3.147) (2.235) (3.140) (−1.95)
Accomplish mission (2011) squared 0.718 1.003 1.066 1.021
0.820
(0.290) (0.399) (0.281) (0.399) (2.01)
Female 0.0695 0.0199 −0.0547 −0.0273 −0.0375
(0.0136) (0.0169) (0.0138) (0.0142) (−2.11)
Minority 0.0154 0.0365 0.126 −0.0340 0.0148
(0.0107) (0.0109) (0.0122) (0.0143) (1.30)
Pay category −0.0327 −0.0564 −0.0214 −0.0500 −0.0133
(0.0147) (0.0187) (0.0152) (0.0151) (−0.72)
Tenure −0.0181 −0.0384 −0.0391 −0.0518 −0.0409
(0.00785) (0.00844) (0.0105) (0.00868) (−5.82)
Job satisfaction 0.690 0.662 0.714 0.618 0.827
(0.00913) (0.00727) (0.00609) (0.00879) (81.99)
N 111,774 108,773 110,477 112,546 113,235
Footnotes

1 The model also suggests that managers will become risk
averse when facing bankruptcy, but because public
organizations do not face organizational death in the same way
as private firms, we focus here on the other
expectations offered by the model.
2 See also studies that separate concepts of entrepreneurship,
innovation, and risk taking (e.g., Covin and
Slevin 16; Morris and Jones 54).
3 However, it is important to note literature that suggests that
networks and collaboration may help
organizations manage uncertainty, and by extension risk, under
certain conditions (see, e.g., Moynihan 56).
4 See Levine (37) for the argument that managers may also
engage in retrenchment during periods of decline.
5 Creating our independent and dependent variables using
responses from different groups within an
organization in different years also helps overcome the common
source bias problem. It is important to
** ** *** ** *
** ** *** ** *
*** *** * *
*** *** **
** *** ***
** *** *** *** ***
*** *** *** *** ***

47cc-ae30-f2627ec8909b%40sdc-v-
acknowledge that it may still exist, however, because of shared
experiences by managers and employees
within the same unit.
6 Ordinary least squares allows us to show one plot of the
predicted category into which each respondent falls
on each dependent variable across the range of the performance
measure rather than the separately plotting
the conditional probability of being in each category.
7 Results do not change substantially if we create different
indicators for each response category. “Strongly
disagree” is not statistically distinct from “disagree” in the
coefficient plot.
8 Related Content: Stanton (PAR July/August 2017)
References
Altshuler, Alan A., and Marc D. Zegans. 1997. Innovation and
Public Management: Notes from the State
House and City Hall. In Innovation in American Government:
Challenges, Opportunities, and Dilemmas, edited
by Alan A. Altshuler, and Marc D. Zegens, 68 – 80.
Washington, DC : Brookings Institution.
Amburgey, Terry L., Dawn Kelly, and William P. Barnett. 1993.
Resetting the Clock: The Dynamics of

Organizational Change and Failure. Administrative Science
Quarterly 38 ( 1 ): 51 – 73.
Barnett, William, and Glenn Carroll. 1995. Modeling Internal
Organizational Change. Annual Review of
Sociology 21 : 217 – 36.
Bawn, Kathleen. 1995. Political Control versus Expertise:
Congressional Choices about Administrative
Procedures. American Political Science Review 89 ( 1 ): 62 –
73.
Bendor, Jonathan, Amihai Glazer, and Thomas Hammond. 2001.
Theories of Delegation in Political Science.
Annual Review of Political Science 4 : 235 – 69.
Berry, Anthony J. 1994. Spanning Traditional Boundaries:
Organization and Control of Embedded Operations.
Leadership and Organization Development Journal 15 ( 7 ): 4 –
10.
Bowman, Edward. 1982. Risk Seeking by Troubled Firms. Sloan
Management Review 23 ( 4 ): 33 – 42.
Boyne, George A. 2004. A “3Rs” Strategy for Public Service
Turnaround: Retrenchment, Repositioning and
Reorganization. Public Money and Management 24 ( 2 ): 97 –
103.
9 Boyne, George A., and Richard M. Walker. 2004. Strategy
Content and Public Service Organizations.
Journal of Public Administration Research and Theory 14 ( 2 ):
231 – 52.
10 Bozeman, Barry, and Gordon Kingsley. 1998. Risk Culture
in Public and Private Organizations. Public

Administration Review 58 ( 2 ): 109 – 18.
11 Bromiley, Philip. 1991. Testing a Causal Model of Corporate
Risk Taking and Performance. Academy of
Management Journal 34 ( 1 ): 37 – 59.
12 Bromiley, Philip, and Shawn Curley. 1992. Individual
Differences in Risk Taking. In Risk‐ Taking Behavior,
edited by J. Frank Yates, 87 – 132. Chichester, UK : Wiley.
47cc-ae30-f2627ec8909b%40sdc-v-
13 Carpenter, Daniel P. 2001. The Forging of Bureaucratic
Autonomy: Reputation, Networks, and Policy
Innovation in Executive Agencies. Princeton, NJ : Princeton
University Press.
14 Chen, Wei‐ Ru. 2008. Determinants of Firms’ Backward‐
and Forward‐ Looking R&D Search Behavior.
Organization Science 19 ( 4 ): 609 – 22.
15 Cohen, Steven, and William Eimicke. 1998. The Use of
Citizen Surveys in Measuring Agency Performance:
The Case of the New York City Department of Parks and
Recreation. Paper presented at the Annual Meeting
of the American Society for Public Administration, Seattle,
WA, May 9–13.
16 Covin, Jeffrey G., and Dennis P. Slevin. 1991. A Conceptual
Model of Entrepreneurship as Firm Behavior.

Complexity theory and public management a ‘becoming’ field

Complexity theory and public management a ‘becoming’ field

Recommended

Recommended

More Related Content

Similar to Complexity theory and public management a ‘becoming’ field

Similar to Complexity theory and public management a ‘becoming’ field (20)

More from LynellBull52

More from LynellBull52 (20)

Recently uploaded

Recently uploaded (20)

Complexity theory and public management a ‘becoming’ field