Theories of Governance and New Public Management:
Links to Understanding Welfare Policy Implementation
Jo Ann G. Ewalt
Department of Government
Eastern Kentucky University
Prepared for presentation at the Annual conference of the
American Society for Public Administration
March 12, 2001
Note: This paper sketches out a research design, but it does not include analysis of data.
Unfortunately, data were not available in time for the ASPA conference.
If Max Weber and Woodrow Wilson were to suddenly appear on the landscape of
modern public administration, normative theories in hand, it is likely they would be unable to
recognize the field. The comprehensive, functionally uniform, hierarchical organizations
governed by strong leaders who are democratically responsible and staffed by neutrally
competent civil servants who deliver services to citizens (Ostrom, 1973) – to the extent they ever
existed – are long gone. They have been replaced by an ‘organizational society’ in which many
important services are provided through multiorganizational programs. These programs are
essentially “interconnected clusters of firms, governments, and associations which come together
within the framework of these programs” (Hjern and Porter, 1981, pp. 212-213).
These implementation structures operate within a notion of governance about which a
surprising level of consensus has been reached. There is a pervasive, shared, global perception of
governance as a topic far broader than ‘government’; the governance approach is seen as a “new
process of governing, or a changed condition of ordered rule; or the new method by which
society is governed” (Stoker, 1998, p. 17). Similarly, in the scholarship that has followed the
‘Reinventing Government’ themes of public effectiveness, much has been written of New Public
Management practices by which governance theory is put into action (Mathiasen, 1996; Lynn,
1996, 1998; Terry, 1998; Kelly, 1998; Peters and Pierre, 1998).
In this complex, devolved mode of service delivery, the unit of analysis for some students
of policy implementation is the network of nonprofit organizations, private firms and
governments. As Milward and Provan note, in policy arenas such as health, mental health, and
welfare, "...joint production and having several degrees of separation between the source and the
user of government funds...combine to ensure that hierarchies and markets will not work and that
networks are the only alternative for collective action" (2000, p. 243).
The purpose of this paper is to attempt to set forth a theoretical framework for the study
of welfare policy implementation by synthesizing the related and theoretically consistent
concepts of governance, New Public Management, and networks. I then discuss how this
framework can be applied to welfare policy implementation. I follow the lead of scholars who
have attempted to offer coherence and synthesis to a research field that historically has been
dominated by top-down and bottom-up perspectives.
The need to inform implementation scholarship is great. As O’Toole concludes in his
review of the literature on multiorganization policy implementation,
The field is complex, without much cumulation or convergence. Few well-
developed recommendations have been put forward by researchers, and a number
of proposals are contradictory….Two reasons for the lack of development are
analyzed: normative disagreements and the state of the field’s empirical theory.
Yet there remains numerous possibilities for increasing the quality of the latter.
Efforts in this direction are a necessary condition of further practical advance.
(1986, p. 181).
Welfare Policy Implementation
It would be difficult to find a policy arena in which the sheer number of local, regional,
and state organizations involved in the implementation of programs and policies was greater, and
in which the organizational interdependencies and dynamics were more varied. The complexity
of welfare policy structures, as well as the variation in these structures, is due in large part to the
flexibility given to states by the Personal Responsibility and Work Opportunity Reconciliation
Act of 1996 (PRWORA).
Before PRWORA, welfare policy generally referred to the Aid to Families with
Dependent Children (AFDC) program, which had been in place for sixty years. Under the
reformed welfare system, AFDC has been replaced with the Temporary Assistance for Needy
Families (TANF) program. The legislation ends federal entitlements, funds welfare through
capped block grants to the states, and sets a five-year maximum lifetime limit on receipt of
welfare benefits. Perhaps as important as these policy changes, PRWORA gives the states
enormous latitude in setting eligibility, benefit, and sanction policy, and it changes the focus of
welfare from a human capital model emphasizing prolonged education and vocational training to
a work-first approach that stresses the importance of work and self-sufficiency (Hayward, 1998).
While there are many models of welfare implementation, welfare is essentially
administered at either the state level (centralized administration) or at the local or county level
(decentralized administration). Regardless of the administration status, PRWORA has brought
together public, nonprofit, and private agencies that had been largely autonomous in their
operations and more narrowly focused in their organizational mission. For example, in theory,
Private Industry Councils (PICs) and Departments of Employment Services (DES) were to have
been working closely with welfare agencies implementing the Job Opportunity and Basic Skills
Program (JOBS) in the late 1980s and the first half of the1990s. The reality was that for most
states, there was very little involvement from employment-related public agencies in
implementing welfare programs (Ewalt, 1998).
In the current environment, local welfare implementation involves a host of public and
nonprofit organizations. Regardless of whether the organization is centralized or decentralized, a
variety of agencies are contracted with to provide basic services such as eligibility assessment,
needs assessments, job training, employment training, education, transportation, child care, job
retention, and rehabilitation. In addition, although most welfare recipients are children and their
mothers, non-custodial parents (usually fathers) are also an important if ancillary program target.
Important goals of welfare programs relate directly to the legislation, and to state implementing
regulations. All states are subject to statutory requirements about the percent of welfare clients
who must be engaged in "countable" work activities, among other rules, and these requirements
drive many of the organizational relationships states and localities establish. The bottom line for
state and local welfare programs is that they seek to accomplish the following universal goals:
•= Divert potential clients from TANF when other assistance may be more appropriate;
•= Move TANF clients into countable work activities as soon as possible and at least as
soon as prescribed by state and/or federal law;
•= Meet federally mandated work participation rates;
•= Remove barriers such as lack of child care, transportation, appropriate clothing, and
so on, so TANF clients can remain at work;
•= Assist TANF clients in devising a self-sufficiency plan to move off public assistance;
•= Apply client sanctions when noncompliance with program rules reaches a critical
This brief review of welfare policy is intended to introduce the requirements of welfare
reform and its implications for organizational structure. To describe the specific nature of
welfare implementation arrangements, we turn to the literature on networks.
Networks in Action
In meeting program goals, welfare agencies must establish both informal and formal
relationships with numerous organizations (Provan and Milward, 1995; O'Toole, 1997; Agranoff
and McGuire, 1998). In many of these multiorganizational networks of linked agencies and
other units, the linkages are not particularly well established. Rather, they are in a state of
continual reformulation because of shifts in providers, new program emphasis or problems, or
other internal or external shocks to the environment (O'Toole and Meier, 2000).
The practice of contracting out government services to networks of nonprofit (and some
private) organizations has been referred to as the "hollow state" (Milward and Provan, 2000).
Hierarchical bureaucracies are generally considered more predictable and stable because
networks must coordinate, negotiate, monitor, enforce and hold accountable a variety of
organizations with varying funding streams and levels of authority and responsibility. As these
authors have noted elsewhere, human service systems that are stable are better performers
(Provan and Milward, 1995). In addition, they point out that networks may threaten the
legitimacy of governance because of the distance between government and policy action (2000,
p. 242). Another critical issue is whether network arrangements are mandated or emerge as part
of the evolution of program implementation (Alexander, 1995; Alter and Hage, 1993).
In sum, Milward and Provan note that there are at least four perspectives on evaluating
•= Clients and advocacy groups favor flexible network structures at the level of
the service provider.
•= Agency managers and network administrators favor stability, which
presumably leads to agency and network growth and increases in resources.
•= Local officials and community leaders look for structures that promote
efficiency, cost reductions, and the containment or reduced visibility of social
•= Funding sources and regulators favor network structures that permit control
and monitoring and thus reduce the likelihood of their being blamed for poor
outcomes. (2000, p. 255)
The premise of this paper is that the related concepts of governance, New Public
Management, and network theory can bring some clarity to the search for theories of policy
implementation. It also offers hope for one of the most prevalent criticisms of implementation
research: the lack of parsimony. In the section that follows, I review the evolution of
implementation research from a top-down/bottom-up structure to efforts to achieve synthesis. A
governance model of implementation is then explored as a theoretical alternative that offers the
potential to synthesize disparate models and bring networks into the governance paradigm.
Finally, I suggest ways that the governance model can be applied to increase our understanding
of implementation structures in welfare policy.
Moving Beyond Top-Down Bottom-Up Approaches
Policy implementation is what happens after policy makers have decided to do something
new, do something different, or stop doing something, and before the impact of this action. In the
study of implementation, at least until recently, two conflicting analytical perspectives
dominated: top-down and bottom-up views of how one should study the issue. Bottom-up
theorists (Hjern and Porter, 1981; Hjern, 1982; Hull and Hjern, 1987; Lipsky, 1978) study
implementation at the street level, concentrating on service deliverers and policy targets. For top-
down theorists (Van Meter and Van Horn 1975, Sabatier and Mazmanian, 1980, Mazmanian and
Sabatier, 1981, 1983, 1989; Sabatier, 1986) the level of analysis is the relationship between the
authorizing statute or order, the nature of the problem, and the central actors in the
Both perspectives have their critics. As summarized by Matland (1995), the top-down
approach suffers from a lack of specificity about important sub-national contexts, while bottom-
up research tends to ignore centralized policy control and is often so narrowly conceived – in the
form of case studies or small-n studies – that variation is described but generalization is rarely
O’Toole suggests that “virtually all analysts have moved past the rather sterile top-
down/bottom-up dispute, and some helpful proposals for synthetic or contingent perspectives
have been offered” (2000, p. 267). However, as he notes, consensus has not been reached and the
sheer number of explanatory factors has not been much reduced.1 Efforts to combine the two
approaches illustrate that both have merit. Two recent reviews of the synthesis literature offer
distinct but related convergence approaches.
In trying to connect top to bottom, researchers have generally either proposed new
models, or have developed theories of when one approach is more appropriate (Matland, 1995, p.
150). Examples of the former include forward and backward mapping (Elmore, 1982,
1985), advocacy coalitions (Sabatier and Pelkey, 1987; Sabatier, 1991), and Goggin et al’s
communications model (1990). Taking the latter approach, Berman (1980) suggests the top-
down/bottom-up dispute can be settled by viewing the specific policy context: scope of change,
validity of technology, institutional environment, goal conflict and environmental stability.2
Matland offers an ambiguity/conflict implementation model which views the level of conflict
over policy goals and goal ambiguity as dichotomous, producing four distinct implementation
processes with varying reliance on top-down or bottom-up forces.
In O’Toole’s (2000) review of advances in implementation research, he finds progress in
attempts to achieve synthesis. In particular, he points to a number of studies that – while outside
the formal domain of implementation literature – have much to contribute to the field. Among
these are Institutional Analysis and Development, Governance, and Networks and Network
Management. It is his assessment of the governance-implementation link that concerns us here.
Governance theory highlights the multivariate character of policy, considers the design and
operation of policy structures and actions, and focuses on the “multi-layered structural context of
rule-governed understandings, along with the role of multiple social actors in arrays of
1 O’Toole’s 1986 review of the literature uncovers some 300 key variables that affect policy implementation. Meier
has quipped, “I propose…any policy implementation scholar who adds a new variable or a new interaction should
be required to eliminate two existing variables” (1999, p. 5-6).
negotiation, implementation, and service delivery. Addressing governance requires attending to
social partners and ideas about how to concert action among them.” (2000, p. 276). Clearly,
governance theory embraces many implementation themes.
Governance and NPM: Integrating Implementation Questions
The (mostly European) literature on governance and the increasingly international
scholarship on New Public Management3 (NPM) describe two models of public service that
reflect a ‘reinvented’ form of government which is better managed, and which takes its
objectives not from democratic theory but from market economics (Stoker, 1998). While some
use the terms interchangeably (for example, Hood, 1991), most of the research makes
distinctions between the two. Essentially, governance is a political theory while NPM is an
organizational theory (Peters and Pierre, 1998). As Stoker describes it,
[G]overnance refers to the development of governing styles in which boundaries
between and within public and private sectors have become blurred. The essence
of governance is its focus on mechanisms that do not rest on recourse to the
authority and sanctions of government….Governance for (some) is about the
potential for contracting, franchising and new forms of regulation. In short, it is
about what (some) refer to as the new public management. However, governance
…is more than a new set of managerial tools. It is also about more than achieving
greater efficiency in the production of public services (1998, p. 17-18).
Peters and Pierre agree, saying that governance is about process, while NPM is about outcomes
(1998, p. 232).
Governance is ultimately concerned with creating the conditions for ordered rule and
collective action (Stoker, 1998; Peters and Pierre, 1998; Milward and Provan, 2000). As Stoker
2 As should be expected, all efforts to synthesize the literature draw from theories found in the separate traditions.
Berman owes debts to Van Meter and Van Horn (1975, 1976) and Goggin, et al (1990), among others.
3 See Kaboolian (1998) for a description of reform movements in the public sector that collectively comprise “New
Public Management” (NPM). I adopt her definition of NPM as a series of innovations that – considered collectively
– embody public choice approaches, transaction-cost relationships, and preferences for efficiency over equity.
notes, the outputs of governance are not different from those of government; it is instead a matter
of a difference in processes (1998, p. 17). Governance refers to the development of governing
styles in which boundaries between and within public and private sectors have become blurred.
The essence of governance, and its most troublesome aspect, according to its critics, is a focus on
mechanisms that do not rest on recourse to the authority and sanctions of government (Bekke, et
al, 1995; Peters and Pierre, 1998; Stoker, 1998; Rhodes, 1996, 1997).
Stoker (1998, p. 18) draws five propositions to frame our understanding of the critical
questions that governance theory should help us answer. He acknowledges that each proposition
implies a dilemma or critical issue.
Governance refers to institutions and actors from within and beyond government. (But
there is a divorce between the complex reality of decision-making associated with governance
and the normative codes used to explain and justify government).
The question, as it relates to policy implementation, is one of legitimacy. The extent to
which those with decision-making power are seen to be legitimate (in the normative sense) will
directly impact their ability to mobilize resources and promote cooperation and build and sustain
partnerships. Thus, the normative dilemma has pragmatic overtones. Beetham suggests that for
power to be legitimate it must conform to established rules; these rules must be justified by
adherence to shared beliefs; and the power must be exercised with the express consent of
subordinates (1991, p. 19).
Governance identifies the blurring of boundaries and responsibilities for tackling social
and economic issues. This shift in responsibility goes beyond the public-private dimension to
include notions of communitarianism and social capital. (However, blurring of responsibilities
can lead to blame avoidance or scapegoating).