In today’s fragmented and narrowly specialised public sector, different functional areas quickly develop their own distinctive language and practice. Thus, as public management theory and practice has settled into the current era of ‘managing-for-outcomes’ (MfO) and ‘results-based management’, the field of policy analysis and advice has been striving to become better informed by evidence, sensitive to community concerns and interests, forward-thinking, and outward-looking. Over the same period, program evaluators have been developing a wide array of methodologies suited to different kinds of evaluation tasks, as well as experimenting with and adapting techniques drawn from the fields of systems thinking and dynamics and dabbling in complexity theory, while those in charge of public sector service delivery have been seeking to lift their game by way of project management applications and other methods.

Optimistic outsiders gazing down upon this scene may see a rich abundance of public sector energy and creativity which is bound to generate improvement, but their pessimistic counterparts will be quick to note that continual proliferation of new techniques and methods within narrowly defined areas of public sector activity is more likely to build a bigger and better Tower of Babel than it is to improve actual government performance. With policy, management, evaluation, and service delivery experts all pursuing their own discipline-specific, smart-practice innovations, it is no wonder that practitioners in these fields are finding it hard to talk to each other and work together effectively. Intra- and inter-agency cooperative networks may be all the rage in public sector studies, but scant attention has been paid to the ways in which core, discipline-based practices actually impede such collaboration. Exhorting policy advisers, public managers, evaluators, and service deliverers to spend more time building networks, sharing information, and looking for collaborative opportunities will not yield fruit so long as the professionals in these fields are speaking different languages with respect to good practice.

Fortunately, beneath the buzzword-babble of outcomes, systems, complexity, and evidence, one can detect a common core of plain-language meanings from which shared understanding and action may emerge. What unites these various devices and techniques are two simple and familiar, but also revolutionary, ideas about means and ends:

New Zealand’s embrace of managing-for-outcomes (MfO) was justified partly on the grounds of the first idea. Although the big state sector reforms of the 1980s clearly had accomplished much in the way of efficient delivery of programs and services and transparent governance, some critics concluded that the New Zealand model’s strong focus on monitoring and reporting outputs was causing the public service to lose sight of policy objectives (Schick 2001, Steering Group 2002). With ministers officially accountable for outcomes and agency chief executives primarily accountable for outputs (and for reporting on the links between the two), there were concerns that these two essential actors, and their respective institutions, were slowly drifting apart into their own preoccupations.

The common language and shared mission of MfO was therefore needed to reconnect outputs with outcomes and to reconnect government departments with both ministers and the people whom they serve. Central agencies put considerable effort into helping agencies articulate their ‘vital few’ high-level outcomes in ways that would clarify links as well as gaps between the agency’s outputs and intended outcomes, with the ultimate goal of helping agencies rethink and then reconfigure their output mix for greater effectiveness. It is too early to say whether or not MfO has had the desired effect yet. There is always the danger that a focus on outcomes will perpetuate or aggravate the artificial distinction between outputs (means) and outcomes (ends), rather than reconnecting them. Whatever the result, it is worth remembering that restoring the integrity of the output-outcome chain was at least part of the original intention. It is also worth noting that many of MfO’s effects on actual departmental practice may be both powerful and unobservable; the fact that evaluations cannot pick up these effects does not mean they are not present.

The idea that means and ends always travel together also has roots in certain perspectives on policy analysis. Whereas the conventional view of public policy assumes that decision makers set policy objectives and analysts/advisers design various configurations of resources (in the form of programs and policies) for reaching those objectives, Aaron Wildavsky (1987) defined a public policy as a yoked phenomenon – a program together with its goals or a particular set of resources tied to particular objectives. We might call it a sort of means-and-ends package deal. In order to ensure the coherence of this package, it is also necessary to include some kind of theory, rationale, or causal logic that explains why those particular means are expected to generate those particular ends. For example, the policy of providing welfare benefits is linked to the outcome of improved family well being and social solidarity according to a typical social democratic view of causation, whereas a more conservative worldview links the very same policy to an entirely different set of outcomes that includes inter-generational dependency, declining well being, and resentment between the working poor and the beneficiary poor. Thus, means and ends are not enough to specify a policy; some sort of linking theory or logic is also needed.

Viewed in this way, policy choice becomes not only a choice among alternative programs, services, regulatory schemes, or activities, but also a choice among alternative objectives, the achievement of which would constitute each related program’s aim, as well as a choice among causal theories. This redefinition of the policy choice process as a simultaneous and interlocking (rather than sequential or hierarchical) choice of ends, means, and causal theories has significant implications for those who give advice about improving government performance, because if Wildavsky was right, then determining which programs and services are ‘best’ cannot be done by reference to externally derived policy objectives. Instead, advisers need to present policy options in the form of alternative program packages, each designed to achieve a slightly different configuration of objectives via its own combination of multiple instruments and resources, and each supported by its own causal theory. Advisers also need to present decision makers with information that will help them compare the alternative packages in terms of feasibility, costs, risks, uncertainties, distributional impacts, and, where possible, likely overall effectiveness. Decision makers need to know not only what the program is meant to produce and how it is meant to do so, but also whose cooperation and/or compliance is required to make it work and the kinds of circumstances that are most likely to support or undermine it. This approach to policy advice recognises that policy choices are nearly always driven by a combination of technical, broadly political, and more narrowly partisan considerations. It does so by presenting program options as multi-dimensional packages and recognising that different actors and stakeholder groups inside and outside government will focus on different dimensions of the selected package.

Wildavsky’s yoked definition of public policy also has implications for program evaluation. In addition to the familiar problem of attributing specific outcomes to specific actions by government, the desired outcomes themselves are seen as a matter of policy choice and, therefore, subject to interpretation and change over time. Among the various professional groups associated with the public policymaking process, evaluators have been leaders in thinking about how means-ends connections and causal theories should be studied amidst uncertainty and complexity.

Examples of yoked policy choices are not hard to find. In the area of drug policy, choosing between a policy of stricter sentences for drug convictions or a policy of decriminalisation also constitutes a choice between the objective of clearly communicating society’s disapproval of drug-taking behaviour, on one hand, or on the other hand, loosening that message in order to pursue a different objective, that is, reducing the social and financial costs associated with drug law enforcement. In an area such as transport, choices among competing objectives are often much more explicit, as when government must divvy up its limited budget for road works among competing projects with different objectives, such as alleviating big city traffic congestion versus improving market access for rural farmers versus reducing road accidents and fatalities. Virtually all allocative policy choices have this simultaneous ends-means quality about them. This is why very few experts have even attempted to devise technical schemes for solving government’s grand budget allocation decisions – guns vs butter or health vs education vs housing. Examples of competing causal theories are also plentiful. The invasion of Iraq was expected by some to liberate the Iraqi people from oppression and ultimately contribute to Middle East stability and development, while others expected it to fuel civil war in Iraq and destabilise the region.

A particular program or policy’s intended links between means and ends can, of course, be strong or weak, direct or indirect, proximate or distant; the causal theory may be based on evidence, educated guesses, or flights of fancy. Whatever their features, it is these hypothesised links in the cause-and-effect chain that communicate government’s intentions and thereby provide the platform from which policy designers, implementation planners, and operational units can go about their core business of marshalling resources and enabling people inside and outside of government to work together converting resources into desired outcomes.

This brings us to the second core idea signalled earlier. The second unifying theme from recent developments in good practice is that turning outputs into outcomes, or public means into public ends, is fundamentally a process of co-production that requires action by both government and citizens. This principle may seem obvious in cases such as social services, where the effectiveness of social work interventions clearly depends upon the receptiveness and responsiveness of the client. ^[1] But even the least touchy-feely of all public functions – national defence – depends upon people paying their taxes, refraining from gratuitous actions that would undermine the military (such as interfering with military exercises or bombing military bases), and keeping a watchful eye on government’s use of military power. These sorts of activities, in turn, depend upon public support for the norms of parliamentary representation and oversight, a free press, and free speech, to name just a few of the essential prerequisites to healthy democratic governance. Although some of these citizen obligations are backed by laws and the threat of penalties for non-compliance (including voting in Australia and census participation in New Zealand), we assume that the vast majority of people will fulfil their obligations voluntarily. Production of the resulting goods can therefore be attributed in large measure to high levels of cooperation.

Examples of co-production are everywhere. Government determines eligibility for benefits and writes benefit checks, but it is up to individual beneficiaries how they will spend the money to sustain their families. Government builds roads, but citizens decide when and where to drive on them for purposes of work, recreation, socialising, etc. Citizens also choose how fast to drive (with or without regard to posted speed limits) and with what level of care and attention (with or without regard to legal definitions of reckless driving). Government funds universities, but citizens decide whether and what to study, how much effort to expend, and whether to join or resist the brain drain once they graduate. Understanding these co-production processes is central to designing smart policies that will encourage, enable, educate, persuade, bribe, or coerce citizens into making the kinds of choices that contribute most to aggregate public goals.

Although the principle of co-production is not often acknowledged explicitly, it is implicit in current conversations about accountability for outcomes, participatory policy design, and public trust in government. New Zealand central agencies have worked hard over the past several years to reassure chief executives that they are being held accountable for outputs and ‘managing for outcomes’ only (always in inverted commas), and not for actual outcomes themselves. This sort of approach acknowledges the complex web of factors that combine to determine any social or economic outcome, and recognises that government policy contributes only a few sticky threads to that web. It is a small step from this acknowledgment to embracing a co-production model of the output-outcome chain. Those who agitate for more participatory approaches to policy design, through more and better consultation for example, also seem to tacitly understand the role that citizens play (whether they are consulted or not) in the achievement or undermining of policy goals, although the pro-participation advocates often couch their arguments more in terms of democratic principles than co-production realities. Likewise, those who warn about citizens’ declining trust in government clearly understand the central role that citizens play in the work of government even if they do not always acknowledge policy-specific cases of co-production.

Scratching below the surface of current public sector fashions reveals at least two core ideas – the principle of means-ends dependency and inseparability in public policy, and the principle of government-citizen co-production – which may be worthy of further development. A currently popular set of concepts known as program or intervention logic in Australasia (or the results chain in Canada) provides one possible avenue for this development.