The Political Implications of Performance Management and Evidence-Based Policymaking

Introduction

Over the last few decades performance management (PM) has spread to the public sectors of most Organisation for Economic Co-operation and Development (OECD) countries. A host of new managerial mechanisms, such as performance measuring, benchmarking, accrual accounting, and (internal and external) contracts, have been adopted to boost the performance of public service production and delivery. More recently, calls for basing public policymaking on scientific evidence concerning the relation between intervention and effect have proliferated. Evidence-based policymaking (EP) has been particularly influential within public health policies but is rapidly spreading to other welfare policy areas. Both PM and EP promise to deliver more effective and efficient public policies. For this reason, some have tended to see PM and EP as part and parcel of the same reform movement and lumped them together under the heading of new public management (e.g., Head, 2008). Others have distinguished them according to their scientific validity and/or their usefulness for ensuring various desirable goals, such as accountability (e.g., Heinrich, 2007).

This article agrees that PM and EP should be seen as two distinct phenomena. However, to critically assess their political implications, we may want to view PM and EP as particular forms of government in the sense proposed by Michel Foucault in his lectures on governmentality (see below). Such a distinction may not only provide a better understanding of the distinct rationalities and mechanisms by which PM and EP work, it may also provide us with a stronger insight into the political implications of these two ways of governing. By political implications, this article refers to the forms of power and freedom exercised through PM and EP, and to the promises and the dangers they pose in terms of diminishing the space for possible political and managerial actions. Or, in other words, I discuss how PM and EP may in different ways reduce the possibility of unsettling or reversing the power relations supporting the ways in which public administration is conducted in liberal democracies.

By addressing PM and EP as distinct forms of governing characterized by very different constellations of power–freedom relations, the approach taken in this article differs from the bulk of existing critical accounts. The latter tend to evolve around the three themes of (lacking) rationality, accountability, and realism. First, although PM-inspired reforms seek credibility from more or less clearly defined theoretical frameworks, such as public choice, principal-agent theory, and transaction cost theory, the reforms have been criticized for inadequate evidence of their actual effects (Pawson, 2006). Moreover, several studies show that PM reforms may produce several unintended and undesirable effects, such as creaming, parking, and more generally suboptimal gaming behavior (Bevan & Hood, 2006; Hood, 2006; Wilson, Croxson, & Atkinson, 2006). Consequently many PM reforms are based more on theoretical postulates than on documented effects. Although scientific inquiries into the (lacking) rational foundation may serve as a useful basis for criticizing PM, it is impotent with regards to EP. Such a critique will only serve to strengthen EP by propagating demands for rigorous scientific evidence of alleged relationships between intervention and results. Thus, it seems that the critique of PM’s mixed scientific credentials have contributed to the recent demand for EP.

Second, although PM has been the main target of scientific critique, EP has primarily been criticized for disregarding political accountability (Biesta, 2007; Bronwyn, 2003; Heinrich, 2007), local participation (Parsons, 2002) and even devaluing the debate over the moral issues that are part of all policy choices (Hammersley, 2001). This line of critique suggests that EP’s exclusive focus on building up scientific knowledge about the effects of various policy mechanisms has the effect, deliberately or not, of sidestepping political and moral questions important to the functioning of any liberal democracy.

Third, EP has been criticized for not being able to adequately grasp the political reality in which policymaking processes are enmeshed (Head, 2008; Hood & Jackson, 1991; Marston & Watts, 2003). By overlooking the many ideological concerns and pragmatic arguments affecting policymaking, the proponents of EP are not able to grasp why their recommendations often are either ignored by policy makers (Greenhalgh & Russell, 2009) or captured by managerial interests (Learmonth & Harding, 2006). This critique of the inability of the theoreticians and proponents of EP to grasp the political reality in which it operates is very convincing. Still, this critique ignores that the overall purpose of EP (and those endorsing its use) is not to grasp the political complexity of policymaking, but to give advice on which policy mechanisms or tools are most likely to bring about given political goals.

The three points of critique referred to above provide important insights into some of the undesirable political implications of PM and EP. However, they are inadequate. On the one hand, they seem to overstate their critique. Thus, if EP and PM are so bad, then why are they so desirable not only to policy makers, but also to many public managers, civil servants, and citizens? On the other hand, the existing critique seems to overlook the particular kinds of power at play in PM and EP, and how these forms of power variously depend on the freedom of those over whom these powers are exercised.

The main argument of this article is that by conceiving PM and EP as two distinct forms of government, we are not only able to get a better understanding of why they have spread to and proliferated in many/most OECD countries. This understanding allows us to illuminate how PM and EP in different ways may shape and limit the space of possible political and managerial action. It is argued that we ought to be critical of both forms of power: PM because it works through strictly regulated and accountable forms of freedom and EP because it depends not on freedom but a very particular and exclusive form of truth production.

The reminder of this article first outlines some conceptual and analytical reflections around the notion of government. On the basis of this, I then examine the key features of PM and its historical precursors. This is followed by an account of EP and its link to particular forms of governing. Finally, I sum up the findings and discuss why PM and EP are best considered dangerous in the sense that they may reduce the possibility of unsettling or reversing the power relations supporting the ways in which public administration is conducted in liberal democracies, rather than bad in the sense that PM and EP will necessarily have undesirable consequences only. In fact, both forms of governing have important, albeit different, merits.

Liberal Government

We may provide a more adequate, critical understanding of PM and EP by viewing these as (distinct) forms or mechanisms of power. By power, I am thinking neither of a capacity or resource held by a given actor to impose her will upon someone else, nor of an abstract (discursive or economic) structure working behind the backs of such actors. Instead, I am referring to the manifold administrative and calculative techniques, procedures, and practices by which someone is trying to shape the conduct of someone else. This is a form of power that the French philosopher Michel Foucault named government (Foucault, 2008). Government refers not to the institutional apparatus of the state or other political authorities, but to the rationalities (modes of reasoning) and technologies (devices of intervention) engaged in the art of governing individuals, groups, populations, or nation states. Foucault’s shorthand definition of government, the conduct of conduct, points out that government is both about the ways in which somebody is being governed by others and by him or herself.

More precisely, the term government implies analytical attention to the contact point between the governing of others and the governing of the self by the self. This is to say, the exercise of power by A over B does not necessarily exclude the freedom of B. On the contrary, what distinguished government from other forms of power, Foucault argues, is that it more or less systematically depends on the freedom of those over whom it is exercised (Foucault, 1982). Freedom is understood here not as the absence of power, but as the set of techniques of practices whereby individuals or organizations govern themselves. Except for extreme cases, such as Robison Crusoe on his island before the advent of Friday, such practices of freedom or self-government always take place under the influence of various social norms and power. Unlike almost all other political and social theorists, Foucault insisted on an open-ended conception of freedom as the ongoing and unfinished activity by which the self reflects upon and act upon itself (Foucault, 1994). These reflections and actions may take all kinds of forms depending on the historical and societal context. The only (minimum) criteria that Foucault adopts is that such self-practices are allowed a certain choice or room for maneuver. This very broad understanding of freedom has important analytical implications. Unlike almost all other theoreticians seeking to explain liberal or soft forms of governing, the Foucauldian conception avoids the problem of having to attribute false consciousness (Lukes, 1974) or misrecognition (Bourdieu, 1991, pp. 163-164) to those subjected to noncoercive forms of power. The whole idea of academics having a truer understanding of the interests of those subjected to power, or what freedom really is, is deliberately rejected by Foucault not only because this amounts to academic hubris, but also because it blocks for a more convincing understanding of why soft or liberal forms of power work. Accordingly, Foucault suggested that this relationship be grasped as one in which (liberal or soft) power works through freedom. More precisely, liberal forms of governing may be defined as the thoughts and mechanisms seeking that spur political goals, such as wealth, efficiency, and welfare, by latching on to and facilitating the ways in which individuals, groups and organizations act on themselves. In brief, this is a form of power that works by nurturing and structuring the freedom of those over or through whom it is exercised.

Foucault locates this kind of power (government) in the emergence of classical liberalism in the early 19th century and more recent forms of neoliberalism in the United States and West Germany after World War II. Liberalism here is to be understood neither as an ideology nor as a party political movement, but instead as a distinct rationality of government or governmentality, that is, a particular way of reasoning about how government should be exercised. This kind of rationality may be characterized by its problematizations (the problems it formulates and sets out to handle), the kind of knowledge that informs its way of reasoning, and the kind of technologies that support it. Rationalities of government then are not just mental representations or political ideas but ways of reasoning that are by definition always inscribed in particular ways of governing. Classical liberalism as a specific form of rationality is characterized by the assumption that civil society has a natural ability to govern itself and by a concern over excessive state intervention undermining this ability. In contrast, neoliberalism is characterized by a certain constructivism in the sense of facilitating and regulating particular forms of freedom (Foucault, 2008; Rose, 1999). In other words, the problem that neoliberal government seeks to address is not the dangers of excessive state intervention, but instead: How can we nurture and direct the self-governing capacities of individuals, groups, and organizations in ways that contribute to societal wealth and well-being? This difference between classical liberalism and neoliberalism is crucial because the latter in principle has no political limitations (Triantafillou, 2012, p. 179). Thus, we can always govern ourselves better, exercise our freedoms in a more effective or desirable manner. Accordingly, there is always a moral argument for governing better and more with a view to further stimulating the ability of individuals, groups, and organizations to exercise their freedom in desirable ways.

In the following, I use Foucault’s notions of power, freedom, and liberal government—as a particular constellation of power and freedom—to provide a critical account of PM and EP. I do so not by undertaking a full genealogy of PM and EP, but by teasing out their rationalities and how these are inscribed in more or less concrete policy programs and governing techniques. Thus, PM and EP are regarded here as particular forms of power informed by particular rationalities of government.

If PM and EP may be understood as distinct forms of power, then what kind of governmental rationalities do they link up to? In the following, I will substantiate the claim that PM adheres closely to neoliberal rationalities of government in as much as they try to operate through the facilitated and regulated freedom of public agencies and citizens. In contrast, EP sits somewhat uneasily with neoliberalism because of its fundamental assumption that policymaking should be based not on freedom but on authoritative knowledge about what works and what does not. Yet, as I will also show, EP has so far managed to link up with a politics of choice and, more broadly, with neoliberal rationalities of government.

PM: Governing Through Regulated Freedom

The systematic management and measuring of the performance of individuals and organizations can be dated back at least to the beginning of the 19th century (Chandler, 1977; Foucault, 1977; Hoskin, 1996). By the 1950s, such managerial techniques had spread to most major, private corporations in the capitalist, industrialized world. From around 1950, the U.S. Federal government would launch several initiatives to improve public planning and budgeting through new budgeting mechanisms, employee incentives and performance measures (Van Dooren, 2008).

Inspired by the early reform experiences in New Zealand (Boston, Martin, Pallot, & Walsh, 1996), PM measures spread in various forms and degree to many but not all OECD countries during the 1990s (Pollitt & Bouckaert, 2000; Pollitt & Bouckaert, 2004). Key elements in the New Zealand reforms included centrally defined performance goals and monitoring systems, economic incentive systems, contracts and competitive tendering, the separation of advisory, regulatory and delivery functions, and new financial management systems based on accrual accounting.

In the United States, President Clinton signed the Government Performance and Result Act 1993 (GPRA) to measure progress and hold federal agencies accountable for their results. Under the Bush administration, efforts were made to step up results oriented public policies and budgeting. Accordingly, in 2003 the U.S. Office of Management and Budgets launched the Program Assessment Rating Tool to evaluate the degree to which more than one thousand federal programs met their stated objectives (Gilmour, 2007). With the advent of the Obama administration, Program Assessment Rating Tool was replaced by the GPRA Modernization Act of 2010, which seemingly places more emphasis on creating leadership commitment and routines securing the actual use of performance information for making resource priorities.

In Europe, the United Kingdom seems to have been the front-runner with a number of PM reforms during the 1980s. Yet PM soon began to inform numerous public management reforms in continental Europe as well (OECD, 1995; Pollitt & Bouckaert, 2004). In particular, budget intensive policy areas, such as education, employment, health, and social security were subjected to often quite profound reforms in which PM played an important role. Presently, the debate seems less to be whether PM should be used or not, but rather how it may be adopted and integrated into public policies in the most effective manner and in ways that alleviate undesirable side effects (Moynihan, 2008; Van Thiel & Leeuw, 2002).

The practices taking place under PM are quite diverse. These include different types of performance measuring (which may include very different measures processes, outputs, or outcomes and involve very different indicators), various systems of accounting for performance, and diverse procedures for rewarding or punishing public agencies. It is this diversity that makes PM suit quite diverse political circumstances or seasons (Common, 1998; Hood, 1991). Of course, this diversity also makes it difficult to speak of PM as a distinct form of governing. Nonetheless, the delegation of administrative discretion of public service delivery to executive agencies, under more or less strict regimes of performance monitoring, seem to constitute a defining element of most PM reforms. Ever since the publication of Osborne and Gaebler’s Reinventing Government (Osborne & Gaebler, 1992, p. 25), the mantra that politicians should be doing the steering and others the rowing has been a key feature of public management reforms. The form of steering envisaged by Osborne and Gaebler is a specific liberal form of power that works not by (rules and regulations) determining the behavior of public agencies, but by invoking the self-governing capacities of public managers and frontline workers within a more or less strict regime of performance measures and targets.

The governing of public health may serve to illustrate how PM works by governing at a distance. Hospitals, pharmacies, and general practitioners have seen the introduction of performance measures and moves to those responsible for managing budgets (Ham, 2009, pp. 30-32). In Britain and many other OECD countries, hospitals were to be managed by a corps of general managers with executive powers and responsibility for devolved budgets (Dent, 2005; Harrison & Wood, 1999; Hasselbladh, Bejerot, & Gustafsson, 2008; Reichard, 2002). Simultaneously, public health managers and medical staff were urged to improve the quality of public health services through the publication of their achievements in meeting given standards and indicators. This information the quality of health services was to facilitate the creation of a quasi-market in service providers were to compete for patients now given the possibility to choose the unit providing the best quality. Two points can me made about the object of this knowledge. First, the more or less exclusive focus on production of knowledge pertaining to the conduct of the health providers, rather than the patient or citizen. Whether this knowledge is dubbed surrogate variables (Osborne, 1997, p. 185) or formal variables (Rose, 1996, p. 54), the point is that the knowledge produced under PM focuses almost exclusively on the processes or activities providing health services. In contrast, substantive knowledge on the health of the citizen or the population, such as epidemiology, no longer plays the key role it did under earlier forms of public health policies. This shift of focus in the production of knowledge is linked to the second point, namely that the key object of government is not the citizen or the population, but government itself. It is this folding back of government upon itself that is implied by the term governmentalization of government (Dean, 1999, p. 193). More precisely, what is being governed is not the health of citizens and populations, but those organizations in charge of governing and promoting public health. To the extent that PM is seeking to govern citizens at all, it is with a view to fostering their capacities as mature consumers able to make informed choices of public health services. In the case of public health, recent PM reforms may be interpreted as a form of neoliberal government that seeks to steer public health authorities (government) at a distance in the sense that it by a variety of methods seeks to make these authorities exercise and develop their freedom in ways that will result in “better” health services.

The public health sector reforms are, I think, indicative of a wider change of public administrations toward setting these free to and responsible for implementing politically defined goals in the ways they find most efficient. In contrast to EP, which will be examined below, PM often articulates a pragmatist ideal to problem solving and policymaking. Public managers, frontline workers and the citizens they serve should engage in ongoing collective experiments testing ways in which various mechanisms and solutions are able to meet changing local needs (Dewey, 1957; see also Sanderson, 2009). The ambition of such processes is not to identify scientifically tested and politically final solutions, but to constantly produce locally useful accounts suited to drive ongoing projects and reforms in a socially desirable direction through the active participation of local citizens and relevant groups. This pragmatist strain informing PM should not be exaggerated. PM is often overtly focused on checking the extent to which the activities of public administrations adhere to a set of rigid performance targets, that is, a style of policymaking squarely at odds with pragmatist ideals (Box, Marshall, Reed, & Reed, 2001; Stivers, 2000). Also, in practice PM measures have in many instances been informed by more or less clearly articulated desires to produce knowledge about what works, though in manners very different from EP. Nevertheless, the fact that the pragmatist ideal is often not realized does not mean that PM is antithetical to pragmatist approaches to problem-solving. The persistent debate over PM’s frequent short-circuiting of managerial self-determination and experimenting in practice may be taken as a sign of the strength of this regulative ideal rather than its weakness.

PM then may be regarded as a mode of governing that depends upon steering through the responsibility and the freedom of public agencies, private contractors, citizens, and others who are seen to be able to contribute to realization of political goals. As noted by Peter Miller and others, the calculative practices, such as accounting, that make up PM and measurement are seeking to harness, rather than repress the self-regulating capacities of public and private actors (Miller, 2001). Of course, the kind of freedom at play in PM is not an unregulated one, but one that is explicitly and often quite narrowly regulated by political goals and procedures that hold administrators accountable for their actions on the basis of various technical and professional standards (Shore, 2008). Public managers and frontline workers are urged and often legally obliged to provide accounts of their choice of methods and actions to their superiors and often to the wider public. In short, it is a freedom that allows governing at a distance (Rose & Miller, 1992). We should not overlook this dimension of regulation and power intrinsic to PM, nor should we dismiss the fact that it depends fundamentally on the freedom or self-steering capacities of civil servants, private contractors, citizens, and others. In sum, what distinguishes PM as a form of government is the way in which power is exercised through the nurtured, regulated, and accountable form of freedom of those held responsible for bringing about politically desirable goals. PM seems to be strongly informed by neoliberal rationalities of government in the sense that PM operates through the facilitated and regulated freedom of public agencies and citizens.

EP: Technocratic Governing

EP is grounded on the apparently simple and innocuous assumption that politics should be founded on the best available knowledge. The notion that political governing will benefit from being based on truthful (scientific) knowledge is not very new (Comte, 1875-1877). In fact, the building of modern territorial states has depended on scientific in general and statistical knowledge in particular (Desrosières, 1998, chapters 5 and 6; Scott, 1998). More recently, the dramatic rise of government and military funding of social science research during the 1950s gave U.S. social scientists a strong incentive not only to apply methods emulating the natural sciences, but also to prove the relevance of these for policy interventions (Steinmetz, 2005, pp. 296-299). By keeping more or less strictly to positivist methods of evaluating the effects of various political reforms, social scientists could simultaneously claim political relevance without being tied up to specific political goals: they were supposedly simply delivering rational tools to promote the goals chosen by democratically elected governments (Mirowski, 2005, pp. 164-165).

Although the idea that political interventions should be based on scientific knowledge may not be new, EP today differ from previous experiences in at least two ways. First, like PM it is part of the move to govern government. EP today is not only about identifying methods shaping the conduct of citizens in more effective ways, it is also employed to govern public agencies with a view to make them adopt more such effective methods. If EP shares with PM the ambition of governing government, it differs from PM by governing not through the freedom and experimentation of public agencies, but by urging them to adopt proven methods. This leads us to the second novelty of contemporary EP. Thus, the knowledge required to proof the efficacy of a method is linked to a fairly novel regime of knowledge production. Starting in the field of medical interventions during the late 1940s (Timmermans & Berg, 2003, pp. 89-90), the randomized controlled trial (RCT) has slowly become the predominant ideal for (the Golden Standard) producing knowledge about the effect of not only medical but also a whole range of other policy interventions. The RCT is a very distinct technique of veridiction or way of producing knowledge that purports to be truthful. It essentially implies that a group randomly selected from a larger population is subjected to a particular intervention and then compared—in terms of its behavior or other characteristics—with another randomly selected group that was not subjected to the intervention. Because this technique of truth production is very demanding and often not possible to apply in practice, alternative techniques are also accepted. However, the further these other techniques or methods deviate from RCT the less credibility is granted to their results and their potential use in policymaking. Consequently, we have an evidence hierarchy ranging from RCT at the top, over various more or less controlled designs in which statistical methods are applied to compensate for background variables, to diverse case studies and anecdotal knowledge (Sackett, Rosenberg, Gray, Haynes, & Richardson, 1996). Accordingly, the systematic reviews conducted since the 1990s of existing studies of particular interventions systematically valuates the studies reviewed according to their proximity to the RCT. Interestingly, systematic reviews have now themselves become not only part of but inserted themselves in the very top of the evidence hierarchy (Wayne, 2001).

The evidence hierarchy sketched above may be regarded as a regime of truth in which certain ways of producing the truth are granted more authority than others (Foucault, 1980, p. 132). This is not a matter of whether some forms of knowledge are truer than others, but a question of which procedures and standards of truth production are granted authority and which are not. Over the last two decades or so we have seen the diffusion this regime of truth migrate from the field of medical interventions into the field of education, employment and crime prevention in several OECD countries (Hansen & Rieper, 2009). In the United States, the idea that the selection of political means should be determined by the evidence of their effects has for some time informed a number of employment programs (Bloom et al., 1997; Maynard, Brown, & Schore, 1979). As exemplified by the federal Workforce Investment Act 1998, RCTs and other methods topping the evidence hierarchy were not the only accepted standards for producing evidence (Heinrich, 2007, p. 271). Yet, the tendency seems to be that evidence hierarchy seems to be playing an increasing role in that attempt to prove effects, not least because the hierarchy serves as a criterion for the screening of grant applications within various community programs, such as education, employment, crime, and drug abuse prevention (Weiss, Murphy-Graham, & Birkeland, 2005; Weiss, Murphy-Graham, Petrosino, & Gandhi, 2008). This regime of truth was reinforced by the Coalition for Evidence-Based Policy established in 2001 to promote the use of RCT to strengthen the effectiveness of social policies (Coalition for Evidence-Based Policy, 2011). In the following years, the Coalition managed to convince Congressional legislators into making evidence of effects an integral part of a number of federal social programs.

In Europe, developments have been somewhat slower, but the tendency seems to be the same. In the United Kingdom, where the attempts to pursue EP seem to have gone furthest, Blair’s New Labour government invested quite some political capital in adopting the Aspen Institute’s theory of change approach to policymaking (Strategic Policy Making Team, 1999). This involved the extensive use of theories and evidence in the design of public policies, though numerous other political concerns turned out to influence policymaking (Sullivan, 2011). Britain also saw the emergence of systematic reviews within the field of the social care and health conducted or commissioned by the Social Care Institute for Excellence. The Evidence for Policy and Practice Information and Co-ordinating Centre (EPPI), which was established as early as 1993, assists in the design of systematic reviews and disseminate results from these in the areas of education, public health, and various social policies (EPPI Centre, 2012; Oakley, Gough, Oliver, & Thomas, 2005). Finally, Prime Minister David Cameron formed The Behavioural Insights Team—or just the “Nudge Unit”—with a view to applying the principles of behavioral theory in the design public policy (Cabinet Office, 2011). On the basis of scientific knowledge preferably produced in ways akin to RCT, the unit is urging British public agencies to implement choice architectures designed to make citizens choose a behavior more in line with politically desirable goals (Haynes, Service, Goldacre, & Torgerson, 2012).

EP today then entails a form of policymaking that ideally is based on a form of truthful knowledge produced in a very particular way. This form of policymaking is politically dangerous for at least two reasons. First, it bypasses the freedom of the public administration to choose how—by what means—it will try to implement political goals. EP does not work through the freedom of public administration. Public managers, professionals, and frontline workers are supposed to adopt instruments and procedures based neither on their professional experience nor on the needs or expressed desires of the citizens who are the target of the intervention. Instead public administrations should adopt those instruments and procedures that according to knowledge produced as close to the Golden Standard as possible show that they are likely to bring about the politically desirable effects. The freedom of the public administration is no longer to include the choice of the instruments of intervention, but confined to consulting and applying the knowledge disseminated through the Campbell Collaboration and similar institutions. Accordingly, EP seems to be at odds with neoliberal forms of government to the extent that the latter is defined as a form of rule that systematically works through the freedom of those over whom power is exercised.

Second, EP may influence and narrow down the scope of possible political approaches and goals. This has to do with the kind of evidence that EP is based on. RCT and its neighbors in the top of the evidence hierarchy all rely on the availability of objective and quantifiable measures. The effects of a given intervention must be gauged on the basis of objective measures to ensure replication and they must be quantifiable to allow for evaluating the magnitude of a given effect. By implication, EP must be able to translate its goals into objective and quantifiable measures, how else can one check whether and to what extent a given intervention contributes to that goal? However, most political goals and visions are for a number of reasons often quite vague. Think of such goals as fairness, equality, social cohesion, trust, and being empowered. To translate such goals into objective and quantifiable measures may not be impossible, but in the process of doing so they are likely to take a much narrower and perhaps even different meaning than intended. Such political goals are very difficult to translate into simple quantifiable measures. For example, it has been argued that the systematic monitoring and benchmarking of employment services with a view to providing evidence on which kind of interventions are most effective in getting unemployed into work is creating a technological lock-in that favors one particular political approach and goal only, namely work-first (Triantafillou, 2011). Other employment policy approaches, such as human capital development, reduced work time, or negative income taxation that may serve alternative political goals such as securing political citizenship, reducing work related stress, promote social cohesion, and promoting economic and social equality, are systematically devalued or ignored. Similarly, Yanow has argued that the kind of evidence required by EP excludes other modes of research and bodies of knowledge that may feed into policymaking (Yanow, 2009). In general, if it is difficult to develop measurable indices of a desirable outcome, then it is by definition impossible to test whether specific interventions are generating this outcome. Accordingly, if the evidence of EP implies knowledge generated through RCT or similar procedures only, which seems to be increasingly the case, then the policy of EP may be reduced a quite narrow set of goals and approaches. In brief, EP may only be preaching what means should be adopted, but it may—unwittingly—also influence political approaches and goals.

Concluding Discussion

I have argued that PM and EP may be regarded as two distinct forms of governing that possibly, but not necessarily contradict each other. On the one hand, PM and EP work within a broadly neoliberal rationality of government in the sense that they are both part of the ambition of governing government, that is, the attempt to improve the functioning of public agencies by subjecting these to various governing measures. On the other hand, they differ importantly in terms of the problematizations, forms of knowledge and governing technologies constituting them. Whereas PM essentially works through the regulated and accountable freedom of public managers, frontline workers and citizens, EP depends on the systematic production of knowledge about the causal effects of policy intervention. The efficacy of the type of policy interventions endorsed by EP ultimately rests with the labor of experts and scientists, not with the freedom and participation of public agencies and the citizens they serve. In terms of their governmental rationalities, PM seems to constitute the paradigmatic example of neoliberalism whereas EP is informed not only by neoliberal but also by technocratic rationalities.

To better understand the proliferation and endurance of PM over the three decades in many OECD countries, we should pay attention to the particular kind of problematization informing it, namely: how to facilitate the self-steering and problem-solving capacities of public agencies? The diverse answers to this problem of government entail the use of governing technologies that not only depends upon, but systematically facilitates and structures the freedom of those over whom power is exercised. Contracting out, internal contracts, performance measurement, benchmarking, separation of purchaser and provider all seek to delegate administrative discretion over the production and delivery of services from political decision makers to a variety of public and private agencies. The knowledge produced performance measurement, benchmarking, and other monitoring technologies is attuned not to prove the causal efficacy of a certain intervention, but rather to demonstrate that the public agency—whatever its mode of operation—is achieving politically defined results (output standards). Some have rightly criticized this delegation of administrative discretion to be inadequate and to be subjected to (too) rigorous monitoring and centrally defined accountability procedures. Yet, the ideal of governing through more local autonomy has come with strong moral support and the ideal has been used by public agencies (and private contractors) to effectively create a room for deciding on which instruments and procedures should be employed. It is this room of more or less closely regulated freedom that makes it more suitable to characterize PM as dangerous than outright bad.

If the promise of freedom is what makes PM so attractive, it is also the particular kind of freedom that makes it problematic if not dangerous. First, the kind of freedom propagated and facilitated through PM is not the freedom to do just anything. PM urges public administrators to govern themselves according to a more or less narrow set of norms, standards, and targets favoring particular forms of conduct and particular outcomes of that conduct. Civil servants are expected if not obliged to govern themselves in very particular ways: they must constantly ask themselves how their services may be improved, how they can become more productive, how they can engage more actively with citizens to accommodate their needs, and finally they must provide accounts for their decisions and actions taken. I will not pass judgment on whether such a freedom is morally desirable or not, but instead note that many do not think so (Gay, 2000) and that it at least excludes many other forms of conduct, that is, other kinds of freedom. Second, for those public administrators who turn out to be unwilling or unable to manage their freedom in line with norms of activism and productivity a nonliberal or authoritarian set of interventions may apply. As argued by Dean and Rose authoritarian mechanisms of rule are not inimical to neoliberal government (Dean, 2002; Rose, 2000). It is exactly because the norm of freedom is so central to neoliberal government that anyone—be that a citizen, a company, or a public administration—who turns out not to be willing or able to exercise his or her freedom in a proper manner is regarded as a serious threat to society, a threat justifying the use of authoritarian means. For public administrators unwilling to constantly work to enhance performance such means may range from public harassment (naming and shaming) over sacking of individual employees and the closing down of entire public agencies to the legal prosecution of public managers who have not used public money cost-effectively.

We may now turn to EP. We saw above that EP potentially is at odds with neoliberal rationalities of government because it works not through the (regulated) freedom of public administration, but instead by deducing the selection of policy interventions from a very particular form of truth production. The specific problematization of government that EP is addressing is not that of how to stimulate the self-steering capacities of public agencies, but rather how to secure that public interventions are producing a given political goal? However, we should not exaggerate EPs conflict with neoliberal rationalities of government. The recent British attempt to apply the findings of behavioral science to deploy and test the efficacy of “choice architectures” shows that an apparently highly technocratic device ideally based on scientific evidence only, is employed within a mode of political rule dominated by choice. It is not up to politicians or scientists to decide how citizens, companies and communities should act. Instead, informed by the work of scientists, politicians are supposed to implement architectures that enable citizens and others to exercise their freedom and choose a lifestyle that is in line with current norms of civility, health, and industriousness. Thus, EP sits somewhat uneasily with neoliberal rationalities of government but if would be wrong to regard them as antithetical.

The attractiveness of EP is fairly obvious. If only we have more accurate knowledge of that which is being governed and the efficacy of various political instruments, then we may create a better society. In particular, EP may be a sound corrective to political beliefs in the wonders of certain favored policy instruments. For example, EP may temper the increasing use of PM by showing that the attempt to govern public services by a range of performance indicators may be counterproductive to longer term program impacts and efficiency (Heckman, Heinrich, & Smith, 2002). In Denmark, for example, the medical profession resisted, though in vain, the introduction of accreditation in the Danish health services on the ground of high economic costs and the absence of any scientific evidence of the clinical benefits of accreditation (Knudsen, 2011). It could even be argued that EP may pave way for more democratic forms of rule (Timmermans & Berg, 2003, p. 21). In fact, Donald Campbell’s Experimenting Society intended the continued development and testing of new ideas and policy measures as mechanisms that would invigorate democracy and challenge existing dogmas and authorities (Campbell, 1998). Until a few decades ago, authority in fields such as medicine, education and research, and social work was nearly monopolized by select professions who based and protected their authority on the basis of specific and exclusive forms of education. EP promises to break up this monopoly and create a playing field defined by a particular way of producing evidence which in principle is open to anyone. At the very least this kind of knowledge can be used by the public to hold the professions accountable to their decisions and actions within the said policy fields. At the very best EP may be regarded as a political technology in support of a more open and collaborative but not less dangerous form of governing that some have termed info-liberalism (Catlaw & Sandberg, 2012).

If EP holds certain benign potentials, it also comes with at least two distinct political dangers. First, it may entail that the freedom of public agencies is minimized. EP, at least in its existing forms, hinges on a very particular form mode of truth production that must be used by those forms of policymaking seeking to justify themselves as evidence-based. Ultimately, the freedom of public managers and frontline workers is reduced to ensure that this kind of knowledge is used as systematically as possible in the selection of policy mechanisms. It could be argued that public managers may gain influence at the expense of the professional frontline workers (doctors, educators, social workers) because of the possibility of holding professional decision making accountable to more or less universal and public available standards deducible from the systematic review studies. However, this influence over the professions comes at the expense of the kind of freedoms propagated by PM, notably the set of reflections, judgments, and actions by which public managers (and other civil servants) decide on how best to secure the quality and efficiency of public service delivery.

Second, the scope of political goals and approaches may be reduced those that are susceptible to be objectified and quantified. If only goals and approach amenable to RCT type testing are allowed, then a number of political programs, reforms, and visions may be discarded. So far, this danger seems remote. At least, the use of EP in the sense of policies being based on the methodological procedures of RCT remains rather limited within the social areas. True, the existing experiences with EP in the United States and the United Kingdom almost invariably show that when EP is put into practice it is not conducted “properly.” Yet rather than debating what constitutes true evidence, we may note the (increasing) legitimizing force emanating from the proposition that a particular policy intervention must be based on knowledge produced in ways belonging to the top of the (medical) evidence hierarchy, that is, in ways akin to RCT. Conversely, it is becoming increasingly difficult to defend a policy that is being attacked for lacking evidence, that is, a policy informed by forms of knowledge produced differently than those found in the top of the evidence hierarchy.

An important question for future research then is just how PM and EP will unfold and intertwine with each other as concrete political and managerial practices in the public administration. Will the endorsement of EP imply that public managers, frontline workers and citizens have less influence and voice in the making and delivery of public services? And if so, who are to be held responsible for cases of implementation deficits or poor service quality? Ultimately, we have to ask ourselves if we as social scientists concerned with real life political problems and possible solutions are ready to accept the evidence hierarchy as the only legitimate standard for our research and to replace public administrators as scapegoats when it turns out that policy instruments our research endorsed did not produce the intended results.