emerging issues Better Performance Management Some Single- and Double-Loop Strategies Wouter Van Dooren University of Antwerp ABSTRACT: New Public Management set off a new wave of performance management efforts in government. Recent performance literature has documented the shortcomings of performance management and provided recommendations for improvement. This emerging-issues article revisits the literature, making a distinction between recommendations for better implementation of what are seen as essentially good systems (single-loop learning) and recommendations that target the performance management systems themselves (double-loop learning). It is argued that in complex settings, performance management may benefit from other ways of doing performance management and needs to be more agile, more decentralized, and more political. KEYWORDS: complexity, implementation, performance management, politics of performance management, single-loop and double-loop learning, use of performance information Performance management has had a long ascendancy, and in all probability has a long road ahead. Hatry (2008), for instance, finds it hard to believe that performance management will not continue far into the future. Nonetheless, performance management is not without its problems. Practitioners, management consultants, and academics have sought solutions in response to the paradoxical and often problematic nature of performance management. Based on the recent performance literature,1 the present article identifies some emerging issues for the next round of performance management improvement. The discussion begins by outlining some paradoxes in performance measurement and management that echo some of the main challenges. Next, Argyris & Schn's (1996) distinction between single- and double-loop learning is used to categorize the character of the proposed solutions to the challenges. Single-loop solutions try to mitigate the implementation problems of performance management. The main argument is that better results in performance management 420 Public Performance & Management Review, Vol. 34, No. 3, March 2011, pp. 420-433. 2011 M.E. Sharpe, Inc. All rights reserved. 1530-9576/2011 $9.50 + 0.00. DOI 10.2753/PMR1530-9576340305 Van Dooren / Better Performance Management 421 can be obtained by better implementation. The message recommends having a \"second go\" with an essentially good system. In contrast, the double-loop cluster of responses proposes to change (parts of) the system. The message is not just to try it again, but to do it differently. In complex and ambiguous contexts, double-loop proposals have the greatest potential for improving the added value of performance management. Paradoxes in Performance Measurement and Management New Public Management (NPM) reforms worldwide have introduced a variety of performance measurement and management practices. Performance management has gradually become an integral part of modern governance arrangements (Pollitt & Bouckaert, 2004). In recent years, evaluative research has uncovered some paradoxes in current practice. Four of these paradoxes are discussed below: counting the uncountable, relying on but not trusting professionals, paralysis by analysis, and the difficulty of combining performance management with accountability arrangements. Counting the Uncountable In Albert Einstein's office at Princeton University, it is said, there was a sign stating: \"Not everything that counts can be counted, and not everything that can be counted counts.\" Performance measurement adepts sometimes forget this insight. Managers and politicians have inferred, from the conviction that \"what gets measured, gets done,\" that \"what does not get measured, does not get done.\" This logical but incorrect inference is reinforced by management consultants advocating a quest for the ultimate set of key performance indicators (KPIs). The latter are exemplified by Kaplan and Norton's (1996) work on the balanced scorecard. They suggested that managerial attention should focus on a handful of \"key performances.\" KPIs are almost always only a subset of the total array of services in public sector organizations. Knowing that KPI thinking is the managerial mindset, it is not uncommon for middle managers in large organizations to lobby to get the activities of their division into the KPI set in order to be taken seriously at top managerial and political levels (Van Dooren, 2006). In the past decade, several performance management experts have pleaded for a focus on measuring outcomes instead of outputs or processes (Hatry, 2002; Perrin, 2003). They argue that only outcomes are \"really\" key results. What is key are results that matter for society. It does not matter how many police patrols are negotiating the streets (an output); citizens want safety (an outcome). Therefore, performance measurement should primarily focus on outcomes. Yet, and herein lies the paradox, outcomes are in many instances very hard to count. What is measured gets attention, but many important dimensions are not measurable. 422 ppmr / March 2011 Distrusting Professionals, But Relying on Them A second paradox is the ambiguous attitude of the performance management doctrine toward expertise and professionals. On the one hand, professionals are central to better performance. In fact, the NPM motto \"Let managers manage\" suggests strong confidence in the professionalism of managers (Kettl, 1997). Similarly, managers are expected to be entrepreneurs and leaders who bring out the best in the personnel under their supervisionnot in command-and-control style, but by empowerment (Osborne & Gaebler, 1993). Yet performance management systems seem to express a certain distrust of professionals. Davies and Lampel (1998), assessing performance management in the British National Health Service, argue that managers primarily used performance information in a tactical way, in order to intervene in the doctor-patient relationship. Hence, they developed a vast number of indicators to counterbalance the doctor's professional knowledge. Radin (2006) provides the example of the British Research Assessment Exercise (RAE), which audited the research quality of universities based almost solely on such performance indicators as the number and type of publications. These performance indicators replace professional judgment on the quality of research. Similarly, Radin points to the No Child Left Behind initiative in public schools in the United States. She asserts that the most vigorous critique was based on standardized tests that allegedly did not leave enough room for teacher discretion. There is a clear paradox. On the one hand, trust in professionalism is vital in an increasingly complex society. On the other, we are reluctant to grant this trust and hence fall back on control and audit. In auditor circles, the adage \"In God we trust, the rest we audit\" is well appreciated. Nevertheless, according to Power (1999), audits are to a large extent ceremonialhe speaks of rituals of verification. Many audits create an illusion of control. Similar arguments can be made for other performance measurement regimes in both the public and private sector (Van de Walle & Roberts, 2008). The failure of rating agencies that was exposed by the financial crisis is a case in point for the latter. Paralysis by Analysis Decision-makers have to process a huge body of information found in budgets, audits, impact analyses, evaluation studies, memoranda from interest groups, laws and jurisdictions, personal communications, and other sources. Almost superhuman analytical capacity is required to process all of this. Performance information comes on top of the pile, increasing the risk of an information overload. Although performance management was devised to improve decisions, it may well lead to paralysis. Thus one should not be surprised that practitioners consider selectivity in measurement to be one of the key challenges for implementing performance management (Mayne, 2007). Van Dooren / Better Performance Management 423 If Everyone Is Accountable, then No One Is There is increasing awareness that public organizations cannot be effective on their own. A considerable literature has developed on collaboration, partnerships, and networks.2 As a result of collaboration, the responsibilities for performance are shared. Hence, when many organizations participate, it becomes more difficult to hold a single organization accountable for results. And if many are accountable, there is a risk that no one is taking responsibility for failure and everybody for success. Should we then stick to traditional accountability schemes with one principal and one agent? Probably not. The willingness to collaborate can erode when single-accountability schemes are maintained. Denhardt & Aristigueta (2008) demonstrated that the typical approaches to performance management affect partnerships and collaborations. Performance-based accountability systems tend to undermine collaborative efforts unless they are accompanied by other strategies for providing an impetus for alignment and collaboration across agencies. Single-Loop: Better implementation The usual response to the paradoxes of performance management is to improve implementation. This is single-loop learning. The basic premises of performance management are not put into question. The idea is that if only we try harder, performance management will improve. The discussion that follows considers four strategies that are often proposed for better implementation of performance management. Improve the Quality of Performance Information A first strategy is to improve the quality of the information. The attention paid by organizations to quality matters is not always evident. Research suggests that often only modest attention is paid to quality-assurance practices in the area of performance measurement (Mayne, 2007). Nonetheless, Hatry (2008) argues that an investment in the many dimensions of quality can ratchet up the use of performance information. He mentions validity, accuracy, timeliness, analysis, and presentation as some important quality dimensions. Clear and timely presentation of performance information may remedy the information overload of decisionmakers. Professionals may trust performance information more when validity and accuracy increase. Better analysis may help to focus performance management on what matters and to single out accountability issues. Quality of performance information alone, however, does not guarantee the use of performance information. As argued above, performance information competes with other sources of information for decision-makers' attention. Moreover, 424 ppmr / March 2011 research on gaming has revealed that sometimes information is selected because it fits a predetermined agenda and not because it is intrinsically good (Bevan & Hood, 2006; Perrin, 1998). This should not come as a surprise. Studies on knowledge utilization have made the same point for the use of research evidence in policy-making (Innes, 1990; Weiss, 1977). It should be clear in this context that quality improvements will not help to overcome such issues as disputes about accountability or resistance from professionals. Leadership A survey by the Organization for Economic Cooperation and Development (OECD) found that strong leadership by politicians and managers is a central element in the success of performance management (Curristine, 2005). Someone has to put his or her shoulder behind a performance management effort and develop a measurement strategy. Preferably, this will be a person who carries some weight. Good leadership may be a response to the paradoxes, because good leaders take the uncountable into account, quickly identify core bits of information, motivate professionals, and hold people to account in a fair way. However, as a concept, leadership is poorly defined, and hence interpretation of the OECD findings is complicated. In particular, in survey research such as the OECD's, there is the risk that respondents use leadership as a quick fix when they cannot point to more precise factors. Hence, better implementation through better leadership is not a very actionable recommendation. The issue of leadership certainly raises a host of secondary questions (Van Wart, 2003): Who should the leader be? What traits are important for performance leadership? Where does leadership in performance come from, and how can it be sustained? What should leaders do in which circumstances? It does not suffice to state that better performance management depends on better leadership. It is also necessary to make clear what leadership is and what leaders need to do. Ownership \"Ownership\" is another somewhat magical word in management discourse. Implementation failures are regularly said to be caused by a lack of it. Mayne (2007), for instance, notes that a system built on filling in performance information forms for others, forms of no apparent use to those down the line, is unlikely to be robust and to survive over time. Better implementation of performance measurement and management requires that those who are affected by the system have to accept and internalize the system (Van Dooren, Bouckaert, & Halligan, 2010). In a thrust to ensure ownership, performance management reforms often fall victim to overcommitment (Pollitt, 2008). Many people need to be convinced before a performance management system can be introducedpoliticians, top Van Dooren / Better Performance Management 425 and middle managers, professionals and front-line workers, to name a few. Hence, an understandable strategy is to create high expectations and play down the costs. Yet, although this strategy may prove successful in the short run, it almost definitely will boomerang in the medium term. Typically, the costs of a performance management system are tangible and become apparent relatively shortly after the introduction of the system (Bouckaert, 1993). Benefits, on the other hand, are intangible and may only appear in the longer term. Disillusionment with performance systems that do not (yet) deliver may undermine confidence, and therefore the failure of the performance management effort may become a self-fulfilling prophecy. Ownership strategies are relatively fragile. It takes much more effort to build acceptance of a performance management system than to lose it. Variations of Integration Integration, coordination, formalization, consistency, coherence, routinebuilding, and alignment are some of the most common keywords for those who want to fix performance management without questioning its blueprint. See, for instance, the report of the management consulting firm Accenture (2006) as an example. Bouckaert and Halligan (2008) are on the same line of thinking when they stress the importance of integration in the shift from a model of performance administration to genuine performance management and even performance governance. Although the importance of integration and coordination is undeniable, their limitations must be acknowledged (Laegreid, Roness, & Rubecksen, 2008). Complexity and change regularly tear carefully coordinated systems apart. The desire to coordinate all efforts in advance may lead to delay and even deadlock. In some instances, it may make more sense to remedy the consequences of ill-coordinated performance efforts than to embark on excessively ambitious coordination efforts (Johnsen, 1999). Double-Loop: New Ways of Doing Performance Management The four suggested strategies for better implementation may substantially mitigate the paradoxes identified in the first section, but, as noted, they have some limitations. Recent research on performance management suggests new ways of doing performance management. Some of these new ways are emerging not only in theory but also in practice, as illustrated by the contemporary use of CitiStat (Abramson & Behn, 2006). The purpose of this section is to dig deeper into these proposals. It argues, first, that the main challenge to performance management lies in making it ambiguity-proof. In the next paragraphs, more concrete implications of this argument are proposed. 426 ppmr / March 2011 Making Performance Management Ambiguity-Proof Contemporary public administrationlike society at largeis often described as being ambiguous (Weick, 1995). Ambiguity is characterized by the absence of meaning, rather than by uncertainty and the absence of information (March & Olsen, 1976). Sense-making in society has become multilayered and more fragmented. Arguably, when we all have our own truths, policy-making and management become more complex. Noordegraaf & Abma (2003) add that current performance management, which they label as management by measurement, only fits the rare unambiguous contexts. Defined as such, not many unambiguous situations will be found. Since ambiguity is everywhere, the prospects for performance management are, in this view, rather limited. An alternative to giving up on performance management is to rethink it in order to make it \"ambiguity-proof.\" This can only be done by taking complexity and ambiguity as a given, and changing the practice of performance management on this foundation. Performance information should enable interpretative processes of sense-making. Along these lines, Radin (2006) argues that many problems with performance measurement and management can be attributed to faulty points of departure. She argues that intelligence is not based on universal principles and literal meanings. Rather, multiple sources, situational knowledge, and literal and symbolic meanings are important in sense-making. It should also not be assumed that values, politics, and conflict are not at play in performance management. They clearly are. Finally, the assumption of linear cause-and-effect relations, as well as clear goals and planned change, does not survive the reality check. On the contrary, Radin paints a picture of complexity, interdependence, and unplanned change. Implications What would an ambiguity-proof performance management system look like? Three features are discussed belowthe need for performance management to be agile, decentralized, and political. The Need for Agility Kravchuk and Schack (1996) refer to Ashby, a cybernetics scholar, who posited that only complexity can absorb complexity. Rigid information systems are not able to understand rising complexity in the environment. In the most extreme cases, chaos will appear to reign due to the ever-increasing gap between practical experience and the knowledge base as provided by the information system. Information (what we believe we know) and practice (what we experience) risk becoming separate worldsone orderly, where objectives are set and performance targets are reached, and one chaotic, where people are mainly trying to muddle through the day. The main implication would be that performance indicators (PIs) should be Van Dooren / Better Performance Management 427 used for learning, and less so for accountability (Delancer Julnes, 2008). PI-based accountability requires stability for the period for which targets are set. Yet, as research in Australia and New Zealand has demonstrated, not many fields remain stable for three to six years (Carlin, 2006). As a result, accountability erodes accordingly (Gregory & Lonti, 2008). In addition to stability, accountability requires relatively univocal PIs that do not allow for much interpretation. PIs have to be an accurate and undisputed representation of \"real\" performance. Unlike accountability, learning does not require the same stability and robustness. On the contrary, performance measurement is part of a permanent dialogue in order to make sense of complexity (Moynihan, 2008). Hence, indicators can and should be adjusted in response to contextual changes and new insights. This proposal does not suggest that managers and professionals should not be accountable for performance. It only suggests that accountability through performance indicators (PIs) is probably not compatible with the need to have agile measurement systems. Accountability, however, does not solely depend on PIs. Performance assessment can be qualitative, quantitative, or both. In a complex environment, it might make more sense to hold managers accountable for the way they facilitate learning from performance indicators, rather than for the performance indicators themselves. \"Guerrilla Tactics\" and Decentralization Organizations typically have an undercurrent of repeated decisions they have to make. To these recurrent cycles, a constant stream of unique one-off decisions is added. In recent decades, the relative importance of the stable, recurrent processes has decreased. Kettl (2002) argues that the traditional U.S. public administration boundaries of mission, resources, capacity, responsibility, and accountability must be managed in an increasingly complex and politicized context, necessitating additional negotiation and collaboration between systems and agencies. These complex parallel processes are, in a unique way, shaped by situational requirements of time and place (Pollitt, 2008). Recurrent financial, human resources management, and contract cycles have been the main vehicle for incorporating performance information in decisionmaking (Bouckaert & Halligan, 2008). Without doubt, these cycles will remain the foundation of performance management in the future. Yet top-down performance management on a yearly basis (as in the budget cycle) or a monthly basis (as in many balanced-scorecard systems) will need to be supplemented by flexible efforts to provide performance information on demand. Since complex, unique processes will gain importance, the timing (when the performance information is used) and the locus (where it is used) will be challenged. With regard to timing, performance management in complex contexts may need to resort to \"guerrilla tactics.\"' In complex policy and management processes, the 428 ppmr / March 2011 demand for performance information can arise relatively unexpectedly. At the same time, demand can fade away as quickly as it came about. In such a context, expert personnel are needed to quickly infuse complex processes with performance information. As performance information brokers, they need to be able to both capture the need for and understand the availability of performance information. Like guerrillas, they use hit-and-run tactics to improve the quality of the often unpredictable processes of making sense of performance. Regarding the locus, performance management in complex contexts may benefit from stronger decentralization. Rather than devising top-down systems, performance management needs to be in the hands of the middle managers and front-line supervisors who best understand the situational requirements. This strategy might bring about the capacity of public managers that Behn (2004) calls performance leadership. He compared the performance leadership model to a focus on performance systems and structures. He writes that \"rather than to develop public managers with the leadership capacity to improve the performance of their agencies, we have sought to create performance systems that will impose such improvements\" (p. 3). The same reasoning can be applied within agencies. Middle managers and front-line supervisors need the leadership capacity to improve performance. This includes drawing lessons from performance indicators. For performance budgeting, the decentralized, action-based approach would suggest that performance information should be introduced into budget negotiations at the microlevel rather than systematically reported in the budget document that is voted in the legislature. Since the budget document is mainly an after-the-fact codification of political negotiations that have already taken place, performancebased budget formats run the risk of becoming a bureaucratic exercise (Sterck, 2007). There is some evidence supporting the effectiveness of this decentralized and action-based approach. Bourdeaux (2008), for instance, suggests that performance information mainly enters legislative discourse through policy networks maintained by key committee staff, rather than through executive budget reports and requests. Some confirmation is also found in an OECD survey on performance budgeting showing that countries use performance information to inform, but not to determine, budget allocations (Curristine, 2005). Furthermore, as Curristine (2005) argues, much of the \"linking\" of performance and financial information has taken place simply to provide them in the same report. A study by Melkers and Willoughby (2001) at the local and county levels found that the strongest usefulness of measures within a budget cycle occurs during budget development, with their importance lessening as the budget process proceeds (Melkers & Willoughby, 2005). Political Awareness Some time ago, Innes (1990) observed that the only way to keep data-gathering out of politics is to collect irrelevant data. Good performance information should Van Dooren / Better Performance Management 429 strengthen the evidence base for solving the political problems of who gets what, when, and howthe classic and broad definition of politics by Lasswell (1936). Issues of who gets what, when, and how are at play at all levelsin governmentwide policy-making, in policy sectors and networks, in organizational management, and in micro-management. Performance indicators can elevate the quality of political discussion at all these levels. Therefore, it is not suggested that political institutions (the executive, legislature, parties, etc.) have to interfere with all performance issues at all levels. Rather, the object is to stress the importance of recognizing the political nature of performance management. The first implication is that performance management should involve more rather than fewer actors. In complex settings, as argued above, performance management is mainly about sense-making. Hence, performance management can be expected to have the greatest impact when different perspectives are drawn into the dialogue. Obviously, the interests around the table have to be relevant and the number of participants needs to be workable. A second implication is that performance management should deal with controversy rather than suppress it. Performance information should not be an authoritative argument used to end conflicting views on who gets what, when, and how. Rather, it should underpin a careful argumentation of causes, consequences, and priorities in performance. It should bring controversy to a higher level of argumentationmore evidence based and more focused on output and outcome. The assumption is that high-quality dialogue will lead to improved judgment and decision-making. The preceding paragraphs deal with the political nature of performance management, and not so much with political institutions, such as the legislature, the executive, and the political parties. There are also efforts, however, to strengthen the role of performance information in the political system (Bourdeaux, 2008; Johnson & Talbot, 2008; Moynihan, 2009). Such initiatives will only be successful when they acknowledge the different values and positions of political actors. Performance information that promises to end political debates, to get political argumentation out of the political system, is irrelevant at best, but potentially harmful. Disagreement is essential for the functioning of democracy, and therefore performance information should primarily refocus political debate rather than curb it (Mouffe, 2000). Conclusion Recent performance literature has documented the shortcomings of performance management and recommended ways to improve it. Some of the recommendations prescribe better implementation of performance management (single-loop learning), while others suggest new ways of doing it (double-loop learning). Given 430 ppmr / March 2011 that in complex and ambiguous contexts, performance management risks being decoupled from practice and hence irrelevant for decision-making, this article argues that double-loop learning should be encouraged. New ways are needed to make performance management ambiguity-proof. These may include more agile measurement systems that adapt to a changing environment, managing performance close to the action, and a better understanding of the political nature of performance management. Notes 1. Radin (2006) and Moynihan (2008) studied the American performance management movement, with a particular focus on the Government Performance and Results Act (GPRA) and the Program Assessment and Rating Tool (PART). De Bruijn (2004) mainly analyzed experiences in the Dutch public sector. The UK House of Commons (2003) investigated English practices. The IBM Center for the Business of Government puts out a steady stream of research and publications on how to improve performance management (Boyle, 2009; Kloby & Callahan, 2009; Metzenbaum, 2009). Bouckaert and Halligan (2008) wrote a more theoretical and international comparative evaluation of performance management. Melkers and Willoughby (2001, 2005) published survey evidence at the state, local, and county levels in the United States. 2. See, for instance, Milward & Provan (2000), Vangen and Huxham (2003), Agranoff (2005), and a special issue of Public Management Review on network effectiveness (Mandell & Keast, 2008). References Abramson, M.A., & Behn, R.D. (2006). The varieties of CitiStat. Public Administration Review, 66(3), 332-340. Accenture. (2006). Accenture point of view: Performance management. Washington, DC. Agranoff, R. (2005). Managing collaborative performance. Public Productivity & Management Review, 29(1), 18. Argyris, C., & Schn, D. (1996). Organizational learning: A theory of action perspective. Reading, MA: Addison-Wesley. Behn, R.D. (2004). Performance leadership: 11 better practices that can ratchet up performance. Washington, DC: IBM center for the business of government. Bevan, G., & Hood, C. (2006). What's measured is what matters: targets and gaming in the British health care sector. Public Administration, 84(3), 517-538. Bouckaert, G. (1993). Measurement and meaningful management. Public productivity and management review, 17(1), 31-43. Bouckaert, G., & Halligan, J. (2008). Managing performance: International comparisons. London: Routledge. Bourdeaux, J. (2008). Integrating performance information into legislative budget processes. Public Performance & Management Review, 31(4), 547-569. Boyle, R. (2009). Performance reporting: Insights from international practice. Washington, DC: IBM Center for the Business of Government. Carlin, T.M. (2006). Victoria's accrual output based budgeting system-delivering as promised? Some empirical evidence. Financial Accountability & Management, 22(1), 1-19. Curristine, T. (2005). Performance information in the budget process: Results of OECD 2005 questionnaire. OECD Journal on Budgeting, 5(2), 54. Van Dooren / Better Performance Management 431 Davies, H.T., & Lampel, J. (1998). Trust in performance indicators? Quality and Safety in Health Care, 7(3), 159-162. De Bruijn, H. (2004). Managing performance in the public sector. London: Routledge. De Lancer Julnes, P. (2008). Performance measurement beyond instrumental use. In W. Van Dooren & S. Van de Walle (Eds.), Performance information in the public sector: how it is used (pp. 58-72). Basingstoke, UK: Palgrave Macmillan. Denhardt, K., & Aristigueta, M. (2008). Performance management systems: Providing accountability and challenging collaboration. In W. Van Dooren & S. Van de Walle (Eds.), Performance information in the public sector: how it is used (pp. 106-125). Basingstoke, UK: Palgrave Macmillan. Gregory, R., & Lonti, Z. (2008). Chasing shadows? Performance measurement of policy advice in the New Zealand government departments. Public Administration, 86(3), 837-856. Hatry, H.P. (2002). Performance measurement: Fashions and fallacies. Public Performance & Management Review, 25(4), 352-358. Hatry, H.P. (2008). The many faces of use. In W. Van Dooren & S. Van de Walle (Eds.), Performance information in the public sector: how it is used (pp. 125-140). Basingstoke, UK: Palgrave Macmillan. House of Commons Public Administration Select Committee. (2003). On target government by measurement (Fifth report of Session 2002-2003, vol. 1). Innes, J.E. (1990). Knowledge and public policy: The search for meaningful indicators. New Brunswick, NJ: Transaction. Johnsen, A. (1999). Implementation mode and local government performance measurement: A Norwegian experience. Financial Accountability & Management, 15(1), 41-66. Johnson, C., & Talbot, C. (2008). UK Parliament's scrutiny of public services agreements: a challenge too far? In W. Van Dooren & S. Van de Walle (Eds.), Performance information in the public sector: how it is used (pp. 140-157). Basingstoke, UK: Palgrave Macmillan. Kaplan, R.S., & Norton, D.P. (1996). The balanced scorecard: Translating strategy into action. Boston: Harvard Business School Press. Kettl, D.F. (1997). The global revolution in public management: Driving themes, missing links. Journal of Policy Analysis and Management, 16(3), 446-462. Kettl, D.F. (2002). The transformation of governance. Baltimore: Johns Hopkins University Press. Kloby, K., & Callahan, K. (2009). Moving toward outcome-oriented performance measurement systems. Washington, DC: IBM Center for the Business of Government. Kravchuk, R.S., & Schack, R.W. (1996). Designing effective performance measurement systems under the Government Performance and Results Act of 1993. Public Administration Review, 56(4), 348-358. Laegreid, P., Roness, P. G., & Rubecksen, K. (2008). Performance information and performance steering: Integrated system or loose coupling. In W. Van Dooren & S. Van de Walle (Eds.), Performance information in the public sector: how it is used (pp. 125-140). Basingstoke, UK: Palgrave Macmillan. Lasswell, H. (1936). Politics: Who gets what, when and how? New York: Whittlesey House. Mandell, M., & Keast, R. (2008). Introduction. Public Management Review, 10(6), 687. March, J.G., & Olsen, J.P. (1976). Ambiguity and choice in organizations. Bergen, Norway: Universitetsforlaget. Mayne, J. (2007). Challenges and lessons in implementing results-based management. Evaluation, 13(1), 87-109. 432 ppmr / March 2011 Melkers, J., & Willoughby, K. (2005). Models of performance-measurement use in local governments: Understanding budgeting, communication, and lasting effects. Public Administration Review, 65(2), 180-190. Melkers, J.E., & Willoughby, K.G. (2001). Budgeters' views of state performance-budgeting systems: Distinctions across branches. Public Administration Review, 61(1), 54-64. Metzenbaum, S. (2009). performance management recommendations for the new administration. Washington, DC: IBM Center for the Business of Government. Milward, H.B., & Provan, K.G. (2000). Governing the hollow state. Journal of Public Administration Research and Theory, 10(2), 359-380. Mouffe, C. (2000). The democratic paradox. London: Verso Books. Moynihan, D.P. (2008). The dynamics of performance management: Constructing information and reform. Washington, DC: Georgetown University Press. Moynihan, D.P. (2009). Through a glass, darkly. Public Performance & Management Review, 32(4), 592-603. Noordegraaf, M., & Abma, T. (2003). Management by measurement? Public management practices amidst ambiguity. Public Administration, 81(4), 853-871. Osborne, D., & Gaebler, T. (1993). Reinventing government: How the entrepreneurial spirit is transforming the public sector: From schoolhouse to statehouse, city hall to the Pentagon. Reading, MA: Addison Wesley. Perrin, B. (1998). Effective use and misuse of performance measurement. American Journal of Evaluation, 19(3), 367. Perrin, B. (2003). Implementing the vision: Addressing challenges to results-focused management and budgeting. Paris: Organization for Economic Cooperation and Development. Pollitt, C. (2008). Time, policy, management: Governing with the past. Oxford: Oxford University Press. Pollitt, C., & Bouckaert, G. (2004). Public management reform: A comparative analysis. Oxford: Oxford University Press. Power, M. (1999). The audit society: Rituals of verification. Oxford: Oxford University Press. Radin, B.A. (2006). Challenging the performance movement: Accountability, complexity, and democratic values. Washington, DC: Georgetown University Press. Sterck, M. (2007). The impact of performance budgeting on the role of the legislature: A four-country study. International Review of Administrative Sciences, 73(2), 189-203. Van de Walle, S., & Roberts, A. (2008). Publishing performance information: An illusion of control? In W. Van Dooren & S. Van de Walle (Eds.), Performance information in the public sector: How it is used (pp. 301-323). Basingstoke, UK: Palgrave Macmillan. Van Dooren, W. (2006). Performance measurement in the Flemish public sector: A supply and demand approach. Ph.D. dissertation, University of Leuven. Van Dooren, W., Bouckaert, G., & Halligan, J. (2010). Performance management in the public sector. London: Routledge. Vangen, S., and Huxham, C. (2003). Enacting leadership for collaborative advantage: Dilemmas of ideology and pragmatism in the activities of partnership managers. British Journal of Management, 14(1), 61-76. Van Wart, M. (2003). Public-sector leadership theory: An assessment. Public Administration Review, 63(2), 214-228. Weick, K.E. (1995). Sensemaking in organizations. Thousand Oaks, CA: Sage. Weiss, C.H. (1977). Using social research in public policy making. Lexington, MA: Lexington Books. Van Dooren / Better Performance Management 433 Wouter Van Dooren is assistant professor in the department of political science of the University of Antwerp (Belgium). As a member of the public administration and management research groups, he does research on performance, performance measurement, and management. He is also cochair of the permanent study group on public sector performance of the European Group of Public Administration (EGPA). He recently coauthored (with Geert Bouckaert and John Halligan) a textbook on performance management in the public sector published by Routledge. Copyright of Public Performance & Management Review is the property of M.E. Sharpe Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. emerging issues Better Performance Management Some Single- and Double-Loop Strategies Wouter Van Dooren University of Antwerp ABSTRACT: New Public Management set off a new wave of performance management efforts in government. Recent performance literature has documented the shortcomings of performance management and provided recommendations for improvement. This emerging-issues article revisits the literature, making a distinction between recommendations for better implementation of what are seen as essentially good systems (single-loop learning) and recommendations that target the performance management systems themselves (double-loop learning). It is argued that in complex settings, performance management may benefit from other ways of doing performance management and needs to be more agile, more decentralized, and more political. KEYWORDS: complexity, implementation, performance management, politics of performance management, single-loop and double-loop learning, use of performance information Performance management has had a long ascendancy, and in all probability has a long road ahead. Hatry (2008), for instance, finds it hard to believe that performance management will not continue far into the future. Nonetheless, performance management is not without its problems. Practitioners, management consultants, and academics have sought solutions in response to the paradoxical and often problematic nature of performance management. Based on the recent performance literature,1 the present article identifies some emerging issues for the next round of performance management improvement. The discussion begins by outlining some paradoxes in performance measurement and management that echo some of the main challenges. Next, Argyris & Schn's (1996) distinction between single- and double-loop learning is used to categorize the character of the proposed solutions to the challenges. Single-loop solutions try to mitigate the implementation problems of performance management. The main argument is that better results in performance management 420 Public Performance & Management Review, Vol. 34, No. 3, March 2011, pp. 420-433. 2011 M.E. Sharpe, Inc. All rights reserved. 1530-9576/2011 $9.50 + 0.00. DOI 10.2753/PMR1530-9576340305 Van Dooren / Better Performance Management 421 can be obtained by better implementation. The message recommends having a \"second go\" with an essentially good system. In contrast, the double-loop cluster of responses proposes to change (parts of) the system. The message is not just to try it again, but to do it differently. In complex and ambiguous contexts, double-loop proposals have the greatest potential for improving the added value of performance management. Paradoxes in Performance Measurement and Management New Public Management (NPM) reforms worldwide have introduced a variety of performance measurement and management practices. Performance management has gradually become an integral part of modern governance arrangements (Pollitt & Bouckaert, 2004). In recent years, evaluative research has uncovered some paradoxes in current practice. Four of these paradoxes are discussed below: counting the uncountable, relying on but not trusting professionals, paralysis by analysis, and the difficulty of combining performance management with accountability arrangements. Counting the Uncountable In Albert Einstein's office at Princeton University, it is said, there was a sign stating: \"Not everything that counts can be counted, and not everything that can be counted counts.\" Performance measurement adepts sometimes forget this insight. Managers and politicians have inferred, from the conviction that \"what gets measured, gets done,\" that \"what does not get measured, does not get done.\" This logical but incorrect inference is reinforced by management consultants advocating a quest for the ultimate set of key performance indicators (KPIs). The latter are exemplified by Kaplan and Norton's (1996) work on the balanced scorecard. They suggested that managerial attention should focus on a handful of \"key performances.\" KPIs are almost always only a subset of the total array of services in public sector organizations. Knowing that KPI thinking is the managerial mindset, it is not uncommon for middle managers in large organizations to lobby to get the activities of their division into the KPI set in order to be taken seriously at top managerial and political levels (Van Dooren, 2006). In the past decade, several performance management experts have pleaded for a focus on measuring outcomes instead of outputs or processes (Hatry, 2002; Perrin, 2003). They argue that only outcomes are \"really\" key results. What is key are results that matter for society. It does not matter how many police patrols are negotiating the streets (an output); citizens want safety (an outcome). Therefore, performance measurement should primarily focus on outcomes. Yet, and herein lies the paradox, outcomes are in many instances very hard to count. What is measured gets attention, but many important dimensions are not measurable. 422 ppmr / March 2011 Distrusting Professionals, But Relying on Them A second paradox is the ambiguous attitude of the performance management doctrine toward expertise and professionals. On the one hand, professionals are central to better performance. In fact, the NPM motto \"Let managers manage\" suggests strong confidence in the professionalism of managers (Kettl, 1997). Similarly, managers are expected to be entrepreneurs and leaders who bring out the best in the personnel under their supervisionnot in command-and-control style, but by empowerment (Osborne & Gaebler, 1993). Yet performance management systems seem to express a certain distrust of professionals. Davies and Lampel (1998), assessing performance management in the British National Health Service, argue that managers primarily used performance information in a tactical way, in order to intervene in the doctor-patient relationship. Hence, they developed a vast number of indicators to counterbalance the doctor's professional knowledge. Radin (2006) provides the example of the British Research Assessment Exercise (RAE), which audited the research quality of universities based almost solely on such performance indicators as the number and type of publications. These performance indicators replace professional judgment on the quality of research. Similarly, Radin points to the No Child Left Behind initiative in public schools in the United States. She asserts that the most vigorous critique was based on standardized tests that allegedly did not leave enough room for teacher discretion. There is a clear paradox. On the one hand, trust in professionalism is vital in an increasingly complex society. On the other, we are reluctant to grant this trust and hence fall back on control and audit. In auditor circles, the adage \"In God we trust, the rest we audit\" is well appreciated. Nevertheless, according to Power (1999), audits are to a large extent ceremonialhe speaks of rituals of verification. Many audits create an illusion of control. Similar arguments can be made for other performance measurement regimes in both the public and private sector (Van de Walle & Roberts, 2008). The failure of rating agencies that was exposed by the financial crisis is a case in point for the latter. Paralysis by Analysis Decision-makers have to process a huge body of information found in budgets, audits, impact analyses, evaluation studies, memoranda from interest groups, laws and jurisdictions, personal communications, and other sources. Almost superhuman analytical capacity is required to process all of this. Performance information comes on top of the pile, increasing the risk of an information overload. Although performance management was devised to improve decisions, it may well lead to paralysis. Thus one should not be surprised that practitioners consider selectivity in measurement to be one of the key challenges for implementing performance management (Mayne, 2007). Van Dooren / Better Performance Management 423 If Everyone Is Accountable, then No One Is There is increasing awareness that public organizations cannot be effective on their own. A considerable literature has developed on collaboration, partnerships, and networks.2 As a result of collaboration, the responsibilities for performance are shared. Hence, when many organizations participate, it becomes more difficult to hold a single organization accountable for results. And if many are accountable, there is a risk that no one is taking responsibility for failure and everybody for success. Should we then stick to traditional accountability schemes with one principal and one agent? Probably not. The willingness to collaborate can erode when single-accountability schemes are maintained. Denhardt & Aristigueta (2008) demonstrated that the typical approaches to performance management affect partnerships and collaborations. Performance-based accountability systems tend to undermine collaborative efforts unless they are accompanied by other strategies for providing an impetus for alignment and collaboration across agencies. Single-Loop: Better implementation The usual response to the paradoxes of performance management is to improve implementation. This is single-loop learning. The basic premises of performance management are not put into question. The idea is that if only we try harder, performance management will improve. The discussion that follows considers four strategies that are often proposed for better implementation of performance management. Improve the Quality of Performance Information A first strategy is to improve the quality of the information. The attention paid by organizations to quality matters is not always evident. Research suggests that often only modest attention is paid to quality-assurance practices in the area of performance measurement (Mayne, 2007). Nonetheless, Hatry (2008) argues that an investment in the many dimensions of quality can ratchet up the use of performance information. He mentions validity, accuracy, timeliness, analysis, and presentation as some important quality dimensions. Clear and timely presentation of performance information may remedy the information overload of decisionmakers. Professionals may trust performance information more when validity and accuracy increase. Better analysis may help to focus performance management on what matters and to single out accountability issues. Quality of performance information alone, however, does not guarantee the use of performance information. As argued above, performance information competes with other sources of information for decision-makers' attention. Moreover, 424 ppmr / March 2011 research on gaming has revealed that sometimes information is selected because it fits a predetermined agenda and not because it is intrinsically good (Bevan & Hood, 2006; Perrin, 1998). This should not come as a surprise. Studies on knowledge utilization have made the same point for the use of research evidence in policy-making (Innes, 1990; Weiss, 1977). It should be clear in this context that quality improvements will not help to overcome such issues as disputes about accountability or resistance from professionals. Leadership A survey by the Organization for Economic Cooperation and Development (OECD) found that strong leadership by politicians and managers is a central element in the success of performance management (Curristine, 2005). Someone has to put his or her shoulder behind a performance management effort and develop a measurement strategy. Preferably, this will be a person who carries some weight. Good leadership may be a response to the paradoxes, because good leaders take the uncountable into account, quickly identify core bits of information, motivate professionals, and hold people to account in a fair way. However, as a concept, leadership is poorly defined, and hence interpretation of the OECD findings is complicated. In particular, in survey research such as the OECD's, there is the risk that respondents use leadership as a quick fix when they cannot point to more precise factors. Hence, better implementation through better leadership is not a very actionable recommendation. The issue of leadership certainly raises a host of secondary questions (Van Wart, 2003): Who should the leader be? What traits are important for performance leadership? Where does leadership in performance come from, and how can it be sustained? What should leaders do in which circumstances? It does not suffice to state that better performance management depends on better leadership. It is also necessary to make clear what leadership is and what leaders need to do. Ownership \"Ownership\" is another somewhat magical word in management discourse. Implementation failures are regularly said to be caused by a lack of it. Mayne (2007), for instance, notes that a system built on filling in performance information forms for others, forms of no apparent use to those down the line, is unlikely to be robust and to survive over time. Better implementation of performance measurement and management requires that those who are affected by the system have to accept and internalize the system (Van Dooren, Bouckaert, & Halligan, 2010). In a thrust to ensure ownership, performance management reforms often fall victim to overcommitment (Pollitt, 2008). Many people need to be convinced before a performance management system can be introducedpoliticians, top Van Dooren / Better Performance Management 425 and middle managers, professionals and front-line workers, to name a few. Hence, an understandable strategy is to create high expectations and play down the costs. Yet, although this strategy may prove successful in the short run, it almost definitely will boomerang in the medium term. Typically, the costs of a performance management system are tangible and become apparent relatively shortly after the introduction of the system (Bouckaert, 1993). Benefits, on the other hand, are intangible and may only appear in the longer term. Disillusionment with performance systems that do not (yet) deliver may undermine confidence, and therefore the failure of the performance management effort may become a self-fulfilling prophecy. Ownership strategies are relatively fragile. It takes much more effort to build acceptance of a performance management system than to lose it. Variations of Integration Integration, coordination, formalization, consistency, coherence, routinebuilding, and alignment are some of the most common keywords for those who want to fix performance management without questioning its blueprint. See, for instance, the report of the management consulting firm Accenture (2006) as an example. Bouckaert and Halligan (2008) are on the same line of thinking when they stress the importance of integration in the shift from a model of performance administration to genuine performance management and even performance governance. Although the importance of integration and coordination is undeniable, their limitations must be acknowledged (Laegreid, Roness, & Rubecksen, 2008). Complexity and change regularly tear carefully coordinated systems apart. The desire to coordinate all efforts in advance may lead to delay and even deadlock. In some instances, it may make more sense to remedy the consequences of ill-coordinated performance efforts than to embark on excessively ambitious coordination efforts (Johnsen, 1999). Double-Loop: New Ways of Doing Performance Management The four suggested strategies for better implementation may substantially mitigate the paradoxes identified in the first section, but, as noted, they have some limitations. Recent research on performance management suggests new ways of doing performance management. Some of these new ways are emerging not only in theory but also in practice, as illustrated by the contemporary use of CitiStat (Abramson & Behn, 2006). The purpose of this section is to dig deeper into these proposals. It argues, first, that the main challenge to performance management lies in making it ambiguity-proof. In the next paragraphs, more concrete implications of this argument are proposed. 426 ppmr / March 2011 Making Performance Management Ambiguity-Proof Contemporary public administrationlike society at largeis often described as being ambiguous (Weick, 1995). Ambiguity is characterized by the absence of meaning, rather than by uncertainty and the absence of information (March & Olsen, 1976). Sense-making in society has become multilayered and more fragmented. Arguably, when we all have our own truths, policy-making and management become more complex. Noordegraaf & Abma (2003) add that current performance management, which they label as management by measurement, only fits the rare unambiguous contexts. Defined as such, not many unambiguous situations will be found. Since ambiguity is everywhere, the prospects for performance management are, in this view, rather limited. An alternative to giving up on performance management is to rethink it in order to make it \"ambiguity-proof.\" This can only be done by taking complexity and ambiguity as a given, and changing the practice of performance management on this foundation. Performance information should enable interpretative processes of sense-making. Along these lines, Radin (2006) argues that many problems with performance measurement and management can be attributed to faulty points of departure. She argues that intelligence is not based on universal principles and literal meanings. Rather, multiple sources, situational knowledge, and literal and symbolic meanings are important in sense-making. It should also not be assumed that values, politics, and conflict are not at play in performance management. They clearly are. Finally, the assumption of linear cause-and-effect relations, as well as clear goals and planned change, does not survive the reality check. On the contrary, Radin paints a picture of complexity, interdependence, and unplanned change. Implications What would an ambiguity-proof performance management system look like? Three features are discussed belowthe need for performance management to be agile, decentralized, and political. The Need for Agility Kravchuk and Schack (1996) refer to Ashby, a cybernetics scholar, who posited that only complexity can absorb complexity. Rigid information systems are not able to understand rising complexity in the environment. In the most extreme cases, chaos will appear to reign due to the ever-increasing gap between practical experience and the knowledge base as provided by the information system. Information (what we believe we know) and practice (what we experience) risk becoming separate worldsone orderly, where objectives are set and performance targets are reached, and one chaotic, where people are mainly trying to muddle through the day. The main implication would be that performance indicators (PIs) should be Van Dooren / Better Performance Management 427 used for learning, and less so for accountability (Delancer Julnes, 2008). PI-based accountability requires stability for the period for which targets are set. Yet, as research in Australia and New Zealand has demonstrated, not many fields remain stable for three to six years (Carlin, 2006). As a result, accountability erodes accordingly (Gregory & Lonti, 2008). In addition to stability, accountability requires relatively univocal PIs that do not allow for much interpretation. PIs have to be an accurate and undisputed representation of \"real\" performance. Unlike accountability, learning does not require the same stability and robustness. On the contrary, performance measurement is part of a permanent dialogue in order to make sense of complexity (Moynihan, 2008). Hence, indicators can and should be adjusted in response to contextual changes and new insights. This proposal does not suggest that managers and professionals should not be accountable for performance. It only suggests that accountability through performance indicators (PIs) is probably not compatible with the need to have agile measurement systems. Accountability, however, does not solely depend on PIs. Performance assessment can be qualitative, quantitative, or both. In a complex environment, it might make more sense to hold managers accountable for the way they facilitate learning from performance indicators, rather than for the performance indicators themselves. \"Guerrilla Tactics\" and Decentralization Organizations typically have an undercurrent of repeated decisions they have to make. To these recurrent cycles, a constant stream of unique one-off decisions is added. In recent decades, the relative importance of the stable, recurrent processes has decreased. Kettl (2002) argues that the traditional U.S. public administration boundaries of mission, resources, capacity, responsibility, and accountability must be managed in an increasingly complex and politicized context, necessitating additional negotiation and collaboration between systems and agencies. These complex parallel processes are, in a unique way, shaped by situational requirements of time and place (Pollitt, 2008). Recurrent financial, human resources management, and contract cycles have been the main vehicle for incorporating performance information in decisionmaking (Bouckaert & Halligan, 2008). Without doubt, these cycles will remain the foundation of performance management in the future. Yet top-down performance management on a yearly basis (as in the budget cycle) or a monthly basis (as in many balanced-scorecard systems) will need to be supplemented by flexible efforts to provide performance information on demand. Since complex, unique processes will gain importance, the timing (when the performance information is used) and the locus (where it is used) will be challenged. With regard to timing, performance management in complex contexts may need to resort to \"guerrilla tactics.\"' In complex policy and management processes, the 428 ppmr / March 2011 demand for performance information can arise relatively unexpectedly. At the same time, demand can fade away as quickly as it came about. In such a context, expert personnel are needed to quickly infuse complex processes with performance information. As performance information brokers, they need to be able to both capture the need for and understand the availability of performance information. Like guerrillas, they use hit-and-run tactics to improve the quality of the often unpredictable processes of making sense of performance. Regarding the locus, performance management in complex contexts may benefit from stronger decentralization. Rather than devising top-down systems, performance management needs to be in the hands of the middle managers and front-line supervisors who best understand the situational requirements. This strategy might bring about the capacity of public managers that Behn (2004) calls performance leadership. He compared the performance leadership model to a focus on performance systems and structures. He writes that \"rather than to develop public managers with the leadership capacity to improve the performance of their agencies, we have sought to create performance systems that will impose such improvements\" (p. 3). The same reasoning can be applied within agencies. Middle managers and front-line supervisors need the leadership capacity to improve performance. This includes drawing lessons from performance indicators. For performance budgeting, the decentralized, action-based approach would suggest that performance information should be introduced into budget negotiations at the microlevel rather than systematically reported in the budget document that is voted in the legislature. Since the budget document is mainly an after-the-fact codification of political negotiations that have already taken place, performancebased budget formats run the risk of becoming a bureaucratic exercise (Sterck, 2007). There is some evidence supporting the effectiveness of this decentralized and action-based approach. Bourdeaux (2008), for instance, suggests that performance information mainly enters legislative discourse through policy networks maintained by key committee staff, rather than through executive budget reports and requests. Some confirmation is also found in an OECD survey on performance budgeting showing that countries use performance information to inform, but not to determine, budget allocations (Curristine, 2005). Furthermore, as Curristine (2005) argues, much of the \"linking\" of performance and financial information has taken place simply to provide them in the same report. A study by Melkers and Willoughby (2001) at the local and county levels found that the strongest usefulness of measures within a budget cycle occurs during budget development, with their importance lessening as the budget process proceeds (Melkers & Willoughby, 2005). Political Awareness Some time ago, Innes (1990) observed that the only way to keep data-gathering out of politics is to collect irrelevant data. Good performance information should Van Dooren / Better Performance Management 429 strengthen the evidence base for solving the political problems of who gets what, when, and howthe classic and broad definition of politics by Lasswell (1936). Issues of who gets what, when, and how are at play at all levelsin governmentwide policy-making, in policy sectors and networks, in organizational management, and in micro-management. Performance indicators can elevate the quality of political discussion at all these levels. Therefore, it is not suggested that political institutions (the executive, legislature, parties, etc.) have to interfere with all performance issues at all levels. Rather, the object is to stress the importance of recognizing the political nature of performance management. The first implication is that performance management should involve more rather than fewer actors. In complex settings, as argued above, performance management is mainly about sense-making. Hence, performance management can be expected to have the greatest impact when different perspectives are drawn into the dialogue. Obviously, the interests around the table have to be relevant and the number of participants needs to be workable. A second implication is that performance management should deal with controversy rather than suppress it. Performance information should not be an authoritative argument used to end conflicting views on who gets what, when, and how. Rather, it should underpin a careful argumentation of causes, consequences, and