Abstract
The ambition of this article is to gain a better understanding of the endogenous dynamics of performance management regimes. Based on a review of the literature, we develop a framework that enables us to grasp dimensions and mechanisms of escalation. Hereafter, we demonstrate the use of our framework through two cases. We argue that a range of dimensions of a performance regime may evolve over time, that the dynamics might be repressed by contextual factors and politics, that the dynamics may be reversed leading to de-escalation, and that different dimensions related to the regime do not all necessarily escalate simultaneously.
Introduction
As early as 1989, Pollitt (1989) observed that the functions of a set of performance indicators (PIs) change over time. Since that time, various researchers (e.g., de Bruijn, 2007; Pollitt, 2013; Pollitt, Harrison, Dowswell, Jerak-Zuiderent, & Bal, 2010; Woelert, 2015) have observed the dynamics involved in PI systems, and a terminology to describe these dynamics is gradually being developed. One of these dynamics is that PI systems tend to escalate over time. de Bruijn (2007), for instance, argues that PI systems tend to inflate over time through the “law of mushrooming.” Similarly, Pollitt (2013, p. 353) and Pollitt et al. (2010) point to the notion of a “logic of escalation.” The basic notion is that once a PI system is in place, there is an endogenous dynamic that results in the multiplication of PIs, an increasing technical elaboration of composite indices, a parallel growth in a specialist technocratic community of “performance experts,” and a tighter coupling of PIs to targets and targets to penalties and incentives (Pollitt, 2013, p. 353; see also Kristiansen, 2017, p. 49).
Previous research has thus provided interesting insights into the dynamics of performance management regimes and how they might evolve over time. With this article, we want to add to the existing literature and terminology by providing a more systematic description and understanding of which dimensions of a performance management regime are likely to escalate (or de-escalate) together with a deeper understanding of the modes and mechanisms leading to escalation (or de-escalation). Based on a review of the literature on performance management, the dynamics of performance management systems, and a historical institutional perspective on endogenous changes, we develop a framework that will enable us to grasp the dimensions and mechanisms of escalation.
By means of two case studies from the Danish public sector, we will show that that the tendency of PI systems to “escalate” is contingent upon a wide range of institutional factors. Denmark has a very large public sector. It is a unitary, highly decentralized state with a multi-party system and proportional elections usually resulting in coalition governments (Hansen, 2011). The political-administrative culture is characterized by an egalitarian, consensus-oriented approach to public administration. Denmark has a mode of ministerial responsibility, meaning that a minister can be held politically accountable by the Folketing, the national parliament, for any decision made by the administration under their control (Kristiansen, 2016). The institutional context for public services is characterized by (a) a very high degree of decentralization of public expenditure, (b) a broad portfolio of tasks delegated to local governments, particularly the municipalities, (c) financial discretion at the local level, and (d) corporatist negotiations between central and local government representatives about local taxation and expenditure levels. Public administration in Denmark is often characterized as informal, based on dialogue and trust, and not being thoroughly, systematically, and rationally organized from top to bottom (Hansen & Beck Jørgensen, 2009). Finally, Denmark is characterized by low scores on the Masculinity Index (indicating a feminine society) and having a low power distance, meaning a very egalitarian mind-set (Hofstede, 2016). On this basis, Denmark provides an interesting setting for observing escalation mechanisms, as the administrative culture and government structure differs starkly from, for instance, the United Kingdom, where most of the previous studies of escalation mechanisms have been situated.
Within Denmark, we have selected two different illustrative cases: (a) management by objectives and results (MBOR) in central government and (b) comparative PIs for municipal productivity. Both cases are outside the health care and university sectors, which have previously been the primary research settings when analyzing escalation processes (Pollitt et al., 2010; Woelert, 2015). Moreover, they are not sector specific and cover large parts of the public sector. These different cases are selected for how they illustrate how escalation and de-escalation processes take place in different settings, how contextual factors related to the performance management regimes may enable or constrain endogenous escalation processes, and how endogenous dynamics of performance management regime may be reversed resulting in de-escalation.
Our illustrative cases enable us to show how a range of different dimensions related to a performance regime might evolve and escalate over time. We sketch out a systematic framework to capture these dimensions and discuss how they are interrelated. Moreover, we argue that the endogenous dynamics of a performance regime might be repressed by contextual factors and politics. Furthermore, we argue that the endogenous dynamics of performance management regime sometimes may be reversed and go in the direction of de-escalation. Finally, we argue that different dimensions related to the regime do not all necessarily escalate at the same time; some might escalate while others might remain stable or de-escalate.
The article is structured in four sections. After this introduction, a theoretical framework to grasp dimensions and dynamics of escalation processes in a performance management regime is developed and presented in the “Theoretical Framework” section. In the “Illustration of Escalation Processes: Two Cases From the Danish Public Sector” section, the use of the theoretical framework is illustrated using two cases from the Danish public sector. In the “Conclusion” section, the framework is revisited and conclusions are drawn.
Theoretical Framework
This section develops a framework to grasp various dimensions of escalation and endogenous dynamics of a performance management regime. First, we define what we mean when talking about a performance management regime: What are the core elements in such a regime and how are they interrelated? Second, endogenous dynamics of performance regimes and mechanisms leading to escalation and/or de-escalation over time are discussed on the basis of a historical institutional perspective. Third, we deduce a range of dimensions on which a performance management regime potentially might escalate (or de-escalate) over time, and we discuss the mechanisms leading to escalation and de-escalation.
Performance Measurement, Performance Management, and Performance Regimes
The literature offers an array of definitions of the concepts of performance measurement and performance management. Our point of departure is Van Dooren, Bouckaert, and Halligan (2010, p. 25ff), who define performance measurement as the “bundle of activities of quantifying performance.” These activities include (a) defining a measurement object, (b) selecting PIs, (c) data collection, (d) data analysis, and (e) reporting; the result of these five activities is performance information.
The active use of performance information is highlighted in Van Dooren et al.’s (2010) definition of performance management: “A type of management that incorporates and uses performance information for decision making” (p. 30). A range of different documents—including performance contracts, balanced score cards, standardized quality models, and annual reports—are relevant in incorporating performance information into use, and performance information may be used in various ways (Behn, 2003; Hatry, 2008; Van Dooren, 2006). Vedung (2009) distinguishes between whether information is used for formative or summative purposes, and Van Dooren et al. (2010, p. 101) distinguish between whether the information is used by internal or external actors and whether or not sanctions are coupled to the use of performance information. Finally, the use of performance information may affect actors’ behavior in various ways, both functionally and dysfunctionally (Bevan & Hood, 2006; Smith, 1995).
Whereas Van Dooren et al. (2010) present the core elements of a performance management system, Talbot (2010, p. 81) applies a somewhat broader perspective. He introduces the idea of a performance regime referring to a combination of (a) the institutional context of performance steering and (b) the nature of actual performance interventions.
The institutional context of performance steering refers to who has formal rights and other instruments with which to “steer” public organizations and programs. The institutional context may be simple and very much a vertical chain of principal–agent-type relationships or it may be much more polycentric, with multiple principals, including the delegation of performance measurement to a third party inspectorate or audit body. According to Talbot (2010, p. 92), the sets of stakeholders (institutional actors) that may be part of the institutional contexts are (a) central ministries; (b) line ministries; (c) legislature(s); (d) audit, inspection, and regulatory agencies; (e) judicial and quasi-judicial bodies; (f) professional associations; (g) users and formally constituted user bodies; (h) other public agencies as partner organizations; and (i) agencies themselves. Talbot (2010) primarily focuses on potential principals, whereas agents (or professionals) are given less attention. They may be important actors, however, as they might occasionally with success question the expediency of the performance management regime.
The nature of actual performance intervention refers to the actions these stakeholders (actors) take, individually and collectively, to influence the performance of public organizations or programs—through performance contracts, imposed targets, comparative league tables, and other levers (Talbot, 2010, p. 81).
If we combine the definitions outlined above, a performance management regime may be understood as a combination of (a) measurement of performance using PIs; (b) incorporation of PIs and information in contracts, annual reports, rankings, benchmarking figures, and so forth, and use of performance information; and (c) the institutional actors (stakeholders) which are involved in administering such interventions.
The primary dimensions of a performance management regime and how they are interrelated are illustrated in Figure 1. Figure 1 shows how the measurement process might affect how the information is incorporated and used. It might, however, also work the other way around so that the incorporation and use of information affect the measurement process. Moreover, the measurement process, the design of interventions, and/or the use of information may ultimately produce dysfunctional behavior, such as gaming, cheating, and measure fixation (Bevan & Hood, 2006; Smith, 1995). The design and use possibly also crowd out the professionals’ dedication (de Bruijn, 2007) or motivation (Frey, 1997). Thus, PIs showing an imperfect picture of the professional’s performance, or indicators that are easily manipulated might be followed by unintended behavior. Similarly, the manner in which information is incorporated and used possibly also leads to dysfunctional behavior, as sanctions and high impacts related to the judgment of performance information potentially crowd out motivation or trigger gaming and cheating. Such experiences may be important inputs into the design of next year’s system and may be followed by adjustments or changes to the PIs, how information is used, and so on. Finally, the performance management regime and the behavior it stimulates will be interrelated with the stakeholders (in the institutional context), as they will continuously interpret and perhaps adjust the design of the PIs, interventions, or uses.
Dimensions of a performance management regime.
Besides these dimensions and their interrelations, the performance management regime is obviously also situated in an environment in which global financial crises, new management ideas, and so forth occasionally also produce pressure for change. Although performance management regimes are clearly embedded in a larger environment and changes may be set in motion by environmental pressures, the main emphasis here is on endogenous changes in performance management regimes.
Theoretical Perspectives on the Dynamics of Performance Management Regimes
By endogenous changes we mean technical and political factors endogenous to a performance management regime, which once these systems are in place, appear to exert a logic of their own (Pollitt, 2013, p. 353; Pollitt et al., 2010, p. 14). More specific, we define endogenous changes as changes related to the continuous interactions between the different elements of a performance management regime outlined above: (a) measurement of performance using PIs, (b) incorporation and use of performance information, (c) the behavior the performance management system stimulates, and (d) the stakeholders related to the performance management system.
Earlier, most scholars within the literature on institutional change argued that institutional change was a function of factors exogenous to the institutions such as economic or political crises (Mahoney & Thelen, 2010, p. 2; Streeck & Thelen, 2005, p. 7). A sharp line between the logics and the analysis of institutional reproduction and change was suggested, as there was a tendency to see change mostly in terms of dynamics unleashed by some exogenous shift or shock, ignoring the possibility of endogenously generated institutional change that is more than just adaptive (Streeck & Thelen, 2005, p. 7). Thus, the argument in this article is that changes of a performance management regime rather than emanating on the outside are often endogenous. It is, however, difficult to draw a strict line between endogenous and exogenous factors, and the analytical distinction may be difficult to sustain in empirical research, as they often may work together. In this article, we distinguish between endogenous factors related to a performance management regime meaning the performance management system itself and the stakeholders related to it, and the exogenous factors surrounding the performance management regime such as economic cycles and the development of the broader political system and organization of the state.
Streeck and Thelen (2005, p. 8) argue that change is often gradual, endogenous, and sometimes produced by the very behavior an institution itself generates, as unanticipated consequences of the design of institutions require continuous adjustments and revisions (Streeck & Thelen, 2005, p. 16). Moreover, they argue that institutions are defined by continuous interaction between rule makers and rule takers during which new interpretations of institutions will be discovered, invented, suggested, rejected, or—for the time being—adopted. Thus, institutions will continuously be created and recreated by a great number of actors with divergent interests, varying normative commitments, different powers, and limited cognition. In this process, actors will try to achieve an advantage by interpreting or redirecting institutions in pursuit of their goals or by subverting or circumventing rules that clash with their interests (Streeck & Thelen, 2005, pp. 16, 19).
Streeck and Thelen (2005) suggest a number of processes by which gradual changes can occur: (a) displacement, (b) layering, (c) drift, (d) conversion, and (e) exhaustion. Displacement refers to the removal of existing rules and the introduction of new ones. Layering refers to the introduction of new rules on top of or alongside existing ones, whereas drift refers to the shifting impact of existing rules due to shifts in the environment. Conversion refers to the redeployment of old institutions into new purposes through redirection and reinterpretation; and finally, exhaustion refers to the gradual breakdown of institutions over time.
Another perspective focusing on endogenous changes to institutions may be derived on the basis of the lengthy history of writers who see events as cycling rather than sequencing (e.g., cycling between decentralization and centralization, managerial autonomy and tightening accountability, coordination and specialization) (Aucoin, 1990). According to Pollitt (2008, p. 53) the underlying mechanism might be that in a highly uncertain world where there are no firm rules as to the best solution, organizational designers tend to opt for one set of forms until their particular disadvantages become apparent, whereupon they begin looking for opportunities to move back toward a different form that reduces these disadvantages. The cost hereof is the introduction of other drawbacks. Over time, reformers therefore wobble between alternatives, each carrying their own advantages and disadvantages. Pollitt (2008) argues that cycling might primarily be related to specific conditions, such as the absence of any stable, generally accepted design principles and a limited number of apparent options (each with significant advantages and disadvantages). Cycles can also occur within an overall path and be combined with a long-term trend (Pollitt, 2008, p. 58; see also Kristiansen, 2017, p. 53). The theoretical implication of cycling is that escalation might occur in one period of time and de-escalation in another. Thus, on the basis of the literature on cycling events, we will observe whether endogenous changes of performance management regimes may lead to de-escalation, as disadvantages of a more comprehensive regime appears to the reformers (or rule makers) resulting in cycles between escalation and de-escalation.
Escalation of Performance Management Regimes: Dimensions and Dynamics
In the above, a performance management regime was presented as a combination of (a) performance measurement, (b) incorporation and use of performance information, and (c) stakeholders involved. On the basis of these elements, we will now develop a range of dimensions of a performance management regime in which dynamics of (potential) escalation are likely to occur.
Performance measurement: Escalation of scope and dimensions of measurement
Three dimensions of escalation related to performance measurement are outlined in the following.
Number of PIs and targets
The number of PIs is probably the most prominent and best described escalation dimension in the literature. de Bruijn (2007) has described this mechanism as the “law of mushrooming,” which has multiple causes (p. 40). First, professionals learn to exploit the freedom for strategic behavior offered by the system. The natural managerial response to this is refinement, such as adding new layers of indicators or targets. Second, as a measurement system can only focus on a limited number of aspects of professional performance, the professionals may feel unfairly treated because they score poorly in the system, even though the actual performance is considerably better. Consequently, the professionals have a strong incentive to inflate the system with further layers of indicators or targets to create a more realistic view of their performance. Conversely, the cost related to running a performance management regime can increase or professionals may react against the increasing red tape and render simplification necessary by adjusting the system (a de-escalation process).
Escalation and the sophistication of the dimensions of performance covered by indicators
The dimensions of performance covered by PIs possibly also change over time as a result of sophistication. An organization might choose to formulate output indicators at a given point in time. Here, it might be argued that these indicators do not provide an adequate image of the organization’s performance and that the measurement of outputs is likely leading to measure fixation, suboptimization, myopia, and other kinds of unintended consequences of publishing performance information (Smith, 1995) (especially if measurement is coupled to sanctions). One way of countering this problem is to choose more sophisticated outcome measures that minimize the gap between the PIs and the performance itself (in terms of organizational or policy objectives) or to measure multiple dimensions of performance (e.g., a mix of outputs and outcomes). Some of the disadvantages of measuring outcomes may occur at a later point in time, as attributing outcomes to the specific activities provided by the organization can be difficult (de Bruijn, 2007). This possibly leads to de-escalation and a return to measuring less sophisticated dimensions of performance.
The sophistication of indicators and aggregation of indicators into composite measures
Related to the escalation and sophistication of dimensions of performance covered by indicators, the PIs might also become further sophisticated by using statistical controls, more advanced methods, the aggregation of indicators into composite measures, and so forth. When comparing primary school grades, for example, it is possible to control for socioeconomic circumstances, or simple benchmarks can be replaced with more sophisticated Data Envelopment Analysis (DEA), and so on. Such development processes may be related to a parallel growth in a specialist technocratic community of “performance experts” (Pollitt, 2013, p. 353).
Performance management: Escalation of interventions and use of performance information
Escalation related to the performance management phase focuses on two dimensions: interventions and using performance information.
Escalation of interventions and incorporation documents
Interventions such as performance contracts, comparative league tables, and other levers may escalate by adding ever more steering documents to the existing ones to correct the shortcomings of prior instruments and further sophisticate the regime. Thus, escalation might occur through the layering (Streeck & Thelen, 2005) of new elements on top of the existing elements. Some of these elements may later be displaced or deliberately neglected through drift or gradual exhaustion (Streeck & Thelen, 2005), resulting in a de-escalation process over time. Escalation possibly also occurs through standardization or the further sophistication of the interventions or by escalating or converting the purposes (Streeck & Thelen, 2005) of the interventions over time.
Escalation of the use of performance information
Over time, the purposes of a performance management regime and the use of performance information may escalate. According to Pollitt et al. (2010: 20), the use of performance information tends to shift from formative to summative. The use of performance information possibly also changes from internal to external use (from learning and steering to accountability) through the broadening of the external audiences for performance information. Such processes may result from layering, where new types of use are added to the existing and/or by conversion, meaning that the use of performance information is redeployed into new purposes through redirection and reinterpretation (Streeck & Thelen, 2005).
The combination of summative performance assessment with incentives/sanctions naturally raises concerns about data validity, which is why simple single indicators may be subject to standardization or sophistication to achieve more valid comparisons; possibly also contributing to the inflation of PIs. Moreover, as performance information becomes more summative and targets are increasingly linked to sanctions (i.e., incentives, penalties), various forms of “gaming” and cheating are likely to arise in response (Pollitt, 2013). These various forms of gaming may also trigger an escalation of PIs. Thus, escalation in this dimension of a performance management regime may be followed by escalations in other dimensions of the regime.
Institutional context: Escalation of stakeholders
Three dimensions of escalation related to the stakeholders of a performance management regime are outlined in this section.
Institutional autonomization of performance evaluations
Autonomization includes everything done to make the PI systems more “systematic,” “independent,” and “expert-based,” including the “purification” of processes and quality-control mechanisms. It includes institutional devices to support these processes, such as the delegation of performance measurement to a third party inspectorate or audit body. Through institutional autonomization, third party inspectorates assume power, influence, and resources through their role and position as (independent) evaluators. To maintain their position, power, and legitimacy, third party evaluators must continuously maintain, legitimize, escalate, and sophisticate the performance management regime. The underlying institutional interests of the inspectorate or audit body may be an escalating factor in itself.
Change of ownership: From private/frontline ownership to public ownership
Performance management regimes may have different governance structures, including voluntary performance networking, market-based sales of performance management tools, or the hierarchical imposition of performance management by public authorities (Kuhlmann & Jäkel, 2013). If a performance management regime changes from private ownership to public ownership or a performance management regime developed by frontline staff becomes a public-owned regime, the primary purpose(s) of and use of the regime may also change through conversion. If ownership changes to become public, we may expect performance information to become more integrated in government steering and to be used for summative rather than formative purposes. If PIs are published as official government documents, this process is likely to raise concerns about data validity, which may in turn trigger further changes in the scope and dimensions of measurement.
Escalation of stakeholders and roles in relation to performance management regime
Escalation of stakeholders refers to the increase in the number of stakeholders (or principals) related to a performance management regime. According to Pollitt et al. (2010), there is a tendency toward the establishment of a performance “industry” of official regulators, academic institutions, and others, including groups of consultants and analysts who use the regime partly to pursue their own ends (p. 24). The uses of performance information will therefore likely expand, as different stakeholders demand different kinds of information, which will inflate the measurement system. Key stakeholders may argue at a later stage that the performance management regime has become too diffuse and come to include too many stakeholders and too many different purposes, which may reduce the number of stakeholders (de-escalation).
Summing up, we have now identified eight dimensions (and a few subdimensions) in which escalation may occur and potentially also fuel escalation processes in other dimensions of a performance management regime. The eight dimensions are presented in Figure 2:
Eight dimensions of escalation.
Illustration of Escalation Processes: Two Cases From the Danish Public Sector
After having sketched out our theoretical framework, we now illustrate how the framework can be applied using two illustrative cases from the Danish public sector.
Case I: MBOR in Danish Central Government
Our first case focuses on MBOR, 1 which was adopted in Danish central government in the 1980s and early 1990s (Kristiansen, 2015). In 1991, the Danish Ministry of Finance initiated an experiment with performance contracts between parent ministries and their agencies. In 1992, the first seven contracts were established in a pilot project. Performance contracting was not mandatory, so the Ministry of Finance had to persuade departments and agencies to join. Due to its optional status, performance contracting developed slowly. It has since spread, however, and is now nearly universally adopted in central government.
Over time, various forms of intervention have been added to the original performance management regime. In 1995, performance contracts for chief executives in the agencies (related to performance-related pay) were introduced as a pilot project, becoming permanent in 1997. In 1997, performance-related pay was also introduced for employees.
In 1996, a committee published a report calling for increased central control on the grounds that the decentralization of budget responsibility had gone too far. The report recommended “enterprise accounts” (later, annual reports) as a useful tool in the performance-controlling process, as they could lay the basis for the more thorough evaluation of the performance of central government agencies for the parent ministries, the Ministry of Finance, or the standing committees in the Parliament. It was reasoned that the National Audit Office could also use this information in their performance audit. In 1997, “enterprise accounts” became mandatory.
In 2003, a new government report recommended adding a task hierarchy. Agencies were also supposed to publish “clear goals” for their users, and at the same time effort was made to integrate financial resources with results. Accrual accounting and budgeting were therefore introduced.
In the mid-2000s, MBOR was exposed to extensive criticism from rule takers (see, for example, Gjørup et al., 2007), and a working group in the Ministry of Finance started collecting experiences with MBOR toward the end of 2007. Based on this work, some of the previous demands were displaced according to a deregulation agenda (e.g., that agencies should publish “clear goals” for their users). Moreover, contrary to previous Ministry of Finance reports recommending that targets cover all agency activities, it was now recommended that performance contracts should focus on a few, strategically important goals.
Over a 10-year period, the number of performance targets per contract had almost doubled from an average of around 20 in 1995 to 37 in 2005 (Binderkrantz & Christensen, 2009, p. 66). This trend now seems to have been reversed, as the number of performance targets/contract has been cut in half from an average of 37.3 in 2002 to 18.6 in 2014. In the same period, the dimensions of performance have also changed and become more sophisticated, as the share of outcomes increased from 2006 while the share of activities decreased (Kristiansen, 2017).
In 2014, the Ministry of Finance launched a new report arguing that the existing regime had served too many purposes, resulting in excessively comprehensive target sets. Consequently, the Ministry of Finance recommended (a) a focused and simplified approach to MBOR, (b) one purpose for the regime, and (c) a limited number (5-10) of clear targets related to core agency tasks.
Observations
The case shows how the MBOR dynamics result in the escalation and de-escalation over time on many dimensions of the performance management regime.
Escalation of performance measurement
First, the number of targets/contract in the regime escalated, doubling from 1995 to 2005, after which it de-escalated and halved. Second, the dimensions of performance covered by the targets in the regime became increasingly sophisticated over time, as the share of outcome targets increased from 2006, whereas the share of activity-oriented targets decreased in the same period.
Escalation of performance management (interventions and uses)
The regime also gradually became increasingly sophisticated, as further layers of intervention were added to the original performance management regime. Adding new instruments gradually changed MBOR in this period. MBOR escalated, and the regime became more sophisticated, detailed, and comprehensive. In the 2000s, MBOR received criticism from observers and rule takers for having become excessively comprehensive. MBOR was adjusted in response to this criticism, as some of the demands were displaced and exhausted. A simpler regime with fewer interventions has therefore been favored in recent years.
The use of performance information also escalated, as it changed from primarily internal use to also include external use, and from formative to summative use. In the early years, agencies committed themselves to improving efficiency at the same time as managerial discretion increased. It was later argued that more central control (from the parent ministry) was needed, and the National Audit Office also started using the information in their performance audit. Regime-related incentives also increased over time through the introduction of performance-related pay; however, sanctions in the form of punishment are rare.
Escalation of stakeholders
The agencies, their parent ministries, and the Ministry of Finance were initially involved in the regime. The number of stakeholders escalated over time, however, as the National Audit Office started to use performance information in their performance audit of agencies and ministries. Moreover, the State Employer’s Authority became responsible for performance-related pay and performance contracts for chief executives. In the beginning of the 2000s, agency users were articulated as primary stakeholders for the regime, and “clear targets” had to be communicated to them when MBOR was coupled to the government’s modernization agenda “[w]ith the citizens at the helm” (Regeringen, 2002). Employee associations later tried to capture MBOR to secure targets for personnel policies (Statsministeriet, 2007). As the number of stakeholders increased, so did the number of purposes. Recently, however, the Ministry of Finance recommended a simplified approach to MBOR with one primary purpose and the parent ministry as the key stakeholder.
All in all, the case illustrates how escalation and de-escalation processes have developed over time; MBOR cycles between a simple performance management regime, a comprehensive one, and back toward a simpler one. The MBOR regime is always open for interpretation. Over time, different actors with divergent interests tried to integrate their ends and agendas into the regime. Rule makers also receive feedback from rule takers, however, and the regime has been reinterpreted and recreated on the basis of this learning. However, the escalation and de-escalation of MBOR appear within an overall path and within an overall trend toward further institutionalization, sophistication, and standardization.
Case II: Comparative PIs for Municipal Productivity
Our second case concerns the comparative measurement of productivity in Danish municipalities. The development of productivity indicators can be traced back to the local government reform in 1970, which laid the foundation for harmonized budget and accounting systems and comparable expenditure statistics. On this basis, the Ministry of the Interior created the Municipal Key Figures in 1984. Here, budget and accounts data are coupled with other national statistics, providing comparable indicators of expenditure and service outputs in relation to potential users. The system originally contained around 50 PIs, a number that has grown to almost 200 over the years (Sørensen, 2013). The Municipal Key Figures provide comparable figures of expenditures or outputs but do not suggest appropriate benchmarks for each municipality. This was done by another indicator system introduced in 1987 by a private consultancy bureau: the ECO Key Figures. This system uses statistics to identify groups of comparable municipalities in terms of social and financial conditions, thus increasing the relevance and impact of the PIs in local budget discussions. At the turn of the century, most municipalities were using the comparative indicator systems in their internal political discussions.
The figures were also used as a source (among others) of “ammunition” in the annual negotiations between the government’s economic ministries and the local government associations as regards the financing of local services. It was not considered in the common interests of those actors to create a permanent performance regime with a high level of visibility regarding service levels or the productivity of different municipalities. Nonetheless, the existing comparative performance figures stimulated demands from political stakeholders, such as the Liberal Party, that comparative performance analysis should be used more actively to create incentives for productivity improvements, such as the publication of rankings. In 2004, a conservative–liberal government initiated a major reform of local government structures, and several ministers used the occasion to ask for more comparative performance measures of local governments and services. Each initiative was modified in negotiations with local representatives, which led to the establishment of a small evaluation institute and new comparative measurements of local service outputs.
In 2012, this evaluation institute was merged with two research institutes into a more potent Institute for Research and Analysis of Local Governments. The new institute was able to take over the administration of ECO Key Figures and to attract new expertise, and it soon began producing sophisticated comparative performance analyses. It has also published a report that combines data on expenditure and service outputs into a composite measure of the relative productivity of each municipality, based on DEA benchmarking techniques. However, the institute researchers have expressed strong reservations about the validity of the composite indicators, and the PIs have yet to be used for external regulation.
Observations
This case illustrates how the availability of some indicators can create political demands for escalation, but it also illustrates that escalation can be delayed in a system of “corporatist” regulation of local finances.
Escalation of measurement
The introduction of comparative expenditure statistics in the late 1970s triggered the gradual escalation of the number and technical sophistication of comparative indicators. This escalation has not always been driven by public authorities, as they have hesitated to define what makes municipalities comparable, partly out of consideration for local self-government and partly because economic ministries have been satisfied to keep discussions of local service standards at arm’s length from parliament. This has led to a rather fragmented performance regime in which several new PI systems with different administrators have been layered upon the existing supply of comparative expenditure statistics.
The existence of PIs has stimulated an interest and demand from political stakeholders for better and more sophisticated indicators, leading to escalation. Even so, the escalation of measurement has been held back by the difficulty of finding good, uncontested PIs for service quality, and by the limited calibration of the basic expenditure data at the level of individual services. Furthermore, local representatives have been able to convince national policy makers that national PIs should be formulated at a high level of abstraction to avoid the national politicization of local service production (Sørensen, 2013).
Escalation of performance management (interventions and uses)
Over time, there have been initiatives to convert the application of comparative indicators from soft and internal usage in the municipalities toward harder use as instruments for the national regulation of the municipalities. However, these pressures have generally been held back by the common interests of economic ministries and local government associations in not triggering dysfunctional reactions to imperfect indicators. The economic ministries used the recent DEA benchmarking of municipal productivity to argue that there is a large aggregate potential for improving municipal productivity, but it was not used to single out individual low-performing municipalities or to provide incentives for productivity improvements. Escalation is held back by agreement not to encourage the municipalities to maximize (imperfectly measured) productivity at the possible expense of cost-effectiveness.
Escalation of stakeholders
The management of local productivity has many stakeholders, including economic and line ministries, the municipalities and their national association, for not to mention national and local politicians. Over the years, further institutional stakeholders have been added to the regime, including the firm behind the ECO Key Figures. This firm was sufficiently independent of political interests to present its calculations of comparable expenditure figures without disturbing hard-fought consensus on equalization arrangements, but it was not in a position to publish official evaluations of relative productivity. The recent local government reform led to institutional autonomization and public ownership, representing a significant escalation of the stakeholder dimension. This escalation fed back into the dimension of technical sophistication but has yet to lead to significant escalation on the dimension of intervention and use.
Summing up, the case illustrates how the endogenous dynamics of escalation can involve reactions to existing PIs from political stakeholders. However, it also illustrates how escalation dynamics are possibly delayed when there are strong political and economic interests attached to a performance regime and how the 40-year-old budget and accounting system has implications for the current dynamics of escalation. It does not provide sufficiently calibrated registrations of expenditure to make it possible to compare productivity in each individual service area, which limits the usefulness of the PIs with respect to regulating the productivity of individual municipalities and their service departments. In Table 1, the main findings, similarities, and differences across the two cases are summarized.
Escalation and De-Escalation in Two Cases.
Note. PIs = performance indicators.
Conclusion
This article has provided a structured framework presenting a set of concepts, dimensions, and mechanisms to grasp the escalation as well as de-escalation processes of a performance management regime. The framework, which identified eight dimensions of escalation, was applied to two illustrative cases from the Danish public sector.
This framework made it possible to analyze the escalation of PI systems as a differentiated phenomenon. Although all of the dimensions appeared empirically (with the possible exception of increasing sanctions as subdimension to forms of use), both cases were characterized by a unique combination of dimensions unfolding over time.
Although it may be true that there is an escalation in the number of PIs, in dimensions of performance covered, in technical sophistication, and in interventions, documents, and stakeholders, it would be too simple to conclude that there is a general tendency toward escalation and escalation only. The endogenous dynamics of performance management regimes may also sometimes lead to de-escalation, and the institutional context may also function as a “brake” to escalation.
With sensitivity to the multiple dimensions of escalation, we have shown how de-escalation in one dimension possibly compensates for escalation in another. In Case I, for example, the dimensions of performance covered by the targets escalated and grew more sophisticated. As the interventions became more sophisticated and standardized, the number of targets per contract and the number of interventions de-escalated toward the end of the period under observation. Escalation also occurred in Case II, but with considerable delay, and intensified forms of use such as the public ranking of municipalities and economic sanctions were avoided. Key actors, including local government representatives, were influential enough to ensure that the performance regime remained fragmented and was not used in a hard and regulatory manner.
Our main conclusion is twofold: Escalation occurs but at different speeds—and not automatically. It is also a multidimensional phenomenon, meaning that escalation in some dimensions is sometimes accompanied by de-escalation in others.
Our findings invite speculation about the differences between the institutional trajectories found in the two cases. While the mode of change called layering and conversion occurred in both cases, exhaustion and displacement only occurred in Case I. This finding corresponds to the observation of the need to de-escalate the number of targets, interventions, and purposes. We interpret this phenomenon as an instance of learning through feedback from rule taker to rule maker, as the dysfunctions of an ever more comprehensive regime were realized and rule makers were acting on this knowledge. Thus, Case I illustrates how a performance management regime may cycle between a simple one, a comprehensive one, and back toward a simpler one. In one period of time rule makers try to sophisticate the MBOR regime at the same time as different actors tries to integrate their ends and agendas into the regime resulting in escalations on various dimensions through layering and conversion. Later, rule makers receive feedback from rule takers on the dysfunctions of this comprehensive regime, and the regime is reinterpreted and recreated on the basis of this learning resulting in de-escalation on various dimensions through exhaustion and displacement.
In Case II, the escalation of uses and stakeholders was partly based on interactions between the performance regime and its broader political stakeholders (such as the Liberal Party), thus putting somewhat in question the delimitation between endogenous and exogenous dynamics of escalation. However, through most of the 40 years of its existence, the performance regime has developed through gradual interactions between the measurements, their use and their institutional stakeholders. The main institutional stakeholders were able to anticipate unintended behavioral and political effects of an escalation of uses and influential enough to prevent it. The regime has escalated on other dimensions such as the number and sophistication of indicators, but even on these dimensions escalation has remained modest. The historical institutional context continues to hold back the escalation of the regime; both as a result of the continued dominance of a corporatist alliance of actors in central-local relations and as a result of technical limitations of the budget and accounts systems on which the indicators are based.
Theoretically the article contributes to the literature on escalations of performance management regimes by providing a systematic overview of dimension of a regime that may escalate. It also provides a framework that enables us to analyze de-escalation processes. Moreover, it provides theoretical insight into the mechanisms through which escalation as well as de-escalation processes may evolve. Finally, it makes us able to grasp how escalation processes may be delayed as a result of strong political interest.
The lessons learned from our cases should be interpreted in the light of the fact that they unfold in the Danish political and administrative context. The Danish political system is relatively decentralized, the municipalities bearing responsibility for many welfare services. Danish political and administrative culture is often described as informal, based on dialogue and trust. These characteristics are likely conducive to the dynamics that unfold in our cases. We found that various aspects of the escalation of performance management regimes are subject to negotiation and sometimes learning through feedback. In both cases, we have also observed a lack of escalation regarding intense economic sanctions. These observations correspond to our description of Danish political culture as dialogue-oriented, pragmatic, and organized according to strong and universal principles. If there were a context in which we would expect escalation to occur only partly—and sometimes slowly—we would suggest it to be the Danish one. This speculation should only be taken so far, however, as we could have found examples from other sectors and other policy areas that would fit less comfortably with the somewhat simplified national pattern described here.
The time is not yet ripe for a general theory of escalation of PI systems. Our modest conclusion thus far is that while escalation often occurs, it does not occur automatically and at the same speed everywhere. Instead, we have identified eight dimensions of escalation. Escalation in some dimensions may occur parallel to de-escalation in others. While a given national political culture may influence some aspects of escalation (in our situation, the lack of escalation of economic sanctions), mechanisms such as learning through feedback from rule takers to rule makers and political negotiations help explain the trajectories of escalation/de-escalation in individual cases, but the total set of interacting factors is complex.
We therefore encourage future studies to test our framework in other settings (national, sector, and task) and to delve deeper into the mechanisms contributing to escalation and de-escalation.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research is partly financial supported by the Master of Public Governance.