2.1 Accountability through Elections: The Pivotal Role of Retrospective Considerations

Before being a mechanism of ex post accountability that casts its shadow over decision-makers, elections are a mechanism of ex ante authorization to represent the people, which constitutes ‘promissory’ representation based on the prospective evaluation of pledges made by competing parties and candidates (Reference MansbridgeMansbridge, 2003). However, elections are a relatively crude instrument for the formulation of mandates, so we should not expect that the authorization to govern will necessarily determine the content of public policies. Despite the growing use of voting advice applications, we generally have relatively vague profiles of parties and candidates in mind and simply diffuse expectations from our representatives.

Unsurprisingly, then, discretion is inherent to representative government, and principal–agent theory has convincingly demonstrated that delegation by a principal usually leads to the discretionary power of his or her agent (see Reference MillerMiller, 2005): if we apply the contractual metaphor to describe delegation mandates, apart from being implicit, the contract is an incomplete one, with many points remaining unspecified.Footnote ¹¹ As a counterweight to the agents’ discretionary power and related risk of opportunistic behaviour, principals need to hold them accountable to preserve their own interests. Therefore, in representative democracies, elections are also a mechanism of ex post accountability: the threat of electoral sanction should incentivize office-holders to keep their promises to voters so that citizens retain some of their self-determination despite their decision-making power being delegated.

To some extent, electoral accountability is condemnatory, with the importance of so-called negative voting that primarily seeks to punish ‘rascals’ by deauthorizing them to rule, a reflection of the more general prevalence of a negativity bias in politics (Reference SorokaSoroka, 2014). However, the electoral sanction should primarily be seen as instrumental rather than purely expressive, and it is forward-looking in that it is intended to change the course of things even if voters have no clear prospective motives in mind.Footnote ¹² Moreover, its sheer threat should act as a disciplinary device to discourage agents (representatives) from deviating from principals’ (voters’) preferences simply because it is rational for agents to anticipate in their calculations the future judgement of their principals and the potential consequences thereof (Reference ManinManin, 1997: 179). For the effective operation of democratic government, the prospect of electoral accountability and sanction is most crucial, making the ‘law of anticipated reactions’ work (Reference FriedrichFriedrich, 1937). Absent this expectation – if incumbents discount the risk of being defeated in the next election, for example, in the absence of credible alternatives or because their electorate displays strong partisan loyalty – office-holders would have no incentives to remain responsive to the voters who selected them and temporarily authorized them to rule.

In such a democratic perspective on accountability, the latter serves as a mechanism regulating representative government through its ‘thermostatic’ action. This mechanism is designed to ensure the smooth coexistence of ‘policy representation’ and ‘public responsiveness’ (Reference Soroka and WlezienSoroka and Wlezien, 2010): to remain in power, governments are incentivized to adjust their policy choices to citizens’ preferences,Footnote ¹³ and citizens in turn adjust their views to consider the existing supply of policies (the erosion of partisan identities and the prevalence of issue voting facilitating flexibility). In addition, representation is ‘dynamic’ (Reference Stimson, Mackuen and EriksonStimson, Mackuen and Erikson, 1995): governments use policies to respond to perceived shifts in public opinion, and policies generate feedback effects, with voters also modifying their preferences depending on the changes they discern in governmental action and the outcomes thereof.

It is therefore important for democratic control that the electorate votes based on retrospective rather than just prospective considerations. For example, the evaluation of past performance in office provides useful information to assess the credibility of commitments concerning future performance. To what extent are voters motivated above all by their desire to evaluate the incumbent government, and is their retrospective assignation of responsibilities and evaluation of performance based on sound evidence? Research focusing essentially on economic voting – initially in the United States, and then with a comparative approach (Reference AndersonAnderson, 2007; Reference Duch and StevensonDuch and Stevenson, 2008) – established that, whether voters are concerned with their individual well-being (egocentric voting) or with the state of the economy of their country as a whole (sociotropic voting), they do reward or sanction governments depending on their evaluations. However, empirical research also suggests that voters encounter cognitive limitations in their appreciation. Voters’ decisions are the outcome of complex cognitive processes: they are not just contingent on the evaluation of outcomes that voters attribute to policy; they also require that voters judge the responsibility of governments for these outcomes, their skills and their efforts (Reference Hobolt and TilleyHobolt and Tilley, 2014: 123). The necessary copresence of these elements renders the accountability chain fragileFootnote ¹⁴ and subject to bias, especially due to media coverage that can be misleadingly negative (Reference Achen and BartelsAchen and Bartels, 2017: 108).

2.2 Evidence on Voting Patterns: A Mixed Picture

Findings from survey and experimental research indicate that people use shortcuts to make their decisions on political matters: heuristics, such as party cues, the opinions of friends, media frames or the positions of interest groups, are useful in that respect, and even more so for those (generally those who are less educated) who lack familiarity with politics. They thereby reduce upper-class bias in participation in public affairs, but this may lead to inadequate conclusions and poorly informed choices. Shortcuts are instrumental in overcoming informational shortfalls and making decisions despite the complexity of political matters but do not protect against evaluation and attribution errors due to bounded rationality (such as overemphasizing recent events or overweighting incumbents’ broken pledges).

Reference Achen and BartelsAchen and Bartels (2017) challenged what they call the ‘romantic folk theory’ of democracy, according to which voters would choose parties and candidates on the basis of genuine evaluations of policy. They demonstrated that voters’ choices continue to be tainted by social identities and partisan loyalties, with ideology being an information-saving device used as a filter to decode reality. Ideology thus allows voters to surmount cognitive limits to some extent, considering that the retrospective model sets the bar too high with respect to the level of sophistication required from voters. However, as preexisting views of the world serve as perceptual screens, responsibility judgements are biased by prior political beliefs: ‘people prefer to give credit for good outcomes to political actors they already like and are equally willing to blame poor outcomes on political actors they already dislike’ (Reference Hobolt and TilleyHobolt and Tilley, 2014: 49). For example, partisanship colours perceptions of the economic situation and of who is responsible for the good or bad state of the economy (see, among many others, Reference Tilley and HoboltTilley and Hobolt, 2011). Furthermore, voters exaggerate politicians’ influence (Reference Caplan, Crampton, Grove and SominCaplan et al., 2013), and irrelevant events seem to affect evaluations of governmental performance.

However, there are also more optimistic views. Debating the assumption that the government is commonly blamed for people’s misfortune, even if it is not responsible for ‘bad times’ such as floods or droughts, Reference Gasper and ReevesGasper and Reeves (2011) showed that the electorate is able to separate random events such as natural accidents from governmental responses. Using various counterexamples, Reference StokesStokes (2018) concluded that Achen and Bartels’ severe judgement needs to be nuanced. Some citizens are more predisposed than others to tame affect and partisanship with reflection (Reference Arceneaux and Vander WielenArceneaux and Vander Wielen, 2017). For example, evaluating the inadequate governmental response to Hurricane Katrina, more sophisticated respondents to a survey in Louisiana were less prone to focus blame disproportionately on the president as the most obvious national political figure (Reference Gomez and WilsonGomez and Wilson, 2008).Footnote ¹⁵ Findings from a study based on data from twenty-five democracies across an array of policy areas, including the economy, social welfare, immigration and national security, confirm that holding governments accountable for past performance is mainly the prerogative of highly sophisticated citizens (Reference de Vries and Gigerde Vries and Giger, 2014). At the same time, however, the sophistication gap narrows when voters attach a higher degree of salience to a policy area, that is, when the issues at stake are important to them. After all, isn’t that what counts most for electoral accountability?

2.3 Systemic Variables Matter

The effectiveness of electoral accountability also depends on the institutional architecture of the political system. An analysis of elections from seventy-five countries revealed that separation-of-powers systems are superior in terms of accountability; however, this is only true under the condition of strong party competition and with the exceptions of semipresidentialism and episodes of cohabitation (Reference Hellwig and SamuelsHellwig and Samuels, 2008). The reward–punishment model works best when the mechanisms of accountability are simple (Reference AndersonAnderson, 2000), which is not surprising if we think of the sophistication required to allocate responsibility and evaluate performance. Blurring responsibility and strengthening cognitive bias, vertical and horizontal institutional and power fragmentation attenuate democratic accountability and the effect of evaluations on incumbents’ popularity.

Such structural features are considered to negatively impact electoral accountability in federalist political systems in which policies are produced at various jurisdictional levels. Federalism allows for opportunities to designate representatives at multiple levels but complicates effective popular control by generating confusion about who is in charge of policies (Reference LeónLeón, 2018). Lack of clarity concerning responsibility facilitates adherence to the blame-shifting frames produced by political elites to evade responsibility for poor performance, as in the case of the responses to Hurricane Katrina.

Using data from the European Election Studies, Reference León, Jurado and MadariagaLeón, Jurado and Madariaga (2018) found that when partisans of the national government in federal states are confronted with poor economic outcomes, they are more likely to ‘pass the buck’ to the regional level following a blame-shift logic, while this mechanism is absent in non-federal states. Reference CutlerCutler (2008) concluded that federalism is challenging for Canadian voters wishing to reward or punish their governments for policy outcomes. They do not easily distinguish the roles of different jurisdictional levels, and there is little variation across issues; attentiveness to politics only very slightly improves the quality of responsibility attributions and only on issues where responsibility is objectively clearer. A recent study showed that interprovincial policy variation with respect to coronavirus testing in Canada is not correlated with public assessments of the adequacy of provincial testing, confirming that Canadians have difficulty assigning responsibility to the correct level of government despite high levels of issue saliency and media attention (Reference Kennedy, Sayers and AlcantaraKennedy, Sayers and Alcantara, 2022).

These studies all concur that multilevel structures risk undermining responsibility attribution and thus democratic accountability, and we should keep in mind that the complexity of multilevel arrangements usually goes far beyond the formal distribution of power. Competences are not just divided between jurisdictional levels; they are also frequently shared among them, and different levels may need to pool resources to make policies. All this necessitates collaboration between multiple interdependent actors entangled in dense networks with informal interactions that are not easily visible from the outside. In other words, even if citizens are able to allocate formal responsibilities correctly, they may only see the most visible part of the actual policy-making processes, with a possible gap between accountability and the wielding of power (see Section 4).

Moreover, the characteristics of the party system mediate the role of elections as an accountability mechanism. Proportional representation secures a fairer and more pluralist representation of political forces but risks leading to excessive fragmentation of the party system and to a multiparty government that dilutes responsibility. In contrast, first-past-the-post electoral systems set high thresholds for representation but facilitate the formation of single-party governments. They also tend to induce bipolar dynamics, which makes it easier to identify clear alternatives that deserve to be supported. A recent comparative analysis with data from 400 parliamentary elections (Reference Kam, Bertelli and HeldKam, Bertelli and Held, 2020) revealed that electoral gains and losses are more clearly mirrored in the gains and losses of ministerial portfolios in strongly bipolar party systems, while multiparty systems increase the risk of unstable majorities, with the consequence that losers in elections may remain in office. Furthermore, experimental evidence suggests that legislators elected in single-member constituencies have more sustained interactions with their voters and are more reactive to the requests of this particular – admittedly narrow – segment of the electorate (constituency service) (Reference Breunig, Grossman and HänniBreunig, Grossman and Hänni, 2022).

Allocating responsibility may be even more difficult when the government is divided, as in presidential systems with different majorities in the executive and the legislative resulting in a relatively weak government facing a strong ‘veto point’, or when the legislature is bicameral, with two differently composed chambers. However, it seems that the cohesiveness of the national government is the key factor (Reference Hobolt, Tilley and BanducciHobolt, Tilley and Banducci, 2013): government becomes a clearly identifiable target, and according to survey data from twenty-seven European countries, this matters most for clarity of responsibility, regardless of whether institutional power is shared (in upper chambers, other levels of government, etc.).Footnote ¹⁶ This is reassuring because it appears that institutional checks can tame majoritarian discretion without greatly hampering democratic accountability.

2.4 Policy Responsiveness: The Invisible Hand?

Equally reassuring from the democratic perspective on accountability is the fact that public opinion indeed operates as a ‘thermostat’ despite the limits of retrospective voting. This is likely because governments fear voters’ discontent and usually care about the risk of sanctions, even if voters are not always well equipped to evaluate governments’ actions. Popular control of the government may rely on misperceptions, and elections are not an optimal instrument of accountability due both to issue bundling (governments simultaneously conduct multiple policies on which they can be evaluated) and the difficulties of comparing actual governmental performance with the more or less credible pledges of challengers. However, governments apprehend democratic control simply because it would be too risky for them to bet on its weaknesses. The degree of congruence between opinion preferences and policy is indeed remarkable (Reference Rasmussen, Reher and ToshkovRasmussen, Reher and Toshkov, 2019). In return, governmental parties are more likely to be rewarded by voters if they have fulfilled more of their pledges during incumbency (Reference MatthießMatthieß, 2020).

Of course, this virtuous circle is welcome from a democratic perspective on representation. However, the main limitation of the studies that conclude with such positive results is that the processes leading to the congruence between governments and public opinion are not scrutinized in depth (Reference de Wilde and Rauhde Wilde and Rauh, 2019). Policy-making remains a black box, and it is unclear what kinds of tools decision-makers use to recognize the signals emerging from public opinion (or from narrower but more easily identifiable electoral clienteles) and to influence political demand. Research on politicians’ perceptions and strategic moves would also benefit from empirical findings on reputation management by senior staff in public agencies and social psychological research on how individuals experience accountability pressure (see Sections 4 and 6, respectively).

Furthermore, unpacking the black box often reveals actors other than authoritative government officials who influence the course of things without being held politically accountable: think of the roles of experts, interest group representatives or even editorialists. A conception of democratic regimes that focuses on the citizens–elections–representatives triad is no doubt normatively attractive due to the direct line that it posits upwards from ‘we the people’ to the government and downwards from the government to society (Reference Hupe and EdwardsHupe and Edwards, 2012: 182). However, this conception does not adequately capture the way the governmental process works, as the next sections will show.

3.1 Checks and Balances: A Composite Image

The capacity of legislators to oversee governmental activity effectively depends on a wide range of factors (Reference Pelizzo and StapenhurstPelizzo and Stapenhurst, 2014). The institutional setting matters, of course. Taking the case of the United Kingdom’s model of ‘Westminster’-style parliamentarism, the formal accountability chain is simple, with a direct line that has the merit of clarity from government to parliament and from parliament to the electorate. However, this model leaves little space for effective horizontal checks, as power is concentrated in the cabinet or even in the hands of the prime minister, and the governing party usually controls a disciplined majority in parliament. The opposition may ‘bark’ as expected in a logic of adversarial partisan politics but is not able to ‘bite’ (Reference Wright, Feldman and ElliottWright, 2015). More specific factors also matter; for example, cross-country research suggests that the control of parliamentary committee chairs significantly strengthens opposition parties’ ability to engage in legislative review (Reference Fortunato, Martin and VanbergFortunato, Martin and Vanberg, 2019). In non-parliamentary systems, as in the United States’ presidential government, the executive becomes dominant (Reference Posner and VermeulePosner and Vermeule, 2011), but it undergoes significant pressure in its horizontal accountability obligations towards the legislature, despite its powers (Reference Kriner and SchicklerKriner and Schickler, 2017). Particularly in situations of ‘divided government’ in which different parties control the presidency and Congress, the existence of a regime of power separation incentivizes each branch to be sensitive to the preferences of the other branch. This prompts formal or informal interbranch bargaining to avoid policy blockades, but it is difficult to avoid the latter when partisan polarization is strong.

Notwithstanding cross-country differences, the recent transformations of governance styles challenge parliaments’ power and seem to be at the origin of a more or less pronounced ‘deparliamentarization’. In the next section, we will assess the consequences of the fragmentation of power with the advent of collaborative and delegated modes of governance. A competing narrative emphasizes the rampant concentration of power with the ‘presidentialization’ of politics (Reference Poguntke and WebbPoguntke and Webb, 2005) and the advent of ‘court government’ (Reference SavoieSavoie, 2008), dominated by the prime minister, who is surrounded by an informal group of her or his few favourite ministers, and expert staff of policy professionals, public servants and PR specialists (Reference SvallforsSvallfors, 2020). Whatever the most accurate narrative of governance transformations, both the fragmentation of power in polycentric networks of actors and its concentration in the hands of the executive are considered to negatively impact parliamentary influence. In addition, the general trends of the last decades, such as the increased complexity of policy matters or the internationalization of policy-making, have lessened the centrality of legislative bodies in law-making processes and reduced their roles as accountability forums that hold governments – and rule-making entities more generally – accountable.

With policy issues requiring considerable expertise and policy-making moving partly into less visible negotiation arenas (sometimes even beyond the national state), parliaments are confronted with situations of information asymmetry, to their detriment. Moreover, parliamentarians may not just be unprepared to check executives; they may be unwilling to do so, as with members of governmental parties, or when the issues at stake are of low salience to them and not generating any electoral payoffs. Some parliaments have nevertheless reacted to their loss of power and managed to ‘fight back’ (Reference Raunio and HixRaunio and Hix, 2000), as with the development of parliamentary Select Committees in the United Kingdom, which scrutinize, conduct inquiries and report on governmental work (Reference Schonhardt-BaileySchonhardt-Bailey, 2022). However, the strength of parliamentary reactions varies (Reference Auel, Rozenberg, Tacea, Hefftler, Neuhold, Rozenberg and SmithAuel, Rozenberg and Tacea, 2015), and perhaps more importantly, even this countertrend generates trade-offs with regard to accountability because the parliamentary scrutiny of governmental work is most effective when exercised informally. This is detrimental to the external accountability to the general public, especially as parliamentary control power translates into influence and bargaining that develop in secluded arenas (Reference Auel and BenzAuel and Benz, 2005). Informality and lack of transparency aggravate citizens’ difficulty in allocating responsibility (see Section 2).

Unlike democratic accountability to voters, horizontal accountability between institutions mirrors the constitutional perspective on accountability as a mechanism curtailing abuses of power by establishing checks on rulers’ authority. The primary tool in that respect is the possibility to veto decisions, and the more veto points a political system contains (Reference ImmergutImmergut, 1992), the more rule-makers become accountable to veto ‘players’ (Reference TsebelisTsebelis, 2003), that is, actors able to block or turn down decisions that do not match their preferences. In the case of symmetric bicameralism, for example – with two legislative chambers endowed with equal competences on important matters – the support of legislation by both chambers is necessary. To avoid the deadlock arising from disagreement between the two houses and the exercise of mutual veto rights, conference committees are set up in which deliberation and informal bargaining take place. Although this is not formulated as such in the literature on veto points, each side becomes de facto accountable to the other in these arenas, as it needs to convince the other side about its preferences. Moreover, when policy-makers face such an incentive structure that induces self-restraint and mutual adjustment, they are likely to anticipate and incorporate the expectations of veto players who act as de facto accountability forums. In other words, they are incentivized to absorb forums’ preferences and objections even before these forums formulate any feedback.Footnote ¹⁸

A more recent phenomenon is the growing horizontal accountability of rule-makers to institutions with a core oversight function,Footnote ¹⁹ which draw their legitimacy from being independent and non-majoritarian. Their mission is to safeguard the rule of law and protect minorities and individuals from violations of their rights and, more generally, to ensure compliance with principles of good governance. Traditionally, courts are the relevant forums for legal accountability, but auditing institutions have gained importance as forums of financial accountability with the movement towards performance evaluation and measurement (see Section 4). It is also worth noting that the hard accountability of public authorities to this kind of body, with assorted sanctions, is currently frequently accompanied by softer forms of accountability. A prominent example is accountability to ombudsman institutions: although the latter have no direct sanctioning power, they can mandate public organizations to justify their decisions and practices. It is sometimes remarked that ombudsman institutions’ lack of a formal ‘bite’ can be offset through creative operation (for instance, with an active communication policy) that pressures public authorities to comply (Reference Bovens and WilleBovens and Wille, 2021: 859–60). This would be potentially a case of an oversight institution whose de facto power to control and impose a change of practice is greater than the institution’s de jure prerogatives.

3.2 Judicial Activism: Janus-Faced

The remainder of this section is dedicated to the power that the judiciary draws from its growing role as an accountability forum, which resulted from concerns about the risk of tyranny by the majority and more generally of abuses of power by rulers but also out of the strategic calculations of institutional designers (Reference VanbergVanberg, 2015). In recent decades, courts have seen their role gaining importance, especially regarding their capacity to overrule legislation considered incompatible with constitutional or human rights principles. The emergence of a ‘transnational legal order’ (Reference Shaffer, Ginsburg and HallidayShaffer, Ginsburg and Halliday, 2019) embodied in the ‘global spread of constitutional review’ (Reference Ginsburg, Caldeira, Kelemen and WhittingtonGinsburg, 2008) has paralleled the successive waves of democratization, mostly in Europe and Latin America. As forums to which rule-makers are accountable, courts intrude into the policy-making process in response to complaints or on their own initiative, depending on the judicial system.Footnote ²⁰ Judicial activism is a dimension of the broader ‘rise of the unelected’ (Reference VibertVibert, 2007) that is also visible in the diffusion of the agency phenomenon (see Section 4).

The judiciary draws its legitimacy as an accountability forum from its independence, which is nevertheless subject to empirical variation and sometimes the object of political controversy. The so-called judicialization of politics first means that courts become policy actors just by being asked to arbitrate policy controversies, even if they do not aim to advance any particular policy preferences. Beyond that, judicialization is prompted by constitutionalization: judges become de facto co-legislators when courts possess constitutional review rights because legislators are then incentivized to anticipate their verdict. By acting as veto players capable of striking down legislation (regardless of their motivations), courts become pivotal actors and indirectly take over legislative functions. Like the shadow of future elections that incentivizes incumbents to be responsive to voters’ preferences, the shadow of court rulings perceived as likely to reverse legislation prompts elected officials to absorb the preferences attributed to the judiciary. In the United States, for example, a newer area of interbranch bargaining research incorporates the judicial branch into the model of separation of powers: legislators seek to reach decisions that are as close as possible to their own preferences but robust to court interference (Reference Knight and SchwartzbergKnight and Schwartzberg, 2020: 268).

Moreover, although courts are primarily considered ‘negative’ legislators because of their formal veto rights, which allow them to nullify legislation, they may actually be more than that. They make recommendations in their rulings as to how legal acts should be revised, and the official legislators subsequently tend to follow them ex post in addition to being incentivized to take judges’ anticipated opinions into account ex ante. Needless to say, when judges’ preferences conflict with voters’ preferences, elected officials may be caught in accountability dilemmas. These dilemmas are reflected in the debates on judicialization within political philosophy and legal theory, as Reference Wright, Feldman and ElliottWright (2015: 100) observes: ‘While political constitutionalism asserts the primacy of democratic politics as the arena of accountability and the necessary limitations of the role of unelected judges, legal constitutionalism points to the dangers of mere majoritarianism and the need for the protective accountabilities that only an independent judiciary can provide.’

The frictions between the exercise of popular or parliamentary sovereignty embodied in majority rule and the protection of individual and minority rights guaranteed by the rule of law as interpreted by courts have given rise to controversies regarding the quality of democracy. Developments in the direction of judicialization are welcomed by those defending a liberal–constitutionalist conception of democracy, who set the rule of law and protection of rights as primary objectives and criticize ‘defective democracies’ (Reference MerkelMerkel, 2004) for not providing sufficient countervailing powers to unrestrained majority rule. This approach builds on the fact that, when courts are perceived as independent and impartial, they usually enjoy strong legitimacy. Courts as institutions and judges as professionals tend to build credit and consolidate their reputations in the long run (as independent agencies and their expert staff do, as described in Section 4). They usually enjoy high levels of positive opinions from the general public, although – or maybe rather becauseFootnote ²¹ – they constrain popularly elected policy-makers (Garoupa and Ginsburg, 2015). Therefore, legal constitutionalism praises courts’ contribution to good governance when they act as safeguards against democratic backsliding (Reference Blauberger and KelemenBlauberger and Kelemen, 2017), when the judiciary is seen as a favourable venue for the development of rights (Reference CichowskiCichowski, 2013), or when a growing number of international courts act as norm entrepreneurs to strengthen the rule of law and promote democratic practice, including holding states (as well as transnational actors) accountable and inducing them to provide public justifications (Reference Kuyper and SquatritoKuyper and Squatrito, 2017). Obviously, this doctrine is highly critical of situations in which governmental interference undermines the independence of justice, as it recently happened in European Union member states such as Hungary and Poland.

In contrast, critics contend that the rule of law entails the ‘rule of lawyers’ (Reference Kratochwil, Palombella and WalkerKratochwil, 2008). Those who do not feel comfortable with the consequences of judicialization point out the ‘counter-majoritarian difficulty’ that is inherent to it – in other words, the risk that majority rule is undermined by judicial activism. They are worried about the risk of the formal separation of powers being undermined, but more than that, they criticize the advent of government by judges, who are politically unaccountable by design (as a condition of their independence) and most likely unrepresentative. To what extent should unelected accountability forums be able – in the name of impartiality and the public interest – to hold elected politicians accountable and indirectly incentivize elected officials to consider their views? This is an open question, but it can only be answered in light of the evidence regarding the amplitude of judicialization, which is admittedly mixed.

On the one hand, judicialization affects the treatment of ‘mega’ issues such as the importance of secularism, the contours of citizenship and national identity, electoral legislation and outcomes, corruption indictments and impeachment, bans on political parties and, more generally, the shaping of the democratic process itself as in rulings about transitional justice or emergency legislation (Reference HirschlHirschl, 2008). Although many of these issues speak to the core of the polity as such, reflect deeply salient dilemmas and ought to be the object of deliberation in the public sphere, they have recently been framed in several countries as constitutional issues. As political authorities are unwilling or unable to resolve them, they transfer the problem to the judicial circuit; but judicial intervention can also take place unsolicited. Whatever its origin, such a process of depoliticization leading to a ‘juristocracy’ is normatively problematic, especially if one also considers the spread of quasi-constitutional supranational treaties and legal institutions that place several facets of global governance beyond democratic reach (Reference HirschlHirschl, 2013).

The global nature of the trend described by Hirschl should indeed not be underestimated: ‘Please fasten your seatbelts’, wrote the author before enumerating a long list of high-profile cases that illustrate his argument. However, in large part, the trend affects polities that face deadlock or weakly consolidated democracies, sometimes with a politicized judiciary that is instrumentalized by political forces. There is also contrasting evidence, which should rather lead us to conclude that the claims about the rising influence of non-democratically legitimized judges are overstated. Despite judges enjoying discretion, governments can adopt court-curbing measures, implement court rulings selectively (if not ignore or even override them) and ultimately seek to strip courts of their power.

Backlash may be difficult if the constitution must be amended for that purpose and costly if courts enjoy public support, but judicial power is not unrestrained even in well-established constitutional democracies: a comparative study of the United States, France and Germany covering thirty-six years (1974 to 2009) showed that the ability of constitutional courts to act as veto players is subject not just to cross-country but also to temporal variation because it depends on the way judges are selected, the pattern of government control, the rhythm of governmental alternation and the type of legislative procedures (Reference Brouard and HönnigeBrouard and Hönnige, 2017). Moreover, judicial review can also directly or indirectly contribute to policy-makers’ goals, so that an influential judiciary is not necessarily ‘a nuisance that a dominant coalition would want to eliminate’ (Reference VanbergVanberg, 2015: 172). Judges are sensitive (and perhaps even adhere) to the preferences of the majority of public opinion, and even independent judges care about policy-makers’ preferences and feel restrained by the shadow of court packing measures, just as policy-makers care about and seek to anticipate judges’ opinions. In sum, the relationship between the political and judicial spheres is one of an interplay between actors that are interdependent and behave strategically under this kind of constraint.

Overall, decision-makers in established democracies experience different kinds of horizontal accountability constraints. Empirically, it is difficult to observe a clear-cut trend; cross-country variation remains important, and there are developments that hamper interinstitutional accountability. Regarding the normative evaluation of horizontal accountability, the complexity of our societies may justify the existence of multiple checks on power wielders, and horizontal mechanisms are considered more likely to constrain governmental choices than vertical accountability to voters because institutional forums have the capacity to directly oversee governments continuously and apply direct sanctions. At the same time, for horizontal accountability to operate properly, it must be coupled with vertical accountability to citizens and effective societal accountability, regardless of frictions arising between these accountability forms: as Reference Mechkova, Lührmann and LindbergMechkova, Lührmann and Lindberg (2019: 58) observed, ‘Without fully clean elections, autonomous opposition parties and a developed civil society and media, no country in the world has yet achieved fully effective government oversight through independent high courts, vigorous parliaments, or other institutions’.

Ultimately, with more assertive parliaments and with judicialization, executive and administrative power wielders become more accountable to other – elected or unelected – elites: politicians and judges. These interinstitutional modes of accountability may mitigate accountability to the citizenry by design, but they may also undermine it unintentionally. As we shall see in the next section, similar questions can be raised about the accountability of the bureaucracy after numerous administrative reforms.

4.1 Delegation and the Limits of Political Control over the Bureaucracy

For those adhering to the principal–agent model of delegation, accountability is an essential element of the incentive structure that predisposes agents to act in accordance with their principals’ preferences. The model emphasizes the risk of moral hazard by assuming that agents are motivated by the pursuit of self-interest, but the disconnect from the preferences of elected officials may simply be because the administration has its own conceptions of good policy (Reference Peters, Bovens, Goodin and SchillemansPeters, 2014: 211). Hence, for different reasons, elected officials may find that control of the bureaucracy is necessary, but they may also be inconvenienced by it: control necessitates time investments for acquiring expertise and energy in conducting scrutiny, which in turn entails opportunity costs. Is there any functional equivalent?

‘Stewardship theory’ presents itself as an alternative to principal–agent theory: it challenges its narrow economistic assumptions,Footnote ²² and it is grounded by contrast on a psychological understanding of actors’ orientations and on a sociological understanding of the organizational context of action. Its advocates find that agency theory is too cynical and believe that agents behave as ‘stewards’ who deserve to be trusted, as they are animated by a feeling of duty and find their self-realization in behaviour that is collective-minded and oriented towards the attainment of organizational goals (Reference Davis, Schoorman and DonaldsonDavis, Schoorman and Donaldson, 1997).Footnote ²³ To be sure, this trust may be misplaced, but agency drift can be mitigated if agents display strong value commitments, if they are socialized in the values of their principals, or if they adhere to the norm that their principals – whatever their values – have the authority to command. Scrutiny and sanctions then become unnecessary (and even potentially demotivating), as the agent feels that obeying the principal is the right (or normal) thing to do. To use Reference March, Olsen and GoodinMarch and Olsen’s (2013) distinction, compliance is not obtained through a logic of consequences, that is, with the anticipation of rewards and sanctions, but through a logic of appropriateness: it is rule-based and, as such, is normatively valued conduct.

However, the stewardship model is excessively optimistic. Although a sense of responsibility is necessary, internal checks may not suffice without external control, for instance, to prevent or correct errors by loyal servants. However, we also need to be aware of the structural limitations of control due to the length of the delegation and accountability chains, combined with the uncertainty that affects processes in each of their parts. In Max Weber’s traditional conception of bureaucracy, hierarchy plays a crucial role in ensuring compliance, but even low-level frontline professionals enjoy relative autonomy from their superiors because, for example, they can dissimulate information from them. In addition, as they are embedded in webs of informal relations, they may face incompatible expectations, standards and prescriptions for action from their superiors upwards, target groups, service users and regulatees downwards, and their professional peers sideways (Reference Hupe and HillHupe and Hill, 2007; Reference Romzek, LeRoux, Johnston, Kempf and PiatakRomzek et al., 2014).

Of course, deliberate overlaps may enhance the robustness of an accountability regime. There are more opportunities to call to account, the functioning of accountability no longer depends on a single forum, and individual flaws can be compensated for by duplication (Reference Willems and Van DoorenWillems and Van Dooren, 2012). However, if multiple demands and steering signals are uncoordinated, this may expand the bureaucracy’s room to manoeuvre (Reference MoeMoe, 1984: 768–9). Public organizations invest resources to handle accountability challenges and invent coping mechanisms that balance tensions.Footnote ²⁴ Recent experimental research has demonstrated that accountable agents seek to prioritize among conflicting expectations, unsurprisingly, with the threat of sanction as the most important driver of prioritization (Reference Aleksovska, Schillemans and GrimmelikhuijsenAleksovska, Schillemans and Grimmelikhuijsen, 2022). Dilemmatic situations in the face of ‘quasi-anarchically structured’ (Reference SchillemansSchillemans, 2016: 1402) webs of accountability may nevertheless also lead to what Reference KoppellKoppell (2005) aptly called a ‘multiple accountabilities disorder’ (MAD), which is likely to paralyze individuals, organizations or segments thereof and cause tragic operational failures (Reference Romzek and DubnickRomzek and Dubnick, 1987; Reference Romzek and IngrahamRomzek and Ingraham, 2000: see also Section 5.2).

Whatever its side effects, control within the state apparatus seems to have been increasing in formality, complexity, intensity and specialization.Footnote ²⁵ This does not mean that it has necessarily gained effectiveness. Moreover, the description of political-administrative relations must be put into context: multiple country‐specific trajectories of administrative regimes exist and reflect historically rooted regional traditions, such as that of Anglo-American, Scandinavian or Germanic countries or states in Latin Europe that mostly followed the Napoleonic model (Reference PetersPeters, 2021).

4.2 Administrative Reforms and Accountability

Reform trajectories have diverged to some extent across states, as they are conditioned by their domestic institutional context, power balance and path dependencies, but also within them.Footnote ²⁶ However, for the sake of brevity, the emphasis in this section is on the common shifts in accountability patterns that have taken place in recent decades. Many reforms have indeed been driven, inter alia, by the common expectation that enhanced accountability would lead to better performance, and this in turn should provide more legitimacy to the state. These reforms originated in a critique of the bureaucracy, often portrayed as too powerful and obsessed with rules to the detriment of results and efficiency, and aimed at greater responsiveness to the needs of service users. For that purpose, one major change concerning accountability was the shift from legalism to managerialism that occurred under the banner of New Public Management (NPM) reforms.

Without being completely replaced, compliance accountability that traditionally focuses on institutional rules and formal procedures to be followed was supplemented by performance accountability.Footnote ²⁷ Focusing thus on output quality and problem-solving capacity and supported for that purpose by evaluation instruments such as satisfaction surveys and quantitative indicators, accountability also moves ‘downwards’ to become ‘proximate’ with target populations in the role of accountability forums. The feedback of – often loosely defined – societal stakeholders, such as more or less organized ‘users’ or ‘clients’ directly affected by public services, is strongly valued. More or less institutionalized modes of ‘social’ or ‘civil’ accountability should facilitate the formulation of appropriate policy measures and enhance public trust in government (Reference BrummelBrummel, 2021; Reference Sørensen and TorfingSørensen and Torfing, 2021). In a sense, similar results are expected from the accountability of the administration to service users at the microlevel to what we expect from electoral accountability at the macrolevel: better responsiveness to the needs of target populations, leading in turn to ‘output legitimacy’.

Nevertheless, there is no compelling evidence that changes in accountability processes improve performance: indeed, ‘empirical studies observe tensions, ambiguities, contradictions and unintended consequences’ (Reference Jann, Christensen and LægreidJann, 2016: 41).Footnote ²⁸ This happens for various reasons. For example, performance expectations can be unstable and conflict with each other. Furthermore, for such accountability mechanisms to foster improvements in terms of policy, citizens must highly value performance, but they may have other priorities and ignore evaluation opportunities. It is quite plausible that people are not attentive to performance information, especially if they do not feel strongly affected by the assessed organization or service. In addition, their judgement may be subject to bias (as already noted in Section 2) if they have difficulty identifying policy workers (Reference BertelliBertelli, 2016) or if the information provided – such as quantitative indicators – is manipulated. For instance, this may happen when the media publish league tables even if this kind of publicity was not the purpose of performance assessment (Reference Pollitt, Dubnick and FredericksonPollitt, 2015).

More importantly, service users are asked to evaluate performance, while aspects of service provision become less transparent, making evaluation more difficult. Digitization is a case in point: the – sometimes insidious – development of ‘algorithmic regulation’ (Reference Yeung and LodgeYeung and Lodge, 2019), that is, decision-making informed by artificial intelligence, exacerbates information gaps.

Although the problems described by Busuioc raise important accountability questions, it is usually difficult to identify the logic behind algorithmic decisions and to correct the biases and unintended consequences of the use of algorithms. Meaningful oversight and malfunction diagnosis are burdened due to the large deficits in expertise that affect lay audiences and, at times, even the algorithm users within the public service, whose faith in these systems may be undue. Algorithmic outputs result from the value choices of system designers, who thereby indirectly become part of the policy-making process. However, such choices and parameters are usually not explicitly spelled out and made open to assessment by outsiders (e.g., stakeholder groups negatively affected by decisions based on algorithms), so it is difficult for them to understand and contest these choices. As explained by Reference BusuiocBusuioc (2021), algorithmic accountability first implies critically scrutinizing how the algorithmic results are reached, but this process often remains opaque. What is more, one needs to understand the role algorithmic inputs play in human decisions: for example, automated tools can serve as ‘moral buffers’ for humans to shift the blame and evade responsibility over a decision. In other words, accountability becomes very demanding when one has to unpack both algorithmic processes and the human–algorithm interaction. It is therefore not surprising that (as suggested by survey data: Reference König, Felfeli, Achtziger and WenzelburgerKönig et al., 2022) people are willing to trade away algorithmic transparency for small gains in effectiveness, even in sensitive areas such as policing.

Perhaps even more fundamentally, the shift to accountability downwards is not uncontroversial from a democratic point of view, although participatory forms of accountability can be seen as positive at first glance. For instance, one may find it legitimate if feedback is provided above all by those with intense preferences on policy issues and consider that stakeholder voices should count most because they express the views of those directly affected. However, this kind of social accountability may also be normatively problematic regarding the credibility of the claims of affectedness, which can be contested, and regarding inclusiveness, as unlike citizens’ votes, stakeholders’ voices are not equally distributed across society.Footnote ²⁹ It becomes questionable in such a context to what extent the ballot box continues to be an accountability mechanism ‘that has any direct impact on the day-to-day operations of government’ (Reference RockRock, 2020: 26), while it is not certain that participatory accountability provides satisfactory opportunities for redress. In the case of infrastructure policies, for instance, stakeholder accountability gives an advantage to groups claiming to experience negative externalities that are geographically concentrated, with the risk that diffuse interests are neglected. In addition, those who complain about costs and possibly succeed in receiving compensation may come to enjoy benefits in the future that are less foreseeable in the present (Reference Jordana, Wegrich, Kostka and HammerschmidJordana, 2017). This raises important issues about social accountability being forward-looking, having procedures elaborate enough to consider the intertemporal dimension and perhaps needing to be particularly demanding for decisions that are difficult to reverse.

Such concerns can nevertheless be relativized if we keep in mind that downwards accountability to service users does not replace the hierarchical accountability of the bureaucracy upwards, grounded in the traditional delegation chain. Input from service users is primarily expected to provide more accurate information about performance to management and political superiors, who can process this kind of information to distinguish between organizations and employees that perform well and those that do not and who can use incentives (such as performance pay) to reward those that contribute to good performance. Hence, a crucial question is how accountability downwards has impacted control from the top, but the answer is not straightforward. On the one hand, it may be asserted that bureaucracies have gained autonomy in their day-to-day operations to better adjust to the demands of service users: it has been frequently argued that politicians should ‘steer’ but not ‘row’ and should let managers manage. On the other hand, apart from the fact that the focus on performance also empowers auditing institutions as accountability forums, politicians can use performance evaluations for their own strategic goals. Moreover, the formal separation of politics and administration makes it easier for politicians to blame managerial failure for their own errors, especially as managers lose anonymity in a context of increased media attention: the ‘agency strategy’ that is part of ‘blame games’ (Reference HoodHood, 2010).

However, it has also been argued that public managers are inclined to shift the blame upwards to their political principals because their loyalty to their superiors may be less strong than concerns about their own reputations (Reference MoynihanMoynihan, 2012). In addition, managers’ discretion is not necessarily impacted by the risk of a negativity bias. For example, even in situations in which the media and political agendas reinforce each other by focusing on emotionally loaded aspects of policy-making related to public safety issues, the regulatory agenda remains insensitive, focusing on factual information and following its own course (Reference Opperhuizen, Klijn and SchoutenOpperhuizen, Klijn and Schouten, 2020). Moreover, managers’ vulnerability is reduced by the fact that they can game the system. Although ‘measurementitis’ has often involved enormous costs – the growth of an industry of specialists – it frequently has not lived up to expectations, as Reference Pollitt, Dubnick and FredericksonPollitt (2015) showed about the UK National Health Service. Bureaucrats can resort to presentational strategies and stage management, conspicuously displaying signs of conformity to the proclaimed goals that do not reflect actual performance.

These unexpected effects of disaggregation have not remained unnoticed: they made the governmental centre ‘strike back’Footnote ³⁰ and led to new post-NPM waves of reforms, mainly intended to remedy coordination deficits, such as the ‘joined-up’ or ‘whole‐of‐government’ initiatives. New reforms often did not replace previous ones but were added to them in a layering process that followed a piecemeal logic. This harmed the clarity of responsibility because the successive reforms have been associated with different, if not contradictory, goals. The correct operation of accountability mechanisms presupposes that politicians can easily decode managers’ conduct, whereas confusing goal statements obviously complicate the accomplishment of this task (even though produced by politicians themselves!). It suffices to consider that it is very difficult to reconcile the objectives of empowering service users, freeing managers and optimizing political control (Reference Lægreid, Bovens, Goodin and SchillemansLægreid, 2014).

4.3 Network Governance and Accountability

Organizational complexity did not grow only due to the layering of reforms pursuing sometimes contradictory objectives. It also grew because reform endeavours, apart from seeking to delegate operational tasks to administrative segments closer to the frontline, consisted of the horizontal disaggregation of service provision, with public‒private partnerships and outsourcing, and in the delegation of regulatory missions to independent agencies.

The policy process is often interactive, polycentric and unavoidably messy, with policy complexity mirroring and responding to societal and problem complexity. Political decisions are formulated or implemented through bargaining or deliberation between actors who ‘coopete’ in diffuse policy networks, involving not only politicians and administrators but also interest representatives, stakeholders and experts (independent or members of advocacy coalitions). Therefore, governance becomes less public, less vertical and more interdependent (Reference Willems, Van Dooren, Christensen and Lægreid Willems and Van Dooren, 2017). ‘Coproduction’ or ‘collaborative governance’ are the usual descriptions of such policy-making modes, referring to collaboration with non-governmental actors, vertical intergovernmental cooperation across jurisdictional layers or a combination of both (Reference Ansell and GashAnsell and Gash, 2008; Reference Emerson, Nabatchi and BaloghEmerson, Nabatchi and Balogh, 2012).Footnote ³¹ In the face of wicked problems – characterized by unclear or controversial definitions, interdependent and uncertain causal chains, multiple stakeholders with competing values and interests, and unpredictable consequencesFootnote ³² – there are undeniable practical advantages of collaborative governance with regard to policy effectiveness and acceptance. Sometimes, collaborative modes have been adopted to remedy the flaws of purely managerialist governance, and in many cases, collaborative governance also has normative value with respect to pluralism and inclusiveness. However, from a democratic perspective, it is problematic if elected officials are unevenly engaged in collaborative policy-making arenas, as shown by a comparative study of a large number of cases across countries and jurisdictional levels (Reference Sørensen, Hendriks, Hertting and EdelenbosSørensen et al., 2020). The most influential actors may not be officially authorized to make collectively binding decisions: they may not be visible, possibly leading to errors in the allocation of responsibilities, or may be unelected, so that the wielding of power is divorced from democratic accountability.

The movement towards collaborative governance is prominently manifested in the advent of various modes of cooperation in public service delivery. Reforms have often led to the replacement of direct administration by interorganizational arrangements and contractual relationships with third parties: these are suppliers of services that are more or less external to the public bureaucracy, such as hybrid bodies, for-profit and non-profit organizations, to which public policy tasks are outsourced or with which they are conducted in partnership. Such configurations are challenging for the exercise of accountability because the strict lines of command and control are blurred and sometimes disrupted (Reference Nguyen, Rawat, Morris, Sullivan, Dickinson and HendersonNguyen, Rawat and Morris, 2020; Reference Triantafillou and HansenTriantafillou and Hansen, 2022). It becomes necessary in such a context ‘to ask whether private contractors should be treated as part of the government apparatus for the purposes of holding them accountable for the exercise of public power’ (Reference RockRock, 2020: 19). More control may be suitable, as it is plausible that the interests of external suppliers and their staff’s values differ from those of public organizations. At the same time, contract monitoring is complicated. It becomes difficult to resolve conflicts through hierarchical channels, and the foreseen sanctions are difficult to implement (Reference GirthGirth, 2014). What used to be internal bureaucratic disputes become externalized, possibly leading to litigation, so this is yet another driving factor of the empowerment of courts as accountability forums, as discussed in Section 3.

Research on hybrid entities endowed with public-service tasks reveals that such entities adhere to a plurality of normative frames (Reference Denis, Ferlie and GestelDenis, Ferlie and Gestel, 2015; Reference Skelcher and SmithSkelcher and Smith, 2015) so that their accountability obligations combine public with market, vertical with horizontal and formal with non-mandatory elements.

The case of market accountability, which has the particularity of being based on the ‘exit’ option instead of ‘voice’ and deliberation (Reference MulganMulgan, 2000b), is instructive. On the one hand, it is not without merit: in some countries, organizations, including public entities such as hospitals and schools, de facto compete with each other for service provision, and consumer sovereignty imposes discipline on providers and motivates them to meet performance targets (Reference Peters, Bovens, Goodin and SchillemansPeters, 2014: 219). On the other hand, it has been observed that clients use the ‘exit’ option as a threat to force staff to break professional rules (Reference Klenk, Cohen and HupeKlenk and Cohen, 2019). Hence, staff members face dilemmas and feel that they serve many masters. In particular, for-profit service providers must ‘take into account customers’ demands in order to survive in an economically competitive environment’ (Reference Sager, Thomann, Hupe, Sullivan, Dickinson and HendersonSager, Thomann and Hupe, 2020: 810). For example, competing certifiers face strong pressure from regulatees if they are simultaneously their source of income, and regulatees may opt for the most cost-effective among the competing certifiers, leading to a decline of regulatory standards (Reference Thomann, Hupe and SagerThomann, Hupe and Sager, 2018).

Finally, collaborative arrangements in general, and even more so informal policy networks that lack codification and visibility, are particularly prone to the ‘many hands’ problem of shared responsibility (Reference ThompsonThompson, 1980). Mutual accountability may be developed between the multiple actors that participate in governance networks, and it does have epistemic value if these actors interact densely, share information and deliberate. We may be in the presence of a ‘tangled web’ of accountability ties that link actors together (Reference Romzek, LeRoux and BlackmarRomzek, LeRoux and Blackmar, 2012), some of them formal, such as contracts, and others informal, related to norms of reciprocal obligations developed in interpersonal dynamics and to the ensuing unofficial rewards and sanctions. However, although the social pressure to comply with expectations from peers can be a powerful regulator of actors’ behaviour, this form of ‘community’ accountability (Reference Koliba, Mills and ZiaKoliba, Mills and Zia, 2011: 214) is not tantamount and may even be detrimental to public accountability (Reference PapadopoulosPapadopoulos, 2007: 480–3). The exercise of control may prove challenging for outside audiences anyway if they are confronted with wicked issues characterized by complex and uncertain causalities. This exercise is made even more difficult if such issues are addressed by policy networks operating without clear rules (Reference Cristofoli, Douglas, Torfing and TrivellatoCristofoli et al., 2022), regardless of the merits of fuzzy governance structures with regard to creativity and flexibility. The same happens when the most relevant negotiations take place behind closed doors, while public meetings are staged performances with a primarily symbolic function.Footnote ³³ Outsiders may not be able to assess who is responsible for success or failure and how much each actor has contributed to outcomes, so that responsibility may not be allocated in a way that does justice to each participant’s contribution.Footnote ³⁴ In view of such risks, there have even been claims that – similar to the rise of independent agencies – the advent of collaborative forms of governance results from a deliberate strategy of depoliticization that is expected to facilitate evading accountability.Footnote ³⁵

4.4 ‘Agencification’ and Accountability

Another major trend is the delegation of tasks to independent agencies that have been entrusted with considerable authority, most notably over economic and risk regulation but also more generally over service delivery and policy implementation. In particular, regulatory agencies – such as competition authorities or those in charge of the utilities sector – become crucial players with strong law-making roles in their area of competence (Reference Carpenter and KrauseCarpenter and Krause, 2015), so that their growth has been portrayed as ‘the rise of the unelected’ (Reference VibertVibert, 2007). The influence of NPM doctrines is visible in several agency characteristics, such as the organizational disaggregation of centralized bureaucracies into specialized units that are considered to be more efficient, more ex ante operational flexibility for agency managers given their familiarity with the sector of operations, and performance contracting with the ensuing emphasis on ex post accountability for results (Reference Verhoest, Roness, Verschuere, Rubecksen and MacCarthaighVerhoest et al., 2010).

With ‘agencification’, political and epistemic credibility are so prioritized – and politicians considered to lack both – that large agency autonomy is considered necessary, following the model of independent central banks in charge of monetary policy. Thus, agencies become ‘trustees’ rather than mere delegates as envisaged by the principal–agent model: they enjoy ‘fiduciary’ competences, and their independence appears legitimate (Reference MajoneMajone, 2001). Much of the reputation of independent agencies as contributors to the common good is supposed to derive precisely from their professionalism. They are thereby considered more likely to safeguard the public interest than self-interested politicians: according to this narrative, collective problem solving requires ‘consistency, long-sightedness and rationality’ (Reference Eriksen and EriksenEriksen, 2022: 20), while politicians primarily seek re-election and are consequently tempted to prioritize short-term considerations and constituency service. Ideally, agencies should be shielded from the intrusion of politics (Reference Miller and WhitfordMiller and Whitford, 2016). However, political reality can be more gloomy: legal provisions do not suffice to guarantee the actual independence of agencies (Reference Hanretty and KoopHanretty and Koop, 2013); the latter can be captured by special interests;Footnote ³⁶ and the ‘agencification’ process may primarily serve to evade ministerial responsibility with agencies operating as magnets that attract blame in case of misfortune (Reference HoodHood, 2010).Footnote ³⁷

In practice, agencies are highly diverse in the kinds of tasks that they carry out, their architecture and their degree of decoupling from the traditional administrative hierarchy. Similar to other managerial reforms, agencification is subject to variation across nations and sectors. In a path-dependent logic, organizational creations are often ‘acclimatized’ to previously existing politico-administrative arrangements so that one encounters multiple ‘implementation habitats’ (Reference Verhoest, Roness, Verschuere, Rubecksen and MacCarthaighVerhoest et al., 2010: 4). Not only the autonomy but also the accountability of agencies and their chief executives are context dependent.Footnote ³⁸

In general, the doctrine welcomes a shift from direct accountability to political superiors to more plural forms of accountability, especially for agencies endowed with regulatory tasks that should not be vulnerable to hierarchical pressure likely to undermine their autonomy. However, some authors refer to an ‘autonomization paradox’: autonomy is frequently accompanied by more stringent controls, so that agencies perceive themselves as being more controlled than before, especially if they manage substantial financial resources or deal with salient topics such as food safety or finance regulation (Reference Verhoest, Roness, Verschuere, Rubecksen and MacCarthaighVerhoest et al., 2010: 263). There are multiple ways in which political principals continue to control agencies ex post and influence their operation ex ante, for example, through political appointments, despite formal hierarchical subordination being loosened. Political authorities continue to be primus inter pares in their role of accountability forums, and the extent of the shift from hierarchical accountability may have been overestimated (Reference Hill and LynnHill and Lynn, 2005).

Overall, the correlation between political independence and public accountability is relatively weak, as evidenced by a large-N survey including 799 regulatory agencies in 115 countries and 17 policy sectors (Reference Jordana, Fernández‐i‐Marín and BianculliJordana, Fernández‐i‐Marín and Bianculli, 2018). At the same time, agencies, as typical examples of output-oriented organizations, need to justify their choices and convince various audiences about the achievement of their targets. Formally, they are subject to managerial surveillance by agency boards, financial surveillance by auditing institutions and legal surveillance by courts. Accountability to multiple forums is not without its challenges: there is a risk of redundancies that involve costs (Reference SchillemansSchillemans, 2010) but also of coordination deficits leading to conflicting claims that cause dilemmas (Reference Schillemans, Overman and FawcettSchillemans et al., 2021a). In the European Union, for example, national agencies become part of a multilevel administrative space through their participation in EU-wide rule-enforcing and coordination networks. This amplifies centrifugal trends within national executives (Reference Bach, Ruffing and YesilkagitBach, Ruffing and Yesilkagit, 2015) because national agencies operate at arm’s length from governments and tend to become ‘double-hatted’ by developing loyalties with respect to EU institutions (Reference Egeberg and TrondalEgeberg and Trondal, 2009). The ‘many eyes’ problem, which may undermine the effectiveness of individual agencies’ accountability, combines with the ‘many hands’ problem. In such a shared administrative space, responsibility is diluted, and informational asymmetries that obstruct public scrutiny are particularly difficult to overcome in multilevel settings (Reference Brandsma, Moser, Scholten, Brenninkmeijer and StraussBrandsma and Moser, 2020: 70–2).

Agencies also need to provide justifications to stakeholders (such as firms from regulated sectors or consumer associations) and to the media, especially in the case of policy failure on hot issues: in other words, to forums that have no direct sanctioning power but whose support is necessary for their legitimacy. Reference TuckerTucker (2018: 259) wrote that ‘enjoying insulation from day-to-day political pressures does not shield independent agency leaders from debate and challenge of various kinds’ and called this ‘discursive’ accountability. However, even though reason-giving and deliberation are beneficial to accountability, when the targeted forums are composed of actors other than the democratic principals of the delegation chain (citizens, members of parliament and executives), this cannot be considered a functional equivalent of democratic control. For example, in the case of Dutch executive agencies, accountability towards stakeholders contributes to organizational learning but does not truly enhance democratic control (Reference SchillemansSchillemans, 2011), and critical scrutiny is impeded by the propensity of citizen councils to sympathize and gradually identify with the agency (Reference BrummelBrummel, 2021: 1063–4). Moreover, and similar to the problem addressed above about the accountability of public organizations to stakeholders in general, the participation of those affected by agencies’ activity in their scrutiny runs the risk of being limited to organized minorities that enjoy concentrated benefits or suffer concentrated costs from agencies’ decisions (such as the producers or consumers of particular goods and services). Furthermore, power asymmetries at this level drive agencies to adjust their accountability practices to them (Reference Apaydin and JordanaApaydin and Jordana, 2020).

Finally, the latest research revealed two interesting trends that should be treated together, as they plausibly balance each other: the accountability activity of agencies is more intense than expected, and the monitoring activity of forums is lower than expected. First, agency management may opt to engage proactively in accountability activities. Depending on the type of audience, this may happen because managers believe they have a moral obligation to be held accountable or because they consider this useful for learning purposes, but management can also be animated by a consequentialist logic (Reference de Boerde Boer, 2022). Thus, agency managers cultivate a calculated use of voluntary forms of accountability for strategic purposes as part of reputation management, credit claiming and the creation of constituency support (Reference Busuioc and LodgeBusuioc and Lodge, 2016, Reference Busuioc and Lodge2017; Reference KarstenKarsten, 2015; Reference KoopKoop, 2014).

Actors not held to account de jure may have good reasons to seek to become accountable de facto – through ceremonials of open public communication but also through secretive exchanges with selected target audiences behind closed doors – because accountability may not just be a constraint but also a resource. Proactive accountability contributes to a good reputation, which is in turn a firewall that shields from criticism, including when reputation is undeserved.Footnote ³⁹ Staging ‘showcase accountability’ (Reference BrummelBrummel, 2021: 1065) discourages critical scrutiny: according to experimental evidence, ‘promotional symbols’ compensate for poor performance by shaping citizens’ attitudes (Reference Alon-Barkat and GiladAlon-Barkat and Gilad, 2017). Once an agency’s reputational authority is consolidated, it can serve as a sort of tranquillizer. If agencies come to enjoy support by influential ‘coalitions of esteem’ (Reference Bertelli and BusuiocBertelli and Busuioc, 2021), attempts at control risk high costs, as they may be considered illegitimate and intrusive, and this induces self-restraint on behalf of accountability forums. In such a view, restraint is dictated by a consequentialist logic, but it can also be dictated by a logic of appropriateness, and in particular by deference (Maggetti and Papadopoulos, 2022). Forums may find it misplaced to criticize an organization or an individual that is highly regarded by the forum itself and in all likelihood by other socially significant audiences.

Second, as mentioned, forums may not play their monitory role, which contradicts the predictions of the principal–agent model, according to which accountability gaps are caused by agency drift. The cause here is forum drift, which often relates to information gaps and paralysis due to goal conflict (Reference Benjamin and PosnerBenjamin and Posner, 2018) or to intentional inactivity because of limited time and attention and of other more pressing priorities (Reference Schillemans and BusuiocSchillemans and Busuioc, 2015). Policy salience seems to be necessary for forums to become active (Reference KoopKoop, 2011) and what counts perhaps even more is the existence of exceptional focusing events that damage the reputation of agencies by subjecting them to ‘emotionalized blame attribution’ (Reference Wood, Matthews, Overman and SchillemansWood et al., 2022). Now, if the problem is not so much of a failing agent but of a failing principal or forum, what happens to agency accountability to the broader public as a forum? Even parent ministries may not have the expertise to fully grasp the implications of the highly diverse and often technically, legally or operationally complex decisions made by agencies, or at least this is what agency managers tend to think (Reference Schillemans, Overman and FawcettSchillemans et al., 2021b). How many ordinary citizens even know about the existence of agencies as distinct organizational ‘species’? Such concerns echo the findings of election studies that challenge citizens’ capacity to hold elected officials accountable with retrospective voting (see Section 2). However, these concerns are even more serious in the case of agencies because several among them, and many bureaucratic organizations and policy networks more generally, are much less visible to the public, notwithstanding their influence on decision-making.

Overall, the developments in this section confirm that, in view of the political role of bureaucracy, addressing issues related to the accountability thereof is necessary to capture the way the governmental process works in practice. However, after decades of research on administrative reform, the image of such accountability is complex, if not at times contradictory. Reform processes have affected accountability relations in different directions, and the outcomes appear to be highly context-dependent and uncertain. Ultimately, it is indeed striking how much the picture of the accountability of administrative organizations and agents contrasts with the relative simplicity of the model of electoral accountability that connects elected politicians to the citizenry. The surprising thing in view of these considerations is that – at a macro level – the ‘thermostatic’ mechanism continues to operate in democratic systems. It triggers governmental responsiveness to the preferences of public opinion (as we saw in Section 2), although studies of electoral accountability do not capture the intricacies of the input, output and feedback processes leading to these adjustments.

5.1 Public Governance in Audience Democracy: Mediatization and Its Limits

In his book, Keane praised the virtues of monitory democracy, which puts governors under stress, but one should be cautious about the genuine democratic credentials of the accountability forums that thereby increase in importance. This applies not only to frequently technocratic guardian institutions but also to cause groups that do not necessarily reflect the public’s preferences. In ‘advocacy’ democracy (Reference Dalton, Cain, Scarrow, Cain, Dalton and ScarrowDalton, Cain and Scarrow, 2003), ‘the advantages of education, income, and other unequally distributed resources are more likely to translate into patterns of over- and underrepresentation’ (Reference Urbinati and WarrenUrbinati and Warren, 2008: 405). Furthermore, frequently neither is there an identifiable constituency that delegates ex ante its oversight role to surrogates, nor are the affected communities in a position to sanction them ex post if they are not satisfied with the way they fulfilled their tasks (this applies even more in the case of elusive reference groups such as ‘future generations’). It may even happen that those who are supposed to be represented are uninformed about the actions that surrogates claim to undertake in their favour, or they may ignore the sheer existence of the defenders of their cause.Footnote ⁴¹

A recent comparative study of governmental accountability portrayed the expansion and rising impact of various accountability forums in different types of regimes (Germany, Russia and Bulgaria) as an ‘accountability turn’ that takes place in the ‘media age’ (Reference DimovaDimova, 2020). Alongside the official watchdog institutions and other self-authorized monitory bodies, the media play a crucial role by acting as ‘fire alarms’. What precisely does this metaphor mean? In their famous study of congressional oversight over the bureaucracy in the USA, Reference McCubbins and SchwartzMcCubbins and Schwartz (1984) showed that, instead of dedicating considerable time and energy to actively controlling bureaucrats through ongoing scrutiny (‘police patrolling’), members of Congress rationally opt for intermittently relying on third parties that alert them and draw their attention to problematic situations. In other words, the surveillance of the fulfilment of de jure delegation mandates is de facto delegated to trusted external actors. Although the media have no formal competence to sanction, when they ‘bark’, they can spark reactions from audiences – such as voters or parliamentarians – that do have the power to activate sanctioning (‘biting’) mechanisms, confirming that accountability forums are resource interdependent.Footnote ⁴² Governing actors operate under the shadow of not only vetoes and formal sanctions but also media criticism; therefore, they also try to anticipate media reactions and to preempt negative coverage by strategically adjusting their modus operandi and sometimes by engaging more deeply in self-reflective processes (Reference Jacobs and SchillemansJacobs and Schillemans, 2016: 29).

Media’s rising influence on politics has been highlighted by Reference ManinBernard Manin (1997) with his emphasis on the shift of representative government from party to ‘audience’ democracy, in which politics is personalized and campaigning becomes permanent in a context of fluctuating and eroding partisan loyalties. Relatedly, political communication specialists refer to the mediatization of politics (Reference Esser and StrömbäckEsser and Strömbäck, 2014), which means political actors are more heavily dependent on media judgement than in the past, and the media is more independent from them.

Critical media are indispensable to a well-functioning democratic polity, and mediatization can be seen as a sign of politics becoming more subject to public scrutiny. However, things are more complicated: the effective accountability exercise requires that forums be provided with accurate information, but the information provided by the media system may be skewed even if the media is non-partisan. The media have their own agendas, and the filter of newsworthiness guides the media logic. This problem is particularly acute with the tabloid press, which, apart from simplifying reality and treating political matters superficially, adopts a predominantly populist style of reporting that consists of attributing blame to elites (Reference Hameleers, Bos and de VreeseHameleers, Bos and de Vreese, 2019). Commercial media overemphasize scandals that testify – according to them – to misconduct and lack of probity, frequently calling for redress that requires ‘heads to roll’. At the same time, media coverage may be of short duration, as the media may be inclined to jump over to the next newsworthy event (Reference Kuipers, ‘t Hart, Bovens, Goodin and SchillemansKuipers and ‘t Hart, 2014: 591). Investigative journalists search for blunders that signal lack of competence or negligence on the part of politicians or public administrators, and seldom do they demonstrate interest in policy success that tends to be seen as business as usual. This has led Reference FlindersFlinders (2012: chapter 6), a leading specialist in the study of accountability, to deplore the advent of a blame culture and to criticize the role of inquisitive and predatory media in fuelling a climate of undeserved suspicion vis-à-vis office holders and propagating cynical views of politics. However, overgeneralizations should be avoided, as the political communication literature concludes that the presence of deleterious media effects is contingent on the media system type (e.g., more or less commercialized, etc.) that prevails in each country (Reference PapadopoulosPapadopoulos, 2012).

Whatever the evaluation of media power, the media can now be seen as a ‘player in their own right’ (Reference GrossmanGrossman, 2022: 446) because the reputation that they have to establish good and bad reputations is something politicians tend to include as a variable in their calculations and choices. This happens even though media power may be overestimated, considering that the public often does not count media among the most highly trusted actors. Politicians and parties compete to best meet what they perceive to be media expectations through presentational strategies of ‘self-mediatization’, which they deploy under the guidance of communication specialists. Moreover, we have seen in Section 4 that public managers can also be targeted by highly emotional blame attribution that causes ‘media stress’ (Reference Schillemans, Karlsen and KolltveitSchillemans, Karlsen and Kolltveit, 2019). They become public figures, sometimes reluctantly drawn into political debates by politicians eager to deflect the blame, but occasionally without the latter’s blessing.

Public sector organizations, which are increasingly under the spotlight, invest substantial resources into media management and public relations tasks (Reference SchillemansSchillemans, 2012), and managers even succeed in challenging politicians’ views and in developing their own persuasive narratives (Reference GrubeGrube, 2019). However, controlling media coverage is a difficult task, and even more so with the new media’s growth that makes accountability processes more personal and direct (Reference Kuipers, Brändström and ThompsonKuipers and Brändström, 2020: 16),Footnote ⁴³ despite the expanding industry of communication specialists that support officials in reaching out to the public through many channels. The information on social media may be pluralist. However, it is above all extremely abundant and unmediated and becomes easily viral in the absence of the traditional media’s gate-keeping role and fact-checking function, so that news management by governing actors is harder to achieve. In particular, the distortion of information is even more prominent in the arena of social media: all sorts of fake news widely circulate and receive credit in echo chambers, as people tend to interact with other people who hold similar beliefs, and confirmation bias makes them more receptive to messages that support and intensify their views. Monitoring is in a sense democratized – Reference Vanhommerig and KarréVanhommerig and Karré (2014) referred to the Internet as a low-cost medium easily accessible to ‘monitorial citizens and armchair auditors’ – but frequently at the expense of reflexivity and genuine deliberation.

Nonetheless, monitory democracy does face structural limitations despite the rise of internet campaigns and, therefore, is not as obtrusive as suggested (for better or worse, depending on one’s views). Media coverage is partial: television news, for example, operates with episodic framing; that is, it concentrates on focusing events, relates them to individuals, and narrates simple stories (Reference IyengarIyengar, 1991). Hence, mediatization does not penetrate the whole policy process, as shown by Reference Adam, Hurka, Knill and SteinebachAdam et al. (2019).

Mediatization encounters limits because part of the policy-making process unfolds backstage and is not visible to the media or because policy outputs are not suitable for media ‘dramaturgy’. One can think of policy measures that are technical, lack a tangible impact on people’s lives or are decided in expert forums, sometimes even (much) beyond the national level. Such policy escapes media attention – perhaps with the exception of some specialized outlets – because investing resources to report on routine and low salience issues is suboptimal in commercial terms (Reference HajerHajer, 2009: 176–7). In addition, most journalists lack the background to apprehend technical issues and to become familiar with cumbersome processes. Moreover, positive treatment by the media is frequently not the primary concern for bureaucrats who care about evidence-based choices and do not seek publicity, and special interests prefer to opt for ‘quiet politics’ instead of going public (Reference CulpepperCulpepper, 2010). Backstage policy-making does not necessarily result from an explicit intention to shield decisions from public scrutiny, even though deliberate depoliticization strategies may also be at work, and conditions of secrecy and informality may be necessary for bargaining. However, if what happens backstage matters for policy, then serious errors in the attribution of responsibilities may occur: a comparative study of public transport and economic promotion policies in eight western European metropolitan areas concluded that elected actors are more often presented as responsible for policies (‘over-responsibilised’) in newspaper reporting and more often blamed for policy failures than other actors (‘over-blamed’: Reference Hasler, Küebler and MarcinkowksiHasler, Küebler and Marcinkowksi, 2016). Such a decoupling between the political spectacle and the reality of policy development indicates a divorce between power wielding and public accountability (Reference PapadopoulosPapadopoulos, 2013). We have observed this with collaborative governance (Section 4), but are things maybe different in cases of conspicuous events such as crises?

5.2 Policy Failure: Crisis Management Strategies and Their Limits

As shown in the previous section, accountability dilemmas are one of multiple causes of operational failure. One of the most prominent examples thereof in the accountability literature is that of the explosion of the space shuttle Challenger in 1986, which has been explained in part by NASA’s efforts to manage diverse legitimate, albeit conflicting, pressures and expectations (Reference Romzek and DubnickRomzek and Dubnick, 1987).Footnote ⁴⁴ The management of natural disasters has also been associated with failures in accountability regimes, such as in the poorly coordinated and insufficient emergency responses and recovery actions in the aftermath of Hurricane Katrina in 2005.

Dealing with the same disaster, Reference MoynihanMoynihan (2012) uncovered that crisis response network members deployed blame avoidance strategies.Footnote ⁴⁵ Following this line of research, this section addresses accountability after perceived failure, regardless of whether this means failure to prevent a crisis or to manage its consequences properly. Does accountability work when it is, arguably, mostly needed? The answer is not straightforward: accountability processes can both unsettle and transform orders as well as maintain them (Reference OlsenOlsen, 2017).

Crisis situations can be defined as ‘highly negative and disruptive events or episodes characterized by threat, urgency, and uncertainty’ (Reference Kuipers, Brändström and ThompsonKuipers and Brändström, 2020: 3). Whatever the origin of events defined as critical (natural disaster, external attack or operational deficiencies),Footnote ⁴⁶ these events undermine public trust in the capability or probity of office-holders if public opinion comes to believe that they are responsible for the mishap or have not dealt adequately with its consequences (Reference Kuipers, ‘t Hart, Bovens, Goodin and SchillemansKuipers and ‘t Hart, 2014: 589). Elected executive members, and increasingly unelected top public managers, are in the spotlight and become an easy target whenever public opinion is dominated by the perception that something went wrong. Obviously, crises can be exploited politically: ‘crises need a face, and political executives provide those’ (Reference Kuipers, Brändström and ThompsonKuipers and Brändström, 2020: 10).Footnote ⁴⁷

Demands for explanations may be addressed by formal bodies, such as inspectorates or ad hoc investigation commissions, which rely on professional standards to identify – normally in a dispassionate manner – the possibly highly complex causal processes that led to failure and to draw lessons for the future. Office-holders can expect scrutiny from these bodies to allow them ‘to get out of the limelight’(Reference Kuipers, ‘t Hart, Bovens, Goodin and SchillemansKuipers and t’Hart, 2014: 596) and, ultimately, to appease protest. However, the independence and discretion that these bodies enjoy in their operations may be disturbing for office-holders, so that the relations with them ‘are full of ambiguity and latent tensions’ (Reference Kuipers, ‘t Hart, Bovens, Goodin and SchillemansKuipers and t’Hart, 2014: 592) and oscillate between cooperation and stonewalling. In addition, even if office holders have the tools to shape and exert control over formal accountability procedures, criticism by investigative bodies is considered more credible than positive appraisal, so that stigma cannot be easily avoided (Reference Kuipers, ‘t Hart, Bovens, Goodin and SchillemansKuipers and ‘t Hart, 2014: 597). Moreover, in audience democracy, the media act as catalysts (Reference Boin, McConnell, ‘t Hart, Boin, McConnell and ‘t HartBoin, McConnell and ‘t Hart, 2008: 21), frequently attracted by the newsworthiness of policy failure and – especially the tabloid press and commercial TV channels – animated by a negativity bias in their tone. Postcrisis accountability processes are usually particularly harsh, as account-holders such as opposition parties or victim organizations formulate highly demanding claims that require sustained efforts in terms of justification and manifest an appetite for drastic sanctions.Footnote ⁴⁸ As disasters are judged intolerable, crises are test cases for the accountability of policy-makers (Reference Boin, ‘t Hart, Stern and SundeliusBoin et al., 2005: 124).

On the one hand, public authorities are exposed to stress caused by an upsurge of accountability demands and sometimes to a spiral of escalating criticism so that they are in a great need to calm protests.Footnote ⁴⁹ On the other hand, they are not powerless: postfailure strategic behaviour ‘is a pivotal task of crisis leadership’ (Reference Boin, ‘t Hart, Stern and SundeliusBoin et al., 2005: 103) and consists of coping creatively with pressing claims for transparency, explanations and possibly sanctions. Crises are extraordinary, often complex and dynamic phenomena, so that ambiguity and uncertainty leave space for conflicts of interpretation: facts are disputed, the causalities of events are difficult to disentangle, the division of tasks between ‘principals’ and ‘agents’ is challenged and the pertinence of evaluative standards – perhaps even that of the existing accountability routines – is contested (Reference OlsenOlsen, 2017). Therefore, one needs to win the battle of meaning-making by imposing one’s definition of the situation in the contests between frames and counterframes regarding the accidental or structural origin of crises, as well as their severity and their implications (Reference Boin, McConnell, ‘t Hart, Boin, McConnell and ‘t HartBoin, ‘t Hart and McConnell, 2008). For instance, it may be unclear or controversial if and to what extent public authorities must be blamed for the casualties, and proof of that responsibility may require time-consuming investigative procedures and even court rulings. As even the functioning of independent public inquiries and their conclusions may be shaped by postcrisis politics, there is ample room (and need) for increasingly professional impression management. Presentational strategies extend from dramatization, escalation and overresponsibilization – by the opposition, critical media and cause groups – to problem denial and minimization with the use of ‘defensive scripts’ (Reference Kuipers, ‘t Hart, Bovens, Goodin and SchillemansKuipers and ‘t Hart, 2014: 593) by those accused of responsibility for failure. Does blame avoidance then help win the accountability battle?Footnote ⁵⁰

Reference Hood, Dubnick and FredericksonChristopher Hood (2015) first made the explicit connection between the literature on public accountability and the literature on blame avoidance. The latter posits that it is crucial for politicians and public managers to manoeuvre to shield themselves and reduce reputational damage. If blame cannot be proactively avoided,Footnote ⁵¹ it may be worth seeking occasions to pass the buck, sacrifice scapegoats, or diffuse the blame and ‘circle the wagons’ (Reference WeaverWeaver, 1986: 388–9). The literature provides nuanced conclusions on the capacity to evade accountability by deflecting blame. For example, the intense and negative media scrutiny that characterizes postcrisis situations correlates with cases of resignation, but politicians with more experience, higher precrisis credibility and a thorough understanding of the rules of the game tend to be better shielded (see Reference Kuipers, Brändström and ThompsonKuipers and Brändström, 2020: 13–15).

Unclear causal relations and ambiguous allocation of tasks are theoretically prone to blame-shift strategies, but comparative research tends to show that the outcome of such strategies is uncertain and context dependent. Hood, Reference Hood, Jennings and CopelandJennings and Copeland (2016) compared how heads of government in the USA, the UK, Ireland and Canada behaved when they faced media blame for their personal conduct. Although the authors found evidence that office-holders admit responsibility only after problem or responsibility denial are exhausted as presentational strategies (‘staged retreat’), they doubt the effectiveness of blame management in dampening the firestorm. In his study of policy controversies that gave birth to blame games in the UK, Germany and Switzerland, Reference HinterleitnerHinterleitner (2020) distinguished between ‘high stakes’ and ‘below the radar’ blame games leading to different outcomes. He found that the public salience of issues and the institutional terrain, including informal and implicit game rules, shape the space in which conflict management occurs by providing blaming gateways and barriers against blame. In the British political system, for example, restrictive conventions about resignation (and more generally responsibility attribution) ‘clearly benefit incumbents during blame games’ with limited incentives for the government to amend failed policies; in contrast, in Germany, extensive conventions of resignation are, among others, ‘conducive to creating a rather aggressive blame game that centers on political incumbents’ (Reference HinterleitnerHinterleitner, 2020: 186). This stands at odds with the conventional knowledge about the conflictual Westminster-style bipartisan politics in the United Kingdom and the more accommodative style of German politics related to coalition government and institutional fragmentation.

Needless to say, the existence of a delegation chain can help deflect blame towards ‘lightning rods’ and thus offers policy-makers a ‘protective coating’ (Reference Boin, ‘t Hart, Stern and SundeliusBoin et al., 2005: 118), as in the United Kingdom, where ‘media attention and opponent attacks overwhelmingly focus on administrative actors and entities’ due to relatively low direct governmental involvement (Reference HinterleitnerHinterleitner, 2020: 186). Blame-shifting frames are also more credible in situations in which multiple jurisdictional levels (Reference Maestas, Atkeson, Croom and BryantMaestas et al., 2008) or networks of actors (Reference MoynihanMoynihan, 2012) are involved, often characterized by lack of responsibility clarity. However, a blame shift may not suffice, and in the end, it may be necessary to admit responsibility for the mishaps and to culminate in a restorative rhetoric (Reference Boin, ‘t Hart, Stern and SundeliusBoin et al., 2005: 123). Interestingly, the most effective strategy, according to a meta-study, is self-disclosure of failures, apologies and corrective action, but taking such steps is much more painful than denying the problem. Hence, denial of responsibility is the framing that is most commonly used in response to threats and criticism, even though it is the least effective one (see Reference Kuipers, Brändström and ThompsonKuipers and Brändström, 2020: 12)

The analysis should also include the long-term consequences of postcrisis accountability, but it is difficult to anticipate if a crisis will be ‘fast-burning’ with a return to ‘business as usual’ or if there will be ‘a crisis after the crisis’. This depends, for instance, upon how quickly public attention will move on to other concerns, and timing matters more generally if we think, for example, of the tighter or looser connection with the electoral cycle (Reference Boin, McConnell, ‘t Hart, Boin, McConnell and ‘t HartBoin, McConnell and ‘t Hart, 2008). According to some literature, the ‘long-shadow’ crisis, in which ‘turning the page’ is impossible, is the norm and elite ‘reinvigoration’ the exception (Reference Boin, ‘t Hart, Stern and SundeliusBoin et al., 2005: 104–5; Reference Boin, ‘t Hart, McConnell, Boin, McConnell and ‘t HartBoin, ‘t Hart and McConnell, 2008: 289, 294).Footnote ⁵² Crises seem to be critical junctures that provide opportunity windows and fertile ground for the claims of opposition actors and reformist advocacy groups: indeed, ‘crises can make, break, or transform political and public service careers, agency mandates and reputations, and policy paradigms’ (Reference Kuipers, ‘t Hart, Bovens, Goodin and SchillemansKuipers and ‘t Hart, 2014: 590).

However, even authors who highlight the enduring effects of crises are nuanced. Although cases of overreaction exist (Reference MaorMaor, 2012), postcrisis action can be restricted to symbolic manipulation and paper exercises, and crises do not necessarily lead to substantial policy adjustments because a return to normalcy may be strongly desired. The need to restore the broken trust in governance capacity does not always translate into genuine learning and into the policy shifts that would be required to avoid repeated misfortune, precisely because of the political nature of the postcrisis phase (Reference Boin, ‘t Hart, McConnell, Boin, McConnell and ‘t HartBoin, ‘t Hart and McConnell, 2008).Footnote ⁵³ Postcrisis change is often incremental rather than radical (Reference Boin, ‘t Hart and McConnellBoin, ‘t Hart and McConnell, 2009): for example, studying various disasters in the United States (e.g., homeland security in the case of September 11, natural catastrophes such as Hurricane Katrina, etc.), Reference BirklandBirkland (2006) found that, for critical events to be conducive to genuine policy change, several conditions are necessary, such as event-induced media attention, issue salience for a large portion of the public, the existence of advocacy groups and the preexistence of policy ideas that can provide inspiration for reform.

Therefore, it is reasonable to subscribe to the prudent observation that accountability after crises is a process that ‘leaves us with inconclusive answers to pivotal questions concerning “what works when, and for whom”’ (Reference Kuipers, ‘t Hart, Bovens, Goodin and SchillemansKuipers and ‘t Hart, 2014: 600). Explanations must take different interacting factors into account, namely, the nature of the crisis, the institutional and political contexts and, of course, actors’ strategies. In addition, one should distinguish among the different outcomes of accountability processes. Competing agendas may coexist, causing goal conflict: for example, this can occur when a forward-looking investigatory body concerned with truth-finding seeks to analyse the causes of accidents to improve future operations, while backward-looking commercial media primarily look for culprits who will make headlines. Sanctioning office-holders is different from triggering policy change, and both may be unrelated; policy change will not have the same scope if it stems from genuine learning from failures or from superficial strategic adjustment.

In this section, emphasis was initially placed on the fact that governmental activity is put under the ‘public microscope’ (Reference Kuipers, ‘t Hart, Bovens, Goodin and SchillemansKuipers and ‘t Hart, 2014: 592) in the context of audience or monitory democracy. Some see this as a virtuous process, but others deplore the perverse effects resulting from the negativity bias that prevails among some accountability forums, such as commercial media or opposition parties. However, many miss that low salience backstage policy-making activities are shielded from public scrutiny, so the trend towards ‘hyper-crutiny’ encounters limits. In addition, even following crises or failures, policy-makers seek to control accountability processes and to deflect the blame. The success of such strategies is not ensured, which is not bad news for the role of public accountability. At the same time, some policy-makers may be under- or over-responsibilized so that the divorce between power and accountability continues to exist. As it is not certain that accountability processes trigger policy improvements, the overall contribution of public accountability to good governance in the context of audience and monitory democracy is ambivalent.

7.1 Lessons Learnt on Power and Accountability

Following the introductory section, the point of departure for Section 2 on elections was the crucial role thereof as the core mechanism of democratic accountability in representative government and, therefore, as a major legitimizing device of democratic systems. Although such a function rests on solid theoretical grounds, is it also corroborated by empirical tests? Overall, competitive elections do not perform badly as the foremost channel of democratic accountability; however, the evidence is mixed. There is no agreement on whether we can consider voters as sufficiently well-equipped (or willing) to evaluate governments’ past performance and subsequently reward or sanction them. For example, voters use heuristics to help them in their evaluative tasks, but it is uncertain if such shortcuts may be misleading. The evidence is also inconclusive regarding the institutional setup’s impact on voters’ capacity, including factors such as electoral and party systems, the composition of the executive, a divided government or the vertical separation of tasks across jurisdictions. It seems that electoral accountability works best when power is concentrated because this is associated with more clarity of responsibility. On the one hand, it may be embarrassing if it is difficult to reconcile democratic accountability with a pluralist power structure. On the other hand, it is reassuring if electoral accountability is an effective counterweight when power is concentrated (Reference OlsenOlsen, 2017: 24).

We can also be reassured by the fact that the ‘thermostatic’ model seems to work in democracies. This happens most likely because governments fear (whether correctly or erroneously) the electoral sanction, are therefore incentivized to be responsive to citizens’ preferences and produce policies flexible enough to adjust for opinion shifts. However, the study of electoral accountability remains relatively superficial because it is limited to the official office-holders’ accountability, regardless of their actual degree of responsibility for policy outcomes, and overlooks the role of other key players, which requires a shift in focus from elections to opening up the black box of policy-making processes.

Doctrines of power separation impose institutional checks on the exercise thereof, which should bring about (self)restraint and thus prevent abuses by office-holders. Hence, Section 3 studies interinstitutional accountability: the vertical accountability of the executive and its members to parliament and the horizontal accountability between institutions that are not in a delegation relationship and are relatively equal in terms of power. Although there is no agreement on whether current trends in governance point towards a concentration of power in the hands of the executive or towards its fragmentation in polycentric networks of actors, it can be expected that both trends negatively impact – albeit for different reasons – the capacity of parliaments to control the executive. ‘Deparliamentarization’ refers to the declining role of parliaments, including their executive oversight function: much like individual citizens, parliamentary institutions may not be able, or even willing, to perform their role as accountability forums, so accountability deficits may not be due to agents evading blame but rather to forum passivity. Some parliaments have reacted to their loss of power, but the countertrend is not uniform. Even when parliaments ‘fight back’, they tend to hold the government accountable informally, with procedures of confidential accountability leading to interinstitutional bargaining that lacks transparency and undermines accountability to the public.

Equally worth mentioning is the growing relevance of rulers’ horizontal accountability to unelected forums, whose coupling with electoral accountability is not without frictions. Horizontal accountability may tame electoral accountability by design, as unelected forums derive their legitimacy from their independence from political interests. Moreover, they can convert their control power into influence on policy: when an institution is horizontally accountable to another institution whose consent is needed, it is incentivized to absorb the preferences of the latter, just as office-holders are incentivized to be responsive to citizens in vertical accountability. This happens most notably with the preferences of courts, whose role as accountability forums has substantially increased with the phenomenon of ‘judicialization’. The empowerment of judges is a core dimension of the ‘rise of the unelected’, and adjudication draws its legitimacy from the judicial process having a reputation of being depoliticized, impartial and fair. The ‘law of anticipated reactions’ is also at work when executives and legislatures become accountable to the courts: they seek to reach decisions that are robust to court rulings without being too remote from their own initial preferences. Although judicialization is driven by concerns about unrestricted majority power and the need to protect fundamental rights, it remains controversial. Judges cannot claim to be representative, and they are unaccountable by design to safeguard their independence. For some, this is a major challenge to the legitimacy of courts as accountability forums that gain influence in the policy proces. However, elected politicians are not powerless in the face of judicial activism; therefore, judges may also find it rational to internalize their preferences, and ‘government by judges’ is not unrestricted either.

Section 4 zooms into another category of key policy actors: the members of the bureaucracy. Political control over the bureaucracy seems to have become more sophisticated; however, it is not without limits, costs and collateral effects. Although upwards hierarchical accountability persists, it is difficult to say if control from the top is more effective now. Successive waves of reforms have sought to make it smarter, but they differ in their content depending on national trajectories and administrative styles. In addition, there is no clear indication that the stronger emphasis on accountability to target populations has produced the expected gains in performance and, indirectly, ‘output legitimacy’. This is hardly surprising considering that administrative reforms have often been animated by conflicting goals, rendering trade-offs unavoidable: ultimately, the empowerment of service users or ‘clients’ has been relative, and it becomes more difficult with the recent development of algorithmic decision-making that is a black box for outsiders. Their empowerment also remains controversial from a normative point of view, so it cannot be asserted that more participatory forms of accountability lead to genuine gains in terms of democratic policy-making.

In addition, collaborative forms of governance and the outsourcing or ‘coproduction’ of public services have grown in advanced democracies. On the one hand, such policy-making styles present epistemic advantages, especially in the face of wicked problems; they may enhance policy acceptance and can be valued normatively for being pluralist and inclusive. On the other hand, the issue of appropriate checks to political power wielded through hybrid governance modes and non-public actors is a real one. From a democratic point of view, it is problematic when elected politicians are not strongly engaged in collaborative policy-making arenas and unelected actors are counted among the most powerful. The exercise of power is then divorced from democratic accountability, and when the key players are unseen, the correct allocation of responsibilities is hampered. When policy-making is conducted through amorphous or informal channels, this leads to the ‘many hands’ problem, impeding public scrutiny. Moreover, even control by professional politicians becomes difficult when command lines are blurred and operators are dissociated from the hierarchical circuit. Furthermore, some participants in networks face multiple expectations from accountability forums with conflicting preferences, and the simultaneous presence of ‘many eyes’ requires developing coping strategies to deal with dilemmas: ‘360-degree feedback’ (Reference BehnBehn, 2001) may lead to fuzziness and unpredictability. This kind of cross-pressure may thus cause damage to the public interest and be detrimental to democratic accountability.

A different facet of shifts in governance – and another dimension of the ‘rise of the unelected’ – is the diffusion of the agency phenomenon. The legitimacy of agencies is grounded not just on their expertise but also, similar to courts, on their impartiality, and their independence is (at least rhetorically) highly valued. In practice, the relationship between independence and accountability is not straightforward. Although the doctrine does not consider them incompatible and may even advocate their combination, the picture is more complex, and it is sufficient to note that accountability may undermine independence and impartiality through the shadow of sanctions. In reality, agencies are accountable to different kinds of forums, but when their accountability web becomes too crowded, this can create problems of duplication or cross-pressure. Furthermore, effective democratic control is not guaranteed because of a lack of public awareness about the role of agencies. However, agencies should not necessarily be incriminated for accountability gaps: they engage in accountability activities more actively than expected (namely, because they care about their reputations), but accountability forums are more passive than anticipated (among other reasons, because they trust agencies’ reputations).

Knowing the role that media play as ‘fire alarms’, does accountability become stronger with the mediatization of politics and the advent of ‘monitory’ and ‘audience’ democracy, the effects of which are studied in Section 5? Governing actors – elected officials and increasingly top public managers – operate under the shadow of media criticism, and they seek to avoid or counteract the frequently negative bias in media coverage with presentational strategies of ‘self-mediatization’. Of course, news management becomes more difficult with the growth of social media platforms, on which unchecked information easily becomes viral. This does not mean that monitory democracy faces no limitations. It is mainly front-stage politics that lends itself to media ‘dramaturgy’ and is put under the public ‘microscope’, so mediatization does not penetrate the whole policy process. Policy issues of low salience and measures that are too technical or lack a broad or tangible impact on people’s lives do not score high in terms of newsworthiness, and many policy debates take place in narrow expert circles, and sometimes in remote arenas. Thus, they are not visible to the media who fail to perform their watchdog role. In other words, what frequently and increasingly occurs is a decoupling between the political spectacle and the reality of policy development, which may lead to errors in the attribution of responsibility. As mentioned previously, the consequence is a divorce between power wielding and public accountability.

It can be objected that this may be true for routine policy-making but not for critical situations: when perceptions that things went wrong dominate, policy-makers face more pressing demands for justification from official investigative bodies and strong claims for sanctions from opposition and stakeholder groups, with the media acting as catalysts. As disasters are difficult to accept, policy failure is an easy target for political exploitation, and postcrisis accountability may be harsh. Policy-makers need to react quickly and creatively to manage stressful situations and reassure public opinion. They are far from being powerless in their attempts to win battles of narratives by seeking to deflect the blame to ‘lightning rods’, so there is no guarantee that the gap between power and accountability decreases even in exceptional situations. However, policy-makers do not necessarily opt for strategies that are most efficacious for their survival, and the outcome of their attempts is uncertain and context dependent. The same can be said about the long-term consequences of postcrisis accountability regarding sanctions for failure and especially regarding the willingness and ability of policy-makers to learn from mishaps.

One finally needs to acknowledge that individuals’ subjective perceptions and experiences of accountability pressure matter (Section 6). In particular, how actors react to accountability overloads (i.e., to pressure that they perceive as excessive) is a matter of debate. Advocates of principal–agent theory stress the risk of accountability leading to bad policy out of conformism but, at the same time, envisage that accountable actors can just simulate conformity by gaming the system. Such a conclusion is confirmed by public administration research, which also warns that public servants tend to shield themselves in the face of accountability pressure by adopting defensive behaviour such as formalism. Furthermore, too strong and too narrow an emphasis on accountability at the organizational level can cause prejudice to collective goal attainment.

The fine-grained picture of the microfoundations of accountable actors’ cognitive and emotional states offered by experimental work in social and organizational psychology is most useful for the understanding of the complex processes of ‘felt’ accountability. This body of work confirms that supervision felt as abusive may lead to defensive bolstering and even risks backfiring; however, the outcome depends on a range of moderating factors, and there is evidence that accountability induces actors to become more self-reflective and self-critical. Overall, we learn that accountability has mostly positive effects on individual decision-making quality, contrasting, for example, with some gloomy predictions of dysfunctions that derive from the principal–agent framework. This relatively optimistic note may no longer be valid if accountability is in excess. Nevertheless, perceptions of this aspect inevitably vary, even when individuals are confronted with the same amount of accountability pressure, and responses to pressure are hard to predict. Moreover, if we begin to understand how public managers feel about their accountability obligations, we still know little about how politicians perceive them. Ironically, it is assumed that people generally behave as politicians when they are held accountable, but how real-world politicians feel and act in such situations has not been studied in detail. Not to mention, it may not be justified to infer behavioural patterns of collective actors, which are of interest to students of politics, from individual-level findings.

7.2 Accountability: Kaleidoscopic

A key conclusion is that political accountability is a truly kaleidoscopic phenomenon. It is not just that policy-making is complex, and therefore, accountability relations exist at multiple points and involve a great number of actors and forums that use diverse modes of interaction. Furthermore, accountability processes are multifaceted, potentially generating friction and entailing trade-offs, and the accountability of the individual and collective actors involved in policy-making processes is characterized by ambiguities in many respects:

• Elections seem to perform relatively well as an accountability mechanism, and they plausibly incentivise office-holders to be responsive to citizens’ preferences. They probably do so because office-holders fear retrospective voting, although there may be a gap between their fears and citizens’ actual behaviour. However, even though competitive elections are considered the primary channel of democratic accountability in representative government, confining the study of accountability to its electoral aspect leads to misperceptions. Moreover, although the coupling of vertical accountability to the citizenry with interinstitutional accountability forms is necessary, it may cause frictions.

• With regard to the latter, parliaments are unevenly able or willing to hold their government effectively accountable, so that lower de facto than de jure accountability rather unexpectedly results from forum passivity and not from agency drift. Even when parliamentary control is effective, it may come at the detriment of accountability to the public. By increasingly acting as instances of horizontal accountability, courts indirectly gain influence in the legislative process, as it is rational for policy-makers to anticipate their views. Such interference by the unelected has been criticized, but one should keep in mind that politicians may be able to limit it.

• Political control over the bureaucracy may have expanded, including paradoxically through participatory forms of accountability. Hence, the diffusion of the latter does not necessarily mean more democratic policy-making, but it remains uncertain if enhanced political control has become more effective. Political control is even openly seen as a threat to the impartiality and independence of the proliferating autonomous public agencies. However, this is not to say that technocratic power is unchecked, and accountability may even be overabundant. Top-level public managers and street-level employees are subject to various forms of cross-pressure resulting from conflicting accountability claims (‘many eyes’), leading to dilemmas and disarray, as well as coping strategies and attempts to evade blame. Furthermore, some public agencies voluntarily strive to become accountable for reputational purposes. Thus, they may be more accountable in practice than de jure, but once an agency’s reputation is established, accountability forums tend to trust it and remain inactive. The strategic use of accountability as a reputation-enhancing resource, intended to subsequently shield from overly invasive controls, is perhaps a more generalized mechanism in bureaucracies that has not yet been sufficiently studied.

• In the probably expanding collaborative forms of governance, public accountability competes with participants’ peer accountability to their network partners. It is also de facto undermined because the allocation of responsibility becomes difficult (‘many hands’), especially if collaborative networks are amorphous, informal and lack visibility, with the crucial rule-making actors remaining unseen to outsiders, even if unintentionally. When rule-making is ‘coproduced’ in hybrid partnerships and, even more so, outsourced to private bodies, then the exercise of power may even be divorced from de jure democratic accountability, which best epitomizes the role of accountability as a legitimacy-enhancing device.

• Relatedly, despite the mediatization of politics in a context of monitory or audience democracy and with the rising role of social media, it is above all front-stage politics, lending itself to media ‘dramaturgy’, that unfolds under the public ‘microscope’. Significant activities of the policy-making process remain shielded from media scrutiny because they lack salience and newsworthiness and thus also lack relevance for the increasingly commercialized media agenda. Moreover, most journalists are insufficiently equipped for critical investigations. Hence, although the media are usually considered powerful ‘fire alarms’, they do not always adequately perform their role as accountability forums.

• The decoupling between the political spectacle and the reality of policy development, which undermines accountability’s effectiveness, is likely to persist even in critical situations. On the one hand, policy failure is an easy target for political exploitation, so that postcrisis accountability may be harsh, and responses by policy-makers may prove inappropriate or insufficient to calm criticism. On the other hand, policy-makers are far from being powerless even in such critical episodes, for example, through their ability to deflect the blame, so that the long-term consequences of postcrisis accountability processes remain uncertain.

• Finally, we need to give credit to the quite nuanced conclusions of psychological research regarding accountability’s impact on the feelings and the behaviour of individuals who are held accountable. This body of work shows that the reactions of individuals are shaped by a whole range of moderating factors, whereas studies of politics are, perhaps unduly, more assertive when they assume, for example, that politicians can decipher the preferences of accountability forums and adjust to them. The same applies to bureaucracy studies, which suggest that public managers and employees manage the stress caused by accountability pressure through various kinds of coping strategies. As the validity of such beliefs about policy-makers’ attitudes and behavioural choices has been insufficiently tested until now, the study of accountability in politics would undoubtedly be improved by a careful incorporation of more evidence from social and organizational psychology.

This Element intended to show that cross-fertilization between various relatively insulated research communities is indeed indispensable to capture the prismatic nature of accountability. Studying policy-makers’ accountability requires navigating arenas of governance that have varying properties, such as codification or visibility. It also requires looking at the role of institutions that we must not forget are populated by actors and, therefore, do not necessarily work as intended on paper, just as they do not always produce the expected results. An interactionist approach highlights the existence of complex accountability webs, involving various types of individual and collective actors in their roles as account-givers and account-holders, whose interdependence stems from different organizing principles that coexist, sometimes in a conflicting way, such as hierarchy, delegation, control, antagonism or cooperation. Furthermore, time matters, and accountability fluctuates between periods of routine and equilibrium and critical episodes of high dramatic intensity, in which vulnerabilities and defences come to light: reputations are called into question, but they can be used as a shield to protect oneself from criticism.

It is undoubtedly the variety of action repertoires and modes of interaction in accountability processes that is most striking. Actors seek to impose their narratives, enter into deliberation, struggle or negotiate. They are confronted with incentives and limitations and, therefore, reflect and calculate, but sometimes misperceive opportunities and constraints, as their prior beliefs and inadequate information generate noise. They are unevenly successful in building and damaging reputations, anticipating and managing expectations, taking preferences into account and reshaping them. They establish and challenge causalities, allocate and assume responsibilities, strategically learn to evade accountability and seek it voluntarily. Not less important than their cognitions are their emotions, including not only stress, fear and anger but also satisfaction and gratitude.

Finally, who should be held accountable, for what exactly, by whom, how and with what kind of consequences is often a matter of political controversy and sometimes not settled in normative debate. To give just a few illustrations of dilemmas and open questions: should actors that are not officially authorized to make collectively binding decisions be held politically accountable, and how can their contributions to policy-making be identified if they are unseen? How can one authoritatively assess to what extent constituencies are affected by policy? Which actors legitimately act as fire alarms and as account-holders? Should political accountability be primarily related to diffuse policy outcomes or to public goods delivered to particular target groups? What if responsiveness and democratic accountability to the citizenry collide with peer accountability and compliance with professional norms? Is the threat of sanctions indispensable to effective accountability, or depending on the purposes, should the approach be more educational than punitive? By highlighting core facets of the exercise and control of power, the study of policy-makers’ accountability raises crucial questions on the quality of democratic government, the ‘elephant in the corner’ (Reference Ingram, Schneider, Peters and ZittounIngram and Schneider, 2016) of public policy theory.