Mandate Slippage , Good and Bad : Making ( Normative ) Sense of Pledge Fulfillment

Despite the substantive findings of existing research, the electoral mandate is still an elusive category in representation theory and empirical political science. The article offers a conceptual framework that promises to properly evaluate mandate fulfillment in general, and pledge fulfillment in particular from the standpoint of the normative theory of representation. In this framework the non-fulfillment of pledges is not necessarily bad for representation since mandate slippage, or the gradual process of abandoning the mandate in the post-election phase, may come in both bad and good forms. The proposed framework also develops an empirical research agenda for measuring the causes of bad mandates and mandate slippage by relying and expanding on the toolkit of empirical pledge research. Outcome oriented pledges serve as a prime example of bad mandates, whereas agency shirking is a major cause of bad mandate slippage.


I. Introduction
Despite the substantive findings of existing research, the electoral mandate is still an elusive category in representation theory and empirical political science. 1 It was not so long ago that Andreas Schedler (1998: 191-192) contended that the 'mandate theory of elections (…) has not commanded too much attention from political science'.Eventually, the concept received the serious treatment it deserved in the form of an edited volume by Manin, Przeworski and Stokes (1999).A serious but less than uncontroversial treatment.
In describing the 'mandate conception of representation' the authors claimed that 'mandate-representation' occurs if 'parties truthfully inform voters about their intentions and the implementation of these intentions is best for voters under the given circumstances' (Manin et al., 1999: 30).While the first part of the definition is in line with most accounts of elections as a means to 'confer the median mandate' (McDonald and Budge, 2005), the second part introduces a somewhat alien element, and a fair amount of tension, into the equation.
On the one hand, informing the citizens of proposed policies and incentives conducive to the implementation of these policies is well mapped in the principalagent literature of delegation and representation (see e.g., Besley, 2006).On the other hand, the injection of a benchmark of citizens' interests in the theoretical framework other than their revealed preferences at the polling station creates conflicting directives for normative evaluation as well as a model that is less suitable for operationalization.Indeed, two decidedly empirical research agendas make only use of the first part of the 'mandate-representation' definition: pledge and saliency research.One way to operationalize the complexities of electoral mandates for empirical research is to look at explicit promises made during the campaigns.Studies following this approach create their databases by extracting relevant information from party manifestos and other electoral communications.
They mainly come in two flavors, which are distinguished based on their respective understanding of mandates.On the one hand, 'saliency'-based approaches (such as the classic study by Budge and Hofferbert, 1990) map general trends in policy-making (often relying on budgetary data) in order to gauge the relevance of party manifestos in setting the direction of government.'Pledge' research, by contrast, singles out individual commitments and the fulfillment thereof, regardless of overarching tendencies in governance (see e.g., Royed, 1996;Thomson, 2001;Thomson et al., 2014).As the latter approach has generated a more extensive literature, in the following pledge research serves as our main focus.
While these studies make use of an increasingly standardized set of variables, they rarely venture into uncharted theoretical territory.They mostly rely on an implicit mandate theory derived from the concept of responsible party government.This 'strictly empirics' line of research, however, is not without its own shortcomings.
Perhaps the most important of these, from the perspective of representation theory, is their dependence on the unpacked concept of pledge fulfillment.In this line of research pledge fulfillment is treated as an inherently positive result for the functioning 1 I am thankful for the comments by two anonymous reviewers.All remaining errors are mine.
SEBŐK, M.: MANDATE SLIPPAGE, GOOD AND BAD: MAKING (NORMATIVE) SENSE OF PLEDGE FULFILLMENT 125 of representative democracy.The problem here is that, as it was highlighted by the dual definition of Manin and his co-authors, the redemption of electoral promises may in fact lead to catastrophic consequences for citizens under changing circumstances.
The present article offers a conceptual framework that promises to properly evaluate mandate fulfillment in general, and pledge fulfillment in particular from the standpoint of the normative theory of representation.In this framework the nonfulfillment of pledges is not necessarily bad for representation since mandate slippage, or the gradual process of abandoning the mandate in the post-election phase, may come in both bad and good forms.The proposed framework also develops an empirical research agenda for measuring the causes of bad mandates and mandate slippage by relying and expanding on the toolkit of empirical pledge research.
The argument unfolds in three steps.First, the baseline principal-agent theory of representation is presented along with a new metaphor of the process of representation: the delegation tree.Second, the concepts of bad mandate and mandate slippage are introduced.Third, the conceptual framework presented in the previous chapters is translated for the purposes of empirical research in order to be able to measure the factors leading to mandate slippage.The final section concludes.
II.The Baseline Principal-Agent Model and its Discontents II.1.A delegation tree with branches in the air It is a common feature of contemporary works on democracy to assume that modern government must derive its 'authority directly or indirectly from the people' (Ferejohn, 1999: 131).It is also clear that variations persist in terms of the exact forms and channels of what we refer to as the program-to-policy linkage (Thomson, 2001).
As the fictional government by the people was suppressed by an indirect government of the people, the concept of representation became inherently linked to the role of elections in a democracy.
Elections are pivotal elements of representative democracy as they establish the core political relationship of the system by linking principals and agents via accountability (Shepsle, 2008: 30).They produce a bundle of winners-by way of sanctioning poor performance or forward-looking selection-and at the same time they also produce mandates.Though it manifests itself in various guises, the latter component is always present in campaigns-even as some authors downplay its relevance in effective electoral control (Fearon, 1999).In fact, classic public choice texts on political accountability share the view that 'if voters vote on the basis of platforms or "issues," politicians have little incentive to do what they promise.Thus, voters might be well-advised to pay attention to the incumbent's performance in office SEBŐK, M.: MANDATE SLIPPAGE, GOOD AND BAD: MAKING (NORMATIVE) SENSE OF PLEDGE FULFILLMENT 126 rather than to the hypothetical promises of competing candidates' (Ferejohn, 1986: 7). 2 This negative general attitude toward a concept of political accountability linked to and based on ex ante authorization 3 spawned a sprawling literature linking popular preferences directly to societal outcomes, without any institutional or policy-related intermediary (see studies on 'opinion/policy consistency/dynamics', such as Monroe, 1998).This results in cutting out the 'middle man', the institutions mediating the content of mandates; in the process the institutional environment that filters this content is removed.Thus, it sheds little light on the mechanism of preference transmission.And in this it presents both empirical and theoretical conundrums: it underestimates the importance-in fact: persistence-of electoral pledges and manifestos in actual campaigns.Perhaps more importantly, it also reduces voters into rational principals similar to corporate shareholders for whom the only relevant metrics is located under the red line of earnings reports.Yet politics resists such simplifications: elections may be won and lost over pledges that turned out to be untrue, or policy switches that contradicted existing party ideology.
There are also persuasive reasons for giving elections (as opposed to, say, public opinion polls) a central role in revealing voter preferences.Election day establishes a political contract between principals and agents that is anything but a fiction or metaphor: it institutes obligations buttressed by constitutional law as opposed to the more informal mechanism of responsiveness.As it happens, this is the root that gives rise to the tree of representative government, and its crown consists of a complex structure of branches and leaves (see Figure 1).Party mandates, then, are best understood as the trunk of this tree of delegation: all future decisions emanate from this original authorization for the party/parties of government.4 2 Even researchers adhering to the mandate tradition acknowledged that mainstream studies had considered parties too weak to 'function effectively as programmatic, policy effecting agents', at least in a system based on the separation of powers (Budge and Hofferbert, 1990: 11). 3 The terms authorization and accountability are used in the sense of the principal-agent model of politics (see e.g.Ferejohn, 1999: 133).Authorization is the 'means by which a representative obtains his or her standing, status, position or office' (Dovi, 2011).Ex ante authorization demands that the authorization takes place before the execution of the task or mission at hand.Accountability is 'the ability of constituents to punish their representative for failing to act in accordance with their wishes (e.g., voting an elected official out of office) or the responsiveness of the representative to the constituents' (ibid.).True, some decisions are not rooted in policy mandates in the substantive sense.The government's relations with social partners and exogenous shocks all shape the general direction of government decisions.Nevertheless, the answers to virtually all of the challenges of governance are embedded in party ideology and the values espoused by a political community of like-minded people (consider, for instance, the motives behind the presidential nominations of Supreme Court judges in the U.S.).These are all part of a broader concept of mandate that is far from devoid of policy content.
The totality of government decisions that involve a modicum of policy content are, therefore, relevant for mandate theory.This is true despite the fact that in many cases no clear path can be discerned connecting the actual policy decision to the original authorization.In other words, government resembles a tree with some branches suspended in the air.These branches may or may not have capillaries emanating from the trunk.Furthermore, some decisions may be in direct contradiction with pledges made in the same policy domain.In this respect, they may disrupt the chain of responsiveness (Powell, 2004) even as they perform some basic government function.The tree of delegation is hence supplemented by branches unrelated to ex ante authorization, and together they form government policy.All in all, mandate models are integral parts of the more general field of political accountability.And in light of these considerations, when it comes to theories of political authorization, they appear simply unavoidable.Following this line of reasoning, in the remainder of this paper it is assumed that party mandates are vital elements of well-functioning representative democracies.5Nevertheless, we are still faced with the problem of defining what exactly constitutes a mandate. 6II.2.The trunk of the tree: The direct flow of the mandate The concept of the electoral mandate has been the subject of interpretations and redefinitions by politicians and political scientists alike.Based on an analysis of the 1984 presidential campaign in the United States, Hershey (1994) discerns three recurring thematic elements in mandate claims by politicians: the party mandate, the personal mandate and the policy mandate.This provides a useful starting point as it highlights the diverse conceptual sources of electoral mandates.The common denominator is the presence of a partially binding content which is associated with ex ante authorization.This partially binding mandate is the essence of representation; and is understood as a counterpoint to the appointment of delegates (with a fully binding mandate) and trustees (with a fully non-binding mandate).
With mandate-based representation thus described, the next challenge is to make sense of the content of the partially binding mandate.In its most general form, the mandate is associated with a beneficiary: a party or person.Insofar as most leaders are members and representatives of their own parties, it is not necessary to distinguish between these two levels.Furthermore, there is an implicit linkage between leader and party: their shared history, decisions and ideology.
In this sense there is arguably no such thing as a mandate without policy content, even if some leaders make no effort to present a manifesto to the public.7 Having said that, parties and politicians in developed countries do have a propensity to publish electoral programs (as witnessed by the main database of the Comparative Manifesto Project).This helps to make the case for the relevance of ex ante authorization in democratic theory and practice.Yet in trying to find the meaning of electoral mandates, manifestos are a part of the problem at least to the same degree as they are part of the solution.What authorization entails still depends on the researchers' point of view, and the prime exhibit which illustrates this phenomenon is the division of the research community along pledge and saliency lines.
Once again, both strands of research rely on a policy-based definition of mandate, only this time they consider explicit pledges and issue emphases as opposed to more implicit party ideology.Explicitness, on the other hand, is not a privilege of manifesto pledges: campaigns are ripe with reactive policy statements, which are SEBŐK, M.: MANDATE SLIPPAGE, GOOD AND BAD: MAKING (NORMATIVE) SENSE OF PLEDGE FULFILLMENT 130 sometimes expressed only verbally, as opposed to the more proactive, written form of communication that is the hallmark of manifestos.Perhaps the best way to understand the overall policy content of campaigns-and, thus, that of the mandate-is to think of it as a word cloud.In this mandate cloud some pledges will appear more often, and in more varied forms and terms, while others retain a small, standalone place in the cloud.This diffuse policy mandate is difficult to interpret and evaluate for social scientists, let alone for the average citizen.That is one of the reasons why research strategies have gravitated towards the piecemeal approach of analyzing manifesto pledges or issue saliency.Studies regarding 'cloud fulfillment' estimates seem like science fiction in light of entrenched research agendas, but they are the next logical step all the same.From the electorate's perspective, such a cloud may come closest to laying out the contents of a contract between themselves (the principals) and the parties (the agents).Insofar as campaigns are informative, and the cloud is filled with pledges and issues that send voters signals about future government policies, it is not an exaggeration to speak of a direct, undisturbed and uncompromised flow of the mandate.
This direct flow represents the trunk of the delegation tree.Although its relation to the root of the tree-voter preferences-is ambiguous8 , the moment of authorization creates a firm link between the two.In formulating the provisions of the mandate contract, it is also unnecessary to take a stand regarding bottom-up or topdown dominance (where the bottom-up approach refers to the dominance of focus groups and opinion polls in constructing party platforms and top-down is a metaphor for elite leadership and, possibly, herethetics).As long as parties offer a selection of campaign contents, a mutually endorsed content for the principal-agent contract is within reach.III.Concepts for Disrupted Principal-Agent Relations III.1.Bad mandates: The missing or ill-defined source While the conceptual development of the delegation tree offers a better metaphor for real-life representative processes, the basic terminology of principal-agent model is still deficient when it comes to explaining disruptions of text-book principal-agent relations.A new dictionary explaining these phenomena could make use of the notions of bad mandates and mandate slippage.
Elections play a pivotal role in setting the content of the mandate, yet the contours of this content are shaped before and after polling day.The pre-election period defines the comprehensiveness of the contract, while the post-election phase determines the rate of contract fulfillment.Taken together, these steps constitute what might be called-by taking a page from Thomson's work-the preferences-program-SEBŐK, M.: MANDATE SLIPPAGE, GOOD AND BAD: MAKING (NORMATIVE) SENSE OF PLEDGE FULFILLMENT 131 policy linkage.As in this model elections are the mechanisms whereby the contents of contracts are endorsed and formally authorized, the preceding and subsequent phases decide if the provisions of approved contracts are actually delivered on.
It follows directly from this framework that the linkage may break down either prior to or following the election.In the former case, bad mandates are the result of bad campaigns or weak authorizations.In the latter instance, disrupted principal-agent relations are the outcome of interactions between self-interested strategic players.The former render proper contract-fulfillment impossible as the preconditions for a principal-agent relationship are not met.The latter entail breakdowns in the execution of ex ante authorizations.Table 1 presents these sources of the breakdown of the preferences-program-policy linkage in a stylized chronological order.Bad mandates become incomplete or void contracts on account of either deficiencies in the pre-election period (the campaign) or weak authorizations provided by the electorate.Among the necessary preconditions for a meaningful contract between principals and agents, the content of the campaign is paramount.The quality of the imparted knowledge (scope, depth, concreteness) determines whether a partly binding mandate is created.Citizens' perceptions of the contents of the contract (or the mandate cloud) are also shaped by party decisions regarding the issues that are emphasized or the relative ratio of rhetorical, ideological and policy statements in the campaign.Bad mandates result from low-information electioneering or from massive overlaps in the platforms of major parties concerning all key issues even as other parties are crowded out.The second source of bad mandates are low-participation elections, for these sever the link between the preferences of the majority of voters (who decided to abstain) and the parties responsible for governing.The third source of bad mandates is the weakness of ex ante authorization.Here the term weakness refers to both the scope of victory (whether it is a plurality, a majority, a supermajority etc.) and the relationship between the share of the popular vote received by the winner vis-à-vis the actual seat allocation in the legislature (even as the form of this relationship may differ between polities).
Bad mandates invariably put a dent in the normative basis of representative democracy.Such deficiencies imply that the potential for meaningful mandatefulfillment is limited.This is also why bad mandates are relevant for normative theory, for they constitute a key area of representative deficit.Coupled with the bad sort of mandate slippage (see below), bad mandates dismantle the preferences-programpolicy linkage which is one of the key elements in the process of providing legitimacy in representative democracies.In sum, input legitimacy may be 'contaminated' right at the source.Nevertheless, anomalies in mandate-based representation are just as common in the later stages of the process.

III.2. Mandate slippage: When the direct flow stops
Modern representative government is built on institutional complexity.This complexity disrupts basic principal-agent relations: multiple principals and various agents crop up throughout the delegation chain and-in some cases-the chain itself is broken.Non-majoritarian institutions draw their legitimacy precisely from the fact that they are disjointed from elected office holders.In light of this complexity, the very usefulness of the 'chain model' is called into question: delegation flows through diverse channels as opposed to just one; and these may further dissociate, to the point of resembling a river delta or the crown of a tree.
The branches can be connected by differing logics of representation.A relationship may be based on delegation in the strictest sense: a transfer of authority with no room for maneuver (Andeweg and Thomassen, 2005).An intermediate form of delegation is based on a partly binding mandate, which allows for some wiggle room in terms of interpreting the contents of the contract.The third form is trusteeship, which is a non-binding transfer of authority when it comes to policy content.
The delegation tree consists of the aforementioned forms of relationships which are created between principals and agents, such as parties, political leaders, ministers and government agencies.They jointly populate the space between the original authorization provided by the people and the policy outcomes that partly serve as the basis for future decisions on authorization.The theoretical consequence of this space between the people who provide ex ante authorization and the people who provide ex post authorization of government performance is that policymakers may become released from direct electoral accountability.This is especially true in the case of long-term trusteeship contracts (as is the case with supreme court or constitutional court appointees in some countries-once again, see Table 1).SEBŐK, M.: MANDATE SLIPPAGE, GOOD AND BAD: MAKING (NORMATIVE) SENSE OF PLEDGE FULFILLMENT 133 This phenomenon may also occur in parts of the tree where initially there is a clear and uncontested direct flow of delegation from the root/trunk.Three such phases merit a more detailed discussion: coalition formation; decisions related to the government structure; and their implementation.
Election results are often ambiguous.In fact, in parliamentary democracies a single-party majority in (both houses of) the legislature is the exception rather than the norm.This sets the stage for a duplication (triplication etc.) of principal-agent relations at the top of the trunk.Various parties run on a wide assortment of pledges and issue emphases, and the coalition formation process blends these policy contents into an unpredictable shape: coalition agreements.In some cases there may be more than one majority constellation (including grand coalitions and the like), and even on its own this fact blurs electoral mandates.Overall, the problem of incomplete contracts manifests itself already at this first stage.
This coalition effect is further compounded by an unbalanced relationship in party coalitions.The hierarchical ordering of parties assigns different weights to partyspecific mandates (as each party proposed a different manifesto with varying issue emphases): the formateur party may have more clout over the coalition agreement, just as a minority party with a great potential for hostage taking (especially when the formateur party only won a plurality) can end up being overrepresented.Either way, the chain of legitimacy (Nullmeier and Pritzlaff, 2010) remains intact while delegation relationships are muddled.
The second step concerns government structure.It is shaped by the (formal or informal) coalition agreement in the form of allocations of control over ministries/departments or by appointments to non-majoritarian bodies and issue emphases in the government program.It is also a manifestation of the underlying constitutional structure, i.e., the specifics of the system of separation of powers.Multiple principals may have jurisdiction over the same policy domain, which may result in obscuring responsibility for the fulfillment of specific pledges.Finally, some pieces of legislation may require a supermajority, which brings in the opposition as a new set of stakeholders.
The third major layer of complex delegation relationships is located at the level of implementation.Bureaucratic/agency behavior may be one of the important reasons behind the failure to fulfill a mandate.Classic causes underlying this include agency shirking and problems related to agency (or more general policy) design.The former refers to cases in which the ideal policy position of the agency is different from that of its principal (the legislature or the executive- Epstein and O'Halloran, 1999).Although the principals have various measures of ex ante or ex post control at their disposal, some level of shirking or laxity may nevertheless persist.
Agency design may make matters worse: multiple missions or fuzzy missions confuse agents as to who their real targets are.Multiple and heterogeneous agents functioning in the same policy domain can replicate this jurisdictional overlap from the policy-maker level (where two departments are responsible for the same or largely overlapping policy area) to the 'street level' of bureaucrats (where e.g., two separate police units are responsible for the same or overlapping geographical area or crime type).And coordination between majoritarian and non-majoritarian institutions (such as those between a finance ministry and an independent central bank) can further SEBŐK, M.: MANDATE SLIPPAGE, GOOD AND BAD: MAKING (NORMATIVE) SENSE OF PLEDGE FULFILLMENT 134 complicate mandate-fulfillment.Again, there is a limit to the extent to which careful policy design can be used to solve the 'problem of many hands' (Thompson, 1980) or the 'difficulty of assigning responsibility in organizations in which many different individuals contribute to decisions and policies'.
The three phases of coalition formation, government structure design and implementation depicted in Table 1-along with many other potential layers along the same lines-represent interruptions in the direct flow of delegation.They break or dissolve parts of the chain of delegation.Thus, they contribute to mandate nonfulfillment, which-in the existing research-is considered a negative outcome from a normative theory perspective.The next subsection makes the case that this perception is somewhat misguided, as mandate slippage may come in two flavors: good and bad.

III.3. Problems of controlling agents through mandates
Breakdowns in the direct flow of legitimacy through the tree of delegation cast a shadow on theories of ex ante authorization.From the perspective of normative democratic theory, this is not all bad news, however.Mandate non-fulfillment or mandate slippage may manifest itself in many different forms during the long process spanning the time when campaign pledges are made to when policy outcomes are first realized.Each phase has its own normative character, which also means that no generalization can be made with respect to unfulfilled promises without the analysis of the normative character of each segment.Together they are a testament to the problems and virtues of controlling agents through mandates.
This 'neglected side' of mandate theory was seldom subjected to a more detailed discussion.In one of these attempts Schedler (1998) contends that electoral accountability is a complex and contested exercise.According to him, the public controversies associated with the very idea of mandate-fulfillment can be settled only by the voter, whom he considers the 'supreme judge.'While this is probably true, there is a possibility to dig deeper and analyze the causes that shape this normative judgment.And these causes may be connected to the position of the individual 'transgression' in the mandate slippage process.
The concept of mandate slippage is best understood as a progressive divergence-realized in the course of governance-from the policy content of ex ante political authorization.It may be the product of strategic agency or an inadvertent consequence of decisions unrelated to mandate-fulfillment.The result is a delegation tree with some branches firmly connected to the root and trunk; and others seemingly up in the air without such attachments.
Not all such breaks in the delegation chain are normatively unattractive.Modern representative government draws its legitimacy from a number of sources, and policy content-related input legitimacy is just one such element.Trustee-type institutions of 'government for the people' take no formal orders from elected leaders (who serve relatively short terms) as they fulfill their long-term mission.And voters have a tendency to acknowledge decisions made in a complex political and institutional environment that includes coalition governments, multi-level government and elaborate structures of government agencies.In this complex environment, the SEBŐK, M.: MANDATE SLIPPAGE, GOOD AND BAD: MAKING (NORMATIVE) SENSE OF PLEDGE FULFILLMENT 135 merits of mandate-fulfillment are relative.It is advisable, therefore, to undertake a piece by piece, phase by phase analysis of the process of 'slippage.'Mandate slippages may be assigned to either of three categories based on their normative status.On the one hand, 'good slippages' allow for the contradiction-laden process of converting heterogeneous voter preferences into actual policy outputs.'Bad slippages' on the other hand are unrelated to plurality and mostly consist of institutional frictions (principal-agent anomalies) stemming from the process of governing.A third category refers to slippages that highlight a trade-off between following voter preferences and other normatively worthy requirements (such as governability or stability).Table 2 provides a summary of these phases of the mandate slippage process.In parliamentary systems the first breakdown in the delegation process usually occurs no later than election night.On the one hand, the policy content of the mandate may be straightforward when single-party majorities emerge (but even in these cases 'skeletons in the closet' may divert the content or emphasis of policy-making).At the same time, multi-party majorities may upset clear expectations with coalition agreements which create a government that pursues only a subset of the pledges featured in each participant's manifesto.Moreover, these pledges may be contradictory, which creates the need to reach a compromise, along with the necessity to insert new or modified policy initiatives into the government program.The literature on the relative merits of single-party and coalition governments underscores the importance of a trade-off between governability and compromise.Ideological distance between coalition parties may be a force for good, but at the same time it may also act as a source of constant tension.Issues related to polarizationwhich are widespread in contemporary two-party systems-may be mitigated by the art of compromise as practiced in multi-party governments and grand coalitions.In light of the conclusions found in this line of research, the least that can be said is that coalition governments may be both natural and appealing outcomes of the electoral systems typically used in parliamentary democracies.
The next phase of mandate slippage concerns policy-related problems stemming from the structure of government.Non-fulfillment in this stage may be due to either incomplete contracts or lack of jurisdiction.First, government programs are not identical with coalition agreements.They address issues that were not emphasized in manifestos simply because campaigns and governments prioritize different issues.In most policy areas there is no mandate content to begin governance with: these contracts are incomplete and may contain no information even for some salient issues.
A solution may lie in the combination of pledge and saliency approaches.The more transparently a party's stance is on a policy domain, the better.But in a situation when campaign strategists made no specific attempts to highlight an issue, the saliency approach may fill in the blanks.Then, non-compliance with this more general policy mandate at this stage is detrimental to mandate-based accountability.
With respect to the structure of government, the second problem of nonfulfillment stems from lack of jurisdiction.Let us assume that parties make pledges in good faith and they are willing to execute them if the opportunity arises.Nevertheless, there is a considerable difference between being willing and being able to execute policies.Portfolio allocation is key in this respect: controlling the relevant chunk of government bureaucracy is almost a prerequisite for solving complex policy problems.While issue-oriented parties (such as Green parties) have a tendency to ask for cabinet positions related to their main area of concern, they may fail to secure them.The fulfillment of pledges concerning policy areas that fall outside the direct policy control of a coalition party should therefore be evaluated against norms that are less strict than the ones one would apply in the case of full portfolio control.
After these party-related issues, the third phase of mandate slippage pertains to government and execution.Formal veto points are less of an issue at this stage, the focus is on more informal factors in the way of mandate redemption.In some cases policy outputs require a mere act of parliament.In other cases fulfillment depends on a complex web of agencies and their cooperation.
Part of the art of governance is policy design and bureaucratic control.In single party settings non-fulfillment resulting from these two factors should be frowned upon from a normative perspective.This also applies to 'agency slippage,' regardless of its source (shirking, opportunism or sheer incapacity).In coalition governments it is more difficult to assign the blame to a specific actor.Having said that, ministerial control should serve as a useful rule of thumb for determining where responsibility lies.
Pledges related to policy outcomes are trickier still.Exogenous shocks may force cabinets to change course and go back on pledges.Such policy switches may nevertheless be tolerated by the electorate in the case of actual disasters that justify the change.In other situations 'partnership non-compliance' might occur: as in the case of a pledge to end a war with a neighboring country.It follows directly from this discussion that pledges targeting outcomes are less enforceable and parties should therefore mostly refrain from them.And for those outcome pledges that are SEBŐK, M.: MANDATE SLIPPAGE, GOOD AND BAD: MAKING (NORMATIVE) SENSE OF PLEDGE FULFILLMENT 137 'unavoidable,' the evaluation of non-fulfillment will have to factor in outside circumstances.
The conclusion that can derived from this cursory analysis leads us back to Schedler's contention regarding the final judges of government performance.Experience also shows that voters make complex decisions regarding incumbents: their evaluation involves the ex post analysis of performance (including 'economic voting', pledge fulfillment, etc.) and also ex ante assessments of future performance.In this larger shape of things, mandate-fulfillment is but one metric that informs the electoral control of agents.Mandate slippages need not involve democratic deficits, let alone representation deficits.From the average citizen's perspective, therefore, mandate slippage may either be a normatively good thing or a normatively negative thing.
This section provided an introduction into the theoretical problems of mandate slippage.The final task is to relate the concept of good and bad mandate slippages to the existing literature on pledge fulfillment so that we can make empirical sense of these theoretical propositions.

IV. Explaining Mandate Slippage with Empirical Variables
Studies in the pledge fulfillment have used various sources of data and produced impressive results regarding the empirical strength of ex ante authorization theories.However, there is no overarching theme in the literature concerning mandatefulfillment besides implicit mandate theory.As the delegation chain is dissected into pieces based on the focus of individual researchers, the literature remains devoid of hypotheses regarding the preferences-program-policy linkage as a whole (with the usual caveat of government/opposition).
Despite its purely empiricist inclination, the literature on pledge fulfillment still serves as the best choice available for making empirical sense of, or indeed measure, mandate slippage.Two aspects of empirical pledge research deserve special attention from the perspective of mandate slippage theory.First, its theoretical sources and normative statements regarding representation.The task here is to relate its explicit conceptual underpinnings or implicit tendencies to the framework presented in the previous chapter.Second, the groups of its empirical variables as related to the phases of linkage breakdown in general, and to bad mandate slippage in particular (see Table 2).By highlighting pivotal variables influencing mandate slippage the measurement of this very concept comes within reach.The theoretical sources of empirical research on the program-to-policy linkage are drawn of a distinctly canonic set of literature.Recent pledge research (Kostadinova, 2013;Toros, 2015;Praprotnik, 2015;Thomson et al., 2014;Naurin, 2013;Dobos and Gyulai, 2015) mostly takes inspiration from a small selection of classic pieces in pledge research (Royed, 1996;Artés, 2013;Artés and Bustos, 2008;Naurin, 2011;Moury, 2011;Mansergh and Thomson, 2007;Costello and Thomson, 2008; and the APSA papers by a group of first generational pledge scholars: Thomson et al., 2010;2012;2014) and the theoretical literature these classic pieces make reference to.
The two main theoretical sources of these classic studies are the literature on responsible party government (following, inter alia, APSA, 1950 andKlingemann, Hofferbert andBudge, 1994; for an overview see: Körösényi and Sebők, 2013) and the theory of parties and coalitions as adapted to European context (such as Laver and Shepsle, 1996;Strøm et al., 2008).Nevertheless, formal models of principal-agent relations are not prominently featured in any of the landmark studies of pledge research.
The discussion of the normative value of the results is even less pronounced.The staple quotation here is by Mansbridge (2003: 515) regarding 'promissory representation'.The works of Manin (1997) and his co-authors (Manin et al., 1999) is also often summoned.In neither cases, however, are empirical hypotheses directly related to these theoretical underpinnings.This is true even as the tension regarding the normative aspects of the results is palpable in some work in the saliency or pledge traditions.Pennings (2005: 14) acknowledges that 'the Dutch case shows that there are several structural barriers for the mandatory role of parties in consensus democracies where mandates are always shared with other parties.'But the next step in the normative analysis of these 'structural barriers' is missing.Similarly, Royed and Borrelli (1999: 115) conclude that 'institutional control is fairly predictive of the relative success of the parties, although other factors are also influential.'Once again, the analysis does not go further into an examination of the relative importance of 'party success' and 'institutional control' for the proper functioning of the delegation chain.
It is important to emphasize that these features of the literature are not to be considered to be shortcomings per se.Pledge researchers follow their specific research agenda, which is empirical in its nature.Furthermore, they do indeed make some progress toward generalizing the content of their preferred variables.With some refinements these can indeed be related to various stages of the linkage process and, therefore, can be factored into normative evaluations of the process of representation as well.

IV.2. Variable groups in empirical pledge research
The preeminent characteristic of empirical pledge research is the heterogeneity of its hypotheses and explanatory variables (see Table 3 in the Appendix) 9 .Perhaps the single quasi-permanent hypothesis in over 25 years of research refers to government/opposition party position, with government parties expected to achieve a higher rate of pledge fulfillment.Other recurring variables include status quo (in pledge content), consensus over a pledge in multiple party manifestos as well as ministerial control related to the pledge in question.
While a degree of standardization has been attained regarding a core set of variables, actual hypotheses are still tailored to the research question of individual papers.Factors related to the institutional setting of government (majority/minority; single party/coalition; affiliation of the chief executive; portfolio control etc.) are wellexplored, the initial and latter stages of the preferences-program-policy linkage make no appearance in most pledge research.From voter preferences regarding various issues 10 to agency and partner slippage the list of important components of mandate slippage are mostly absent from empirical studies of pledge fulfillment.
Only a few studies venture beyond stand-alone variables in order to make sense of variable groups and their role in the linkage process.Royed (1996: 48) conceptualizes the 'factors influencing pledge fulfilment' in terms of 'leadership', 'constraints' and 'decision-making environment'.The first of these is usually not captured by variables in pledge research (such as a presence of a 'programmatic leader').The second is most often represented in the control variables section of models.The third, however, foreshadows the proliferation of institutional variables in comparative studies, introduced from the early 2000s.
Table 3 in the Appendix presents these overlapping ideas in a unified structure along with the pertinent variables in empirical research.Besides exogenous factors (which partly incorporates the 'resources' category) and dummies (mostly used for election years/periods), three major variable clusters emerge: pledge-related, partyrelated and government-related factors.Pledge-related variables are associated with the 9 For saliency studies, the most widely used pairing of variables consisted of the issue emphases in federal expenditure categories (dependent variable) and those of election platforms (which was used as the explanatory variable; in this the authors followed the lead of Budge and Hofferbert, 1990).This research direction is strongly tied to the work of the Manifesto Research Group (MRG)/Comparative Manifestos Project (CMP) and the policy topic coding system used by the latter (see https://manifestoproject.wzb.eu/coding_schemes/1).Recent developments added breadth to the approach, which now focuses on voter preferences as they are manifest in survey data (Pennings, 2005), inter alia.On the dependent variable side, a three-fold description of pledge fulfillment (fully, partially or not fulfilled) has emerged as the standard, even as the causal models in these pieces of research are almost exclusively binary: fulfilled/not fulfilled.(I thank the anonymous reviewer for making this pointfor exceptions see e.g., Toros, 2015: 246.) 10 For an exception see the survey method used by Naurin, 2011.SEBŐK, M.: MANDATE SLIPPAGE, GOOD AND BAD: MAKING (NORMATIVE) SENSE OF PLEDGE FULFILLMENT 140 pre-election phase of the linkage process.Most often they refer to the content of manifestos and pledges (issue area; left/right understood in terms of e.g., tax cuts; status quo or change).Some others touch on the entire corpus of pledges (saliency measures) or the action or outcome oriented nature.Two further groups are relevant for the post-election phase of the linkage process.Party-related variables describe the political clout of parties (legislative control), its history (incumbency) or its future oriented pledges vis-á-vis other parties' pledges (consensus).The third group, government-related factors, resembles previous categorizations most ('government type'; 'decision-making environment').This threefold classification scheme provides a means to relate the underlying empirical variable groups to mandate slippage theory.

V. Measuring Mandate Slippage with Pledge Research Variables
Situating variables in the linkage process is a crucial step toward measuring mandate slippage.First of all, some variables are unrelated to the concept of mandate slippage as they refer to the pre-electoral phase of the process.These variables describe pledges as components of manifestos: the content or saliency of specific promises.Anomalies, such as an overwhelming reliance on outcome pledges (as opposed to output/action pledges), in these cases are only relevant for a discussion on bad mandates, as opposed to the mandate slippage.
As for the post-election phase of the linkage process pledge research offers insights into mandate slippage, at least in some cases.A simple re-ordering of the elements along the tree structure of Figure 1 will immediately shed light on the potential of adapting these empirical variables for the purposes of normative evaluation (see Table 4 in the Appendix).
The linkage process unfolds through five major phases: the formulation of policy preferences; the formulation of pledges; the assignment of party seats in the legislature; government formation; and execution.A brief evaluation of the variables, as situated in this process, from a normative standpoint is as follows.
Forward-looking voter preferences lie at the root of all ex ante theories of representation.However, these may or may not be reflected in party manifestos since bottom-up preference representation is limited by incomplete, transient and manipulated preferences.The actual mandate formation process is less reminiscent of a nationwide poll of opinions on all policy issue than an amalgamation of party stances, ideologies and polls of preferences regarding specific issues.The resulting electoral programs provide a transparent interface for voters' interaction with parties, and for both interested parties the content of the contract is unveiled during the campaign.Information is readily available, at no or negligible cost, and voters make their choice by either using it or opting for rational ignorance about it.
The key normative concept of these first two phases (preferences formulation; pledge formulation) is the bad mandate.A bad mandate is by and large unrelated to mandate content: electoral authorization is always ripe with content as a mandate may include policy, personal and party elements with only blurred lines separating these segments (Hershey, 1994).Manifestos also provide a wide selection of information for Nevertheless, the most widely used variables in this category only derive their normative importance from describing the information level of campaigns not actual policy positions.In this context better information (in terms of its scope, depth, concreteness: see the output/outcome variable) equals better mandates.Similarly, the lack of manifestos may be indicative of bad mandates (even as a counterweight to the lack of explicit manifestos the policy history of parties and candidates should also be factored in).More salient pledges, on the contrary, have a higher visibility in the mandate 'cloud' and, therefore, they are expected to be fulfilled at a higher rate.Finally, economic indicators are factored in both to electoral decisions and policy outcome, yet-as they lag behind political decisions; and an element of luck is ever present-they have no clear normative relevance. 11lection day provides a linkage between the electorate and parties, which forms the basis of a legitimate government (see ex ante authorization, as the first step in the delegation chain).Elections (via turnout, voting and the electoral system) convert policy preferences to party size in the legislature.
The crucial element here is the proportionality of pledge fulfillment to party size.This sets the normative evaluation of a number of overlapping variables (including coalition/grand coalition; chief executive).Of these, government/opposition clearly enjoys a unique position, both in theory and empirical research: it selects the subjects of pledge fulfillment (governing parties) as well as the actual set of pledges to be fulfilled.Consensus pledges, which a number of parties explicitly support, are to be fulfilled at a higher ratio than, say, 'purely' opposition pledges because of their wider support.This reasoning also applies to pledges featured in the coalition agreement.
Perhaps the most developed set of explanatory variables in current mandate research concern government formation.Yet most of these variables are related to the trade-off between governability (having a stable government as opposed to a constant flow of early elections) and pledge fulfillment.This phase gives rise to the idea of unified and divided mandates, with the latter prevalent in the cases of coalition governments.The number of participants (coalition/single party government) and their ideological background (coalition/grand coalition) in itself signifies no normative value.They may be the result of a dominantly proportional electoral system or heterogeneous voter preferences.Furthermore, ministerial control may be misaligned with pledges made, but may be the price to be paid for a stable government.
While variables related to government formation may serve as the key focus of some pledge research, the ultimate phase of implementation is more important from the perspective of normative analysis.In this stage the mandate is already set for each policy area: either by virtue of the initial authorization or the reshuffle of saliency and SEBŐK, M.: MANDATE SLIPPAGE, GOOD AND BAD: MAKING (NORMATIVE) SENSE OF PLEDGE FULFILLMENT 142 control due to government formation.Nevertheless, divided government still plays a role with regards to the problem of input vs. output legitimacy.The wider bureaucratic structure, including the institutions of the system of separation of powers is a 'melting pot', where sources of legitimacy are intermixed.From an institutional perspective, constitutional courts and independent central banks are designed in a way that is expressly meant to give them the ability to withstand public pressure.Thus, input legitimacy (the theoretical basis on which ex ante authorization processes are built) competes with output legitimacy: in the eyes of the electorate, the value of pledge fulfillment competes with 'good' policy outcomes.The trunk of the delegation tree is disconnected from certain policy decisions in the given policy domain: some branches grow out of this trunk, others are hanging in the air.
This presents a complicated scenario for normative analysis.On the one hand, non-fulfillment or the lack of co-operation by non-majoritarian institutions (such as the role a central bank plays in pledges concerning economic growth) represents a tradeoff between input and output legitimacy.On the other hand, agency slippage, in the form of shirking by street level bureaucrats or others, is considered detrimental for the proper functioning of the mandate model.The status of what could be referred to as partner noncompliance is less straightforward: this is typical of outcome-related pledges, as they require the cooperation of extra-governmental actors.In these cases coercing non-obliging actors into joint action may have adverse consequences for representative government as a whole.
Taken together, the above-mentioned factors have a bearing on the rate of pledge fulfillment.In general, pledge fulfillment may be favorable from a normative perspective, but non-fulfillment is not necessarily bad.As a result pledge fulfillment can only be normatively evaluated in the wider context of the process of representation, and by a detailed analysis of each step and variable related to this process.
Mandate slippage can only be measured when provided a good mandate: missing and ill-defined sources are difficult to track through the process of delegation.Variables widely used in pledge research may prove key components in any empirical analysis of mandate slippage.They may fall into good, neutral or bad categories depending on the presence of logics competing with the inherent value of pledge fulfillment (such as governability or output legitimacy).
As for our normative analysis, only a few variables stand out for their normative relevance both when it comes to bad mandates and bad mandate slippage.The composition of the mandate cloud lead to relatively good or bad mandates, as in the case of output and outcome pledges.Bad mandate slippage may result from agency shirking in the implementation phase.In contrast, the positive evaluation of government party fulfillment could lead to a wider category encompassing various forms of popular support and party size (pledges enjoying a wider consensus; put forth by the formateur party/chief executive).

VI. Conclusion
This article introduced the concept of mandate slippage, and the related terminology of bad mandates, as well as good and bad mandate slippage.It relates these concepts to empirical pledge research as the research agenda of the latter presents a unique opportunity to apply these concepts to the reality of contemporary representative government.The argument unfolded in three steps.First, the baseline principal-agent theory of representation was presented.Its various discontents include its reliance on the delegation chain metaphor and its inability to incorporate parallel processes of governance.For these reasons the alternative metaphor of the delegation tree was introduced.This captures more accurately the multiplicity of policy subsystems and the individual paths of various pledges from formulation to implementation.
Second, the concepts of bad mandates, and those of good and bad mandate slippages were introduced in order to account for disrupted principal-agent relations.Third, the conceptual framework presented in the previous chapters is translated for the purposes of empirical research in order to be able to measure mandate slippage.It was argued that the theory of mandate slippage offers a hitherto missing general framework for situating and understanding empirical variables in the preferencesprogram-policy linkage.
Throughout our analysis the point was made that empirical pledge research will only make sense from a normative perspective if it can answer two simple questions: 1. Is non-fulfillment bad for representation?2. If and when it is bad, which factors are responsible for this negative outcome?
In contrast to extant research based on implicit mandate theory, the answer put forth to the first questions is: 'not necessarily'.Indeed, one of the aims of this paper was to show that non-fulfillment may be related to neutral or even beneficial factors in the linkage process leading to 'good' representation.In the most simple rendering, 'good' representation consists of 'good' substance and 'good' process.On the one hand, a properly functioning delegation chain (or tree) provides the latter.On the other hand, it is quite impossible to forecast in a pre-defined manifesto all decisions of a parliamentary cycle of, say, four years.This is why both pledges and issue saliency matter-and also the missing element of underlying party ideology as well as the general trends in the (policy) contents of past decisions.
'Good substance,' therefore, partly results from this more general link between campaign content (best understood as a 'pledge cloud') and government deeds.The other source of good substance is leadership or finding adequate answers in response to a changing environment (see exogenous shocks and the associated policy switches, Stokes, 1999). 12he second research question concerns factors underlying pledge nonfulfillment in a negative normative status.In the analysis two main culprits are picked out: bad mandates and bad mandate slippage.Bad mandates make pledge fulfillment irrelevant as in these cases there are no pledges to start with, or because they are vague, disorienting even.Bad mandate slippage causes the non-fulfillment of pledges in a way that is unrelated to normatively valuable trade-offs between fulfillment and governability or outcome legitimacy.
In conclusion, a normatively relevant pledge (and saliency) research agenda should focus more on variables that have some bearing on bad mandates or bad mandate slippage.This entails more emphasis on the initial and ultimate phases of the delegation process: pledge formulation and implementation.A reoriented research agenda will make great strides toward fulfilling the promise of pledge research as the empirical study of representative government.
Figure1.The flow of legitimacy through a branch of the mandate process
SEBŐK, M.: MANDATE SLIPPAGE, GOOD AND BAD: MAKING (NORMATIVE) SENSE OF PLEDGE FULFILLMENT 138 IV.1.The theoretical sources of pledge research

Table 1 .
Sources of the breakdown of the preferences-program-policy linkage

Table 2 .
Examples of mandate slippage (post-election phase) SEBŐK, M.: MANDATE SLIPPAGE, GOOD AND BAD: MAKING (NORMATIVE) SENSE OF PLEDGE FULFILLMENT 141 voters, including the party's position in the policy space (such as the left-right spectrum), what issues it emphasizes, and the general weight of policy issues as opposed to political statements with no policy relevance.The wide variety of variables in empirical pledge research describing content (context area; status quo; favored groups; expand/cut taxes; policy instruments) reflects the diversity of types of content.

Table 4 .
SEBŐK, M.: MANDATE SLIPPAGE, GOOD AND BAD: MAKING (NORMATIVE) SENSE OF PLEDGE FULFILLMENT 150 The normative status of pledge research variables