Upwards and downwards accountability and policy implementation
An extension of this model that also includes regulatees and other stakeholders can be represented using the veto players graphical notation developed by Tsebelis (2002: pp. 19-24). The relationship between agency, politics and other social actors is illustrated in Figure 5.2. Given the ideal point of the political power (P*), an agency will be created with a certain degree of independence from politics. This is graphically represented by the distance between P* and the agency's ideal point A*, as well as by the agency's circular indifference curve.11 At the same time, the agency's accountability can be symbolized by the radius of the politician's circular indifference curve: the shorter the radius, the higher the accountability. The size of the politician's indifference curve, obviously, has an impact on the agency's policy choices. As long as accountability includes a sanctioning element, indeed, it must affect policy implementation: the fact that the agency explains its conduct and offers information about its decisions is not without consequences. Consultations are aimed not only at making decisions transparent, but also at receiving feedback that will affect present or future decisions (see Chapter 1 of this volume).
Once created, the agency enjoys a certain degree of autonomy. Although it is not a proper veto player (because it is created by the political power, and the political power can dissolve it or change its powers and prerogatives), the regulatory agency is free to choose its most preferred policy, as long as this falls within the politician's indifference curve. The assumption is that if it decided to implement a policy that the principal cannot accept (that is, that lies outside P's indifference curve), the agency would lose independence or power because of the politician's sanction. Since the agency is the agenda-setter as regards the policy choice, it will decide to implement policy p1, or something very close to it. Obviously, if the preferences of the political principal change (for example, when a new government comes into power), the agency will adjust its policy accordingly. It must be noted, however, that the policies commonly delegated to regulatory agencies are not issues on which there is high polarization of preferences in the political system. In other words, changes due to government alternation are not likely to cause big shifts in the principal's preferences.
Figure 5.2 helps us visualize what happens if the agency decides to develop forms of 'voluntary accountability' (Koop 2014) towards reg- ulatees or other stakeholders. Suppose we have a social or economic
Figure 5.2 Effects of accountability (towards politics and towards social and economic actors) on policy implementation actor, 51, with its own indifference curve, to which the agency is not formally accountable - that is, it may be obliged to offer it information, but not bound to take its position into account when deciding the regulatory policy. Even though the agency is not legally bound to consult them, it might find it beneficial to do so for a variety of reasons: because this would increase the likelihood that its decisions would be correctly implemented; because it would enhance its legitimacy; because it would reinforce its position vis-a-vis the political principal. Ultimately, an agency (within the limits set out in the 'delegation contract') aims to secure its position in the institutional system, and its reputation and credibility are crucial in this respect. Therefore, the agency might decide, for instance, to involve S1 in its regulatory decision-making process.
A choice of this kind would have the effect of changing the implemented policy from p1 to p2. The new policy would still be within the politician's indifference curve, and hence the agency would not face any consequences. On the other hand, the agency would gain in legitimacy and reputation. This will happen, of course, provided the indifference curve of the social or economic actor involved overlaps with that of the principal.
A more realistic scenario would consider the presence of more social or economic actors, whose indifference curves might or might not overlap. In the example represented in Figure 5.2, we have two social actors, S1 and S2, between whom the regulatory agency must choose. If it decides to be voluntarily accountable to S2 (as opposed to being accountable to S1), for instance, it will choose a policy implementation like p3 (instead of p2 or p1). The agency could take this decision because it expects a greater pay-off from involving S3 in its policymaking process, or because (with an equal or similar pay-off) p3 is closer than p2 or p1 to its ideal point A*. Agencies are often confronted with several social actors whose preferences are not compatible: monopolist firms and consumers, incumbents and newcomers, and so forth. Hence, they often have to choose to whom to be accountable. Obviously, this argument does not apply if the agency is bound to consult with certain actors. If the legislator prescribes that one or more actors must be involved, it means that it wants the preference of that actor to be constantly taken into account by the agency. This is what McCubbins, Noll and Weingast (1987) defined as 'deck-stacking': the coalition that creates (or reforms) the agency 'seek[s] to ensure that the bargain struck among the members of the coalition does not unravel once the coalition disbands'. Future regulatory choices are made more predictable and stable, and some social actors are given an 'institutional advantage' over others.