Kiewiet and McCubbins: The Logic of Delegation

Kiewiet and McCubbins. 1991. The Logic of Delegation. Chicago: University of Chicago Press.

In Brief

Kiewiet and McCubbins's lasting contribution has been to outline how delegation works, the specific problems it creates for the delegator (the "principal"), and the means by which principals can control their agents.

They apply their "principal-agent" framework to the problem of Congressional delegation. Though they examine three Congressional delegations of authority--to committees, to the executive branch, and to bureaucrats--it is perhaps the delegation to bureaucrats that has attracted the most attention and debate.

The conventional wisdom (the "abdication" hypothesis) says that Congress and majority parties have lost much of their original authority through this delegation. Although some delegation was initially necessary, agents increased their authority and it can't be taken back. Frequently, Congress compounded agency problems by creating a new agent to oversee a rogue one (e.g. the GAO to oversee executive spending, duplicating auditing work by the Treasury). Congress lacks the resources to monitor all these agents and actually enforce its will. At worst, Niskanen was right: bureaucrats have captured Congress, and they are extracting ever-increasing sums.

Kiewiet and McCubbins concede the strength of the abdication argument, and recognize that Congress is quite unable to use post hoc monitoring adequately. But rather than dwell on what Congress can't do, Kiewiet and McCubbins seek to show what it can do. It can, for example, structure things in ways that ensure that agents do more or less what Congress wants, even if Congress doesn't spell out everything perfectly. Crucially, it matters how Congress delegates, not simply whether it does so. These ideas are tested through an investigation of the appropriations process.

Delegation and Agency Loss (Ch 2)

Delegation is a common solution to collective action problems and social choice instability: For example, we (as voters) delegate to Senators to work on problems for us. (This argument draws on economic theories about firm organization: Organization is necessary to overcome some collective action problems.)

Agency Loss

Delegation, in principle, can have enormous productivity gains for both principals and agents. Unfortunately, delegation also creates opportunities for agents to act against their principal's interests--we call this problem agency loss. The authors break agency loss into three categories:

1. Hidden actions: The principal cannot know all that an agent does.
2. Hidden information: The principal cannot know all that an agent knows.
3. Madison's dilemma (Federalist 51): After giving the government enough power to control the people, you must somehow force it to control itself. More broadly, after giving an agent power to do a task for you, you must prevent it from using that power against you.

Collective Principals

If the principal is a collective rather than an individual--e.g. Congress (as principal) delegating to agents--then this collective principal will face additional (internal) problems that undermine its control of its agents:

1. Prisoner's dilemmas
2. Coordination problems
3. Social choice instability (see Arrow, Condorcet)

As a result of all this, delegation is certain to entail agency loss (the agent does things against the principal's interests) and agency costs (the cost of mitigating agency losses).

Preventing Agency Loss

There are four key ways of overcoming agency problems. Two are ex ante (done before delegation occurs) while the others are ex post (done after the delegation occurs).

1. Contract design (ex ante, to prevent moral hazard)
2. Screening and selection mechanisms (ex ante, to prevent adverse selection)
3. Monitoring and reporting requirements (ex post, but these are costly and can be manipulated by agents; monitoring takes the form of police patrols (random audits) and fire alarms (by affected third parties))
4. Institutional checks (ex post, like checks and balances). Institutions can be sticky or not.

Delegation by Congress to Congressional Leaders (ch 3)

Party labels are useful informational labels for re-election campaigns. Developing and protecting the party label entails collective action, coordination, and social choice problems, so parties delegate to Congressional leaders. However, they face Madison's dilemma: How do you prevent the leaders from using their control (especially of the Rules Committee) against you?

Of the four mechanisms in ch 2, the dominant one here is screening and selection. Leaders tend to come from the ideological middle of each party (Table 3.3, pg 52). There are also some institutional checks, though; parties in Congress have occasionally placed checks on the leadership, especially during periods of greater ideological heterogeneity within the party.

Delegation by Congress to Committees (ch 4)

The authors examine the case of Congressional (House) delegation to the appropriations committee, which has significant power over the federal budget.

The conventional wisdom: Each Rep wants to target spending at her own district (to boost reelection odds), creating incentives for huge spending logrolls. Though high spending is rational for a single legislator, if everybody overspends on pork for their district, the budget goes into the red. Committee assignments compound the problems: through self-selection, the committees end up being filled with Reps who want only to boost that committee's spending on projects.

The received view is that the House Appropriations committee exists as the "Guardian of the Treasury": it reviews spending bills and eliminates the pork. In reality, Congress can do nothing to prevent future Houses from pursuing their partisan policy priorities. If Congress wants to spend more, it will override the Appropriations Committee or manipulate the congressional budget process. If it wants to spend less, it will do the same.

Delegation by Congress to the Executive (ch 7)

This chapter considers the problem of allowing executive offices--the budget bureau in this case--to influence the budget, a Congressional power. It also considers the problem of divided government.

Conventional Wisdom: Abdication

When it comes to the presidency, the abdication hypothesis argues that the president has shifted from a "chief clerk" to a "chief legislator." Not only has Congress given the president exclusive authority to propose each year's budget (since 1921), it has also allowed the Budget Office (later the Office of Management and Budget, OMB) to claim a power of "central clearance" over appropriations requests, legislative proposals, and even testimony by bureaucratic officials before Congress. Although Congress made an attempt to undo some of this abdication (e.g. by creating the CBO and GAO), it has been unable to do so.

Delegation, not Abdication

Kiewiet and McCubbins seek to counter this abdication hypothesis, arguing instead that this apparent "abdication" actually reflects an effort by the Congressional majority to pursue its goals. The argument has a few parts.

First, Congress began delegating authority to make a budget proposal long before 1921; for a long period, the Secretary of the Treasury was expected to collect and revise all departmental budget requests, subject to the president's oversight--and presidents as far back as John Quincy Adams used this as an opportunity to influence budget requests. The change in 1921 simply built upon previous practices and created a new executive office--the Bureau of the Budget--to do the job. But expanding the president's control over budget proposals was in line with the Republican majority's strategies (in 1921). Several forces combined to make the Republicans adopt a platform of strict fiscal conservatism at this time--the unprecedented debts left from WWI, the shift under Wilson (1913) to a direct (and highly progressive) income tax, and a bureaucracy filled with Democratic appointees (protected by new civil service reforms). By empowering the Bureau of the Budget, Republicans sought to promote fiscal austerity.

Second, creation of "central clearance" does not significantly undermine Congressional goals. The abdication hypothesis holds that central clearance--especially the Budget Bureau's right to clear all testimony before it was given to Congress--would effectively prevent Congress from gathering information about department needs other than from the Budget Bureau. But such a conclusion is unwarranted. "There is simply no way that the bureau or any other entity in the executive branch can choke off the flow of communication between Congress and the executive agencies" (p 176). First, the Budget Bureau can't credibly commit to punishing a department head (i.e. by cutting his budget) for making a direct appeal to a subcommittee, since the department head could just do the same thing again and override the Budget Bureau. Second, Congress can mandate that agencies must provide information, even if the Budget Bureau wants to stifle it. And although the Budget Bureau can centrally clear prepared testimony, it cannot centrally clear testimony that department heads give to Congress. Central clearance does not exist to hinder Congress, then; "Congress accepts central clearance of executive policy proposals because it serves the interests of its members" (p180).

Divided Government

Divided government is a central variable in understanding this chapter. Conflict over the Budget Bureau and such occurs not because of delegation, but because of divided government. The Bureau is often becomes the locus of conflict between a president and an opposing Congress. As such, it often appears that delegation to the Bureau has enabled the president to thwart Congress (all the time). But the fact that it is involved in conflicts does not mean that it is the cause of the conflict.

The abdication hypothesis argues that the creation of the Budget Bureau appeared to be the abdication of the Congress responsibility over policy proposals to the executive. Nonetheless, the authors argue that it has been a delegation designed to aid congressional parties for pursuing their policy goals...even in periods of divided government.