Principal–agent theory of government

Modern government is a chain of delegation. Voters delegate to representatives, legislators to cabinets, ministers to civil servants, and governments to agencies, contractors and international bodies. At each link a principal relies on an agent to act on its behalf. The difficulty is that agents have their own goals and know things principals do not. Principal–agent theory studies how principals can structure these relationships to reduce agency loss, the gap between what principals want and what agents do.

The framework comes from economics, where it was developed for owners and managers, insurers and the insured, and employers and employees. Political scientists adapted it to elections, legislatures and bureaucracies from the 1980s.

Asymmetric information

Two problems recur:

• Adverse selection (hidden information): before delegation, the principal cannot tell good agents from bad ones.
• Moral hazard (hidden action): after delegation, the principal cannot observe the agent's effort or choices.

Stephen Ross (1973) formalised the principal's problem of designing incentives. Michael Jensen and William Meckling (1976) defined agency costs as the sum of monitoring expenditures by the principal, bonding expenditures by the agent and the residual loss that remains even with optimal arrangements.

Political delegation

Terry Moe's "The New Economics of Organization" (1984) introduced these tools to the study of public bureaucracy. Kaare Strøm described parliamentary democracy as a single chain of delegation from voters to members of parliament, prime minister, ministers and civil servants, contrasting it with presidential systems that have multiple competing principals.

Electoral control

John Ferejohn (1986) modelled elections as a means for voters to control incumbents who have private information and their own interests: voters set a performance threshold and re-elect only those who meet it. Later models distinguish sanctioning bad agents from selecting good types, and Timothy Besley's Principled Agents? (2006) emphasises selection of public-spirited politicians as well as incentives.

Oversight strategies

Mathew McCubbins and Thomas Schwartz distinguished police-patrol oversight, in which legislators actively and continuously monitor agencies, from fire-alarm oversight, in which they set up procedures that let citizens and interest groups raise complaints, and argued that legislators rationally prefer the latter. Other tools include ex ante controls (statutory detail, appointment powers, procedural requirements) and ex post controls (audits, budget cuts, hearings, courts).

Multiple principals and agency drift

When an agent serves several principals (a bureaucracy answering to a legislature, an executive and courts), it can play them against each other. Agencies can also drift as their mandates age and their political sponsors change.

• Audits and elections in Brazil: Claudio Ferraz and Frederico Finan exploited the random selection of municipalities for federal audits and found that public release of audit findings before the 2004 elections reduced the re-election of mayors with documented corruption, especially where local radio could spread the information.
• Independent central banks are designed as agents with clear mandates and protection from short-term political pressure, a deliberate trade-off between control and credibility.
• Inspectors general, ombudsmen and supreme audit institutions institutionalise monitoring of executive agents.
• Performance contracts and New Public Management reforms in the United Kingdom, New Zealand and elsewhere applied agency thinking through targets, contracts and executive agencies.
• International delegation: states delegate to the IMF, the European Commission or trade tribunals and then debate how to keep them accountable.
• AI agents: delegating tasks to automated systems raises a new agency problem in which the agent's objectives may differ from the principal's in ways that are hard to observe.

• Who is the principal? In systemic corruption the supposed principals (citizens, honest officials) may themselves expect others to be corrupt. Anna Persson, Bo Rothstein and Jan Teorell, from interviews in Kenya and Uganda, argued that anti-corruption reforms built on principal–agent assumptions fail because systemic corruption is better understood as a collective action problem.
• Motivation: public service motivation research suggests many officials are not narrowly self-interested; heavy monitoring can crowd out intrinsic motivation.
• Measurement: many public outputs are hard to measure, so performance incentives can distort behaviour (teaching to the test, gaming targets).
• Normative thinness: the framework treats principals' preferences as given and says little about deliberation, rights or what agents owe to non-principals.
• Over-formalisation: models often assume a clear principal with coherent goals, whereas real politics features ambiguous, conflicting mandates.

Principal–agent theory is now standard in the analysis of elections, legislatures, bureaucracies, courts, international organisations and corporate governance. It explains why transparency, audits, competitive elections and clear mandates improve accountability, and it has strong empirical support in settings such as the Brazilian audit programme. Its limits are clearest where principals are missing, divided or complicit; there, collective action and institutional trust become the binding constraints.

• Jensen and Meckling, "Theory of the firm" (1976)90026-X)
• Terry M. Moe, "The New Economics of Organization" (1984)
• McCubbins and Schwartz, "Congressional Oversight Overlooked" (1984)
• Kaare Strøm, "Delegation and accountability in parliamentary democracies" (2000)
• Ferraz and Finan, "Exposing Corrupt Politicians" (2008)
• Persson, Rothstein and Teorell, "Why Anticorruption Reforms Fail" (2013)
• Timothy Besley, Principled Agents? (Oxford University Press, 2006)