Incentive Audit & Risk Checklist

An incentive audit is a safety check for rewards and motivations. Like testing a bridge before allowing traffic, you examine what your system actually encourages people to do—not what you hope it will do. The rat bounty example (called the Cobra effect) shows how well-intentioned rewards can create perverse incentives when people game the system to maximize their payout rather than achieve the intended outcome. This happens everywhere: when teachers are paid based on test scores, they may focus on test prep rather than learning; when sales bonuses reward quarterly numbers, representatives may push products that customers don't actually need. The audit process asks: What does this system truly reward? What shortcuts might rational people take? What unintended behaviors emerge when optimization meets human creativity? Check this: For any important goal you're incentivizing, list three ways someone could achieve the metric without fulfilling the real purpose.

An incentive audit works like this: first, identify who has decision-making power and what rewards them. Then trace through the system step by step, asking what a rational person would do to maximize their reward, even if it harms the overall goal. The audit checks for misalignment between stated objectives and actual motivations, surfaces hidden conflicts of interest, and reveals where metrics might be gamed.

The process draws on principal-agent theory from economics, which examines how one person (the principal) can motivate another (the agent) to act in the principal's interest rather than their own. Incomplete contracts—you can't write rules for every situation—create space for agents to find loopholes. For example, misaligned incentives in financial systems can arise when rewards focus on output volume rather than outcome quality.

Incentive audits apply beyond business. In healthcare, metric-based incentives can create similar gaming pressures, such as when outcome measures influence classification decisions. In education, high-stakes testing pressure has led to cheating scandals and curriculum narrowing, as Campbell's Law predicts: the more a metric is used for decisions, the more it corrupts the process it measures. The checklist approach helps organizations anticipate these problems by systematically examining: who benefits, what's measured, what's ignored, and what rational actors will do when faced with these rewards.

To conduct your own audit, map the full chain of incentives from top-level goals to individual actions. Ask: What happens if this target is hit but the real problem isn't solved? What would a cynical but clever person do? Where are the feedback loops that would catch gaming, and are they strong enough? Then redesign incentives to align individual optimization with collective goals—by rewarding outcomes rather than outputs, using multiple metrics, and preserving discretion for human judgment.

Incentive audits draw on contract theory and principal-agent models developed by Nobel laureates Bengt Holmström and Oliver Hart, who showed how incomplete contracts and information asymmetries create persistent agency problems[5]. Campbell's Law (1976) formalized the corruption of social indicators: any quantitative metric used for decision-making becomes vulnerable to pressure and distorts the process it intends to monitor[2]. Goodhart's Law expresses a similar insight: when a measure becomes a target, it ceases to be a good measure.

• Čihák and Johnston (2013): Proposed incentive audits as a regulatory tool for financial systems, demonstrating how analyzing incentive structures could have predicted and prevented key failures in the 2008 crisis, including misaligned risk-taking incentives and conflicts of interest in credit rating[1].
• Blonz (2023): Studied misaligned incentives in energy efficiency programs, quantifying how contractor behavior affected program outcomes[3].
• Sidorkin (2016): Argued that Campbell's Law reveals ethical limits of measurement, suggesting some social phenomena require qualitative judgment and that taboo against measuring certain domains may protect against metric corruption[2].

The audit approach recognizes that perfect incentive alignment is impossible—contracts are incomplete, information is asymmetric, and human ingenuity always finds new ways to game systems. The goal is not perfection but awareness: identifying the most likely failure modes and building in countermeasures before they cause systemic damage.

Incentive audits have practical constraints. They require honest self-examination, and organizations may resist exposing uncomfortable truths about their own reward structures. The audits can't predict every possible gaming strategy—people are endlessly creative. Some domains resist quantification entirely, making metric-based incentives inappropriate. The approach also assumes rational actors, but behavior is also shaped by culture, norms, emotions, and cognitive biases that incentive models miss. Finally, audits are diagnostic—they identify problems but don't solve them; changing entrenched incentive systems faces political resistance and unintended consequences of its own.