How Incentives Work | Rewards, Costs and Behaviour

An incentive is something that changes the relative attractiveness of one action compared with another.

In one line: incentives work by changing expected rewards, costs, risks or social consequences—but the behaviour they produce depends on what people notice, value, believe and can actually control.

Evidence boundary: Incentives are central to economics, but behavioural research shows they do not operate mechanically. Monetary rewards, penalties, social norms, fairness, timing, autonomy and intrinsic motivation can interact. This article therefore treats incentives as behaviour-shaping signals inside a larger system, not as a universal “pay more, get more” law.

Parents, schools, companies and governments all use incentives. Marks, praise, bonuses, fines, deadlines, privileges, rankings, fees and social recognition can all shift behaviour.

The difficulty is that incentives often change more than the one behaviour designers intended.

What Is an Incentive?

An incentive changes the payoff structure around a choice.

Choice → perceived reward/cost → expectation → action → consequence → learning → changed future behaviour.

The same incentive can affect different people differently because they may value the reward differently, interpret the signal differently or face different constraints.

1. Incentives Change Relative Payoffs

If a behaviour becomes more rewarding or less costly, people may become more likely to choose it. If it becomes more costly or risky, they may avoid it.

A bonus can increase the payoff for reaching a target. A late fee increases the cost of delay. Public recognition can increase the social reward for contribution.

But the designer must ask whether the targeted action is actually under the person’s control.

2. Incentives Work Through Perception, Not Merely Through Their Official Design

An incentive that nobody notices cannot influence a decision. An incentive that is too distant, confusing or uncertain may carry little weight.

People also interpret what an incentive means. A reward may signal appreciation. It may also signal that the task is unpleasant enough that payment is needed to make someone do it.

Behaviour therefore responds to both the payoff and the message around the payoff.

3. Timing Changes Strength

Immediate consequences often influence behaviour more strongly than distant ones.

A student may value examination success in six months yet still choose the immediate reward of a game tonight. A company may know maintenance reduces future failure yet postpone it because the cost is immediate and the benefit is delayed.

Good incentive design therefore considers when the person experiences the reward or cost, not only its size.

4. Measurement Determines What Gets Rewarded

Incentives usually attach to something observable.

If a school rewards only test scores, students and teachers may shift attention toward what raises scores. If a workplace rewards only ticket closure, people may close easy tickets while difficult important work waits.

This creates a central rule:

People can optimise the metric instead of the real job.

5. Rewards Can Increase Effort—But Not Always the Capability Behind It

A reward can increase how much effort someone applies. It cannot automatically supply missing knowledge, skill or tools.

If a student does not understand algebra, promising a reward for a higher mark may increase practice without fixing the conceptual gap. If an employee lacks authority to solve a bottleneck, a performance bonus may increase pressure while leaving the constraint untouched.

Before strengthening an incentive, diagnose whether the problem is actually motivational.

6. Extrinsic Rewards Can Interact With Intrinsic Motivation

People sometimes act because they enjoy the activity, value the purpose, identify with the role or feel responsibility toward others.

Adding external rewards can support these motives, leave them unchanged or sometimes weaken them if the person begins to experience the activity mainly as controlled from outside.

The OECD’s behavioural work explicitly warns that work motivation is not reducible to monetary incentives alone and that poorly designed rewards can carry hidden motivational costs.

7. Fairness Affects Whether Incentives Are Accepted

People compare rewards and penalties with what others receive and with what they believe the rule should be.

A technically efficient incentive can create resentment if people perceive the process as unfair, impossible to influence or biased toward those with better starting conditions.

Fairness changes cooperation because people are not only maximising money. They are also interpreting the relationship and the rule.

8. Social Incentives Matter

Approval, status, belonging, reputation and shame can all change behaviour.

A student may contribute because peers value contribution. A professional may follow a standard because competence is part of professional identity. A person may avoid asking for help because the local culture treats uncertainty as weakness.

Social incentives can be powerful precisely because they act on identity and relationships, which makes them ethically important to design carefully.

9. Penalties Can Deter Behaviour but Also Change the Meaning of the Rule

A penalty raises the cost of an action. This can reduce the behaviour if the penalty is credible, visible and sufficiently relevant.

But a fine can sometimes be interpreted as a price: “I am allowed to do this if I pay.” The moral or social meaning of the rule can shift.

Deterrence therefore depends on enforcement, interpretation and legitimacy as well as penalty size.

10. Incentives Create Second-Order Behaviour

People adapt to the incentive system itself.

If bonuses depend on a threshold, people may concentrate effort just above the threshold. If rankings matter, people may avoid helping competitors. If attendance is rewarded but learning is not measured, presence may improve without capability changing.

Good design anticipates gaming rather than assuming people will respond only in the intended way.

11. Incentives Work Inside Constraints

A person cannot respond to an incentive if the desired action is unavailable.

A scholarship may encourage study, but not if the learner lacks access to the required programme. A productivity bonus may not change output if the bottleneck is a broken machine. A health incentive may fail if the healthy option is unaffordable or inaccessible.

Motivation, capability and opportunity need to be separated.

12. Institutions Are Made Partly of Incentives

Rules change expected consequences. They make some behaviour easier, safer or more rewarding and other behaviour more costly.

This is why governance reform often involves changing incentives as well as writing new rules. The World Bank’s governance framework emphasises that durable change can require shifting the incentives, preferences and contestability of actors with power.

13. Education Should Use Incentives to Build Independence, Not Dependence

External rewards can be useful during early behaviour formation, but the educational target is not a child who acts only while the reward remains visible.

Good educational incentives gradually connect behaviour to competence, responsibility, progress and meaningful goals. The learner should increasingly understand why the behaviour matters and be able to continue without constant external control.

The agency gate matters: incentives should support better choices, not manipulate the child into dependence on approval or fear.

The Whole Incentive Chain

Desired outcome → measurable behaviour → reward/cost signal → person interprets signal → choice changes or does not → behaviour produces consequences → gaming/fairness/motivation effects appear → system measures real outcome → incentive is retained, repaired or removed.

A Useful Metaphor: Incentives Tilt the Playing Field

An incentive makes some routes feel more downhill and others more uphill.

But tilting the field does not teach the player how to run, remove a locked gate or guarantee that the chosen direction is good. Incentives change the local slope. The rest of the system still matters.

Incentives at Three Zoom Levels

Micro: one choice

How does this reward or cost change the person’s immediate decision?

Meso: one organisation

Which behaviours are actually rewarded, punished and measured across repeated work?

Macro: institutions and society

How do laws, prices, taxes, status, norms and access shape the large-scale behaviour of people and organisations?

How Incentives Fail

Wrong target: the incentive rewards a proxy rather than the real outcome.
Capability blindness: motivation is increased when the real bottleneck is skill, tools or authority.
Gaming: people optimise the measurement without producing the intended value.
Crowding out: external control weakens an existing internal motive.
Unfairness: the rule creates resentment or disengagement because opportunities or starting conditions differ.
Delay: the reward or consequence is too distant to influence the current choice strongly.
Dependence: the behaviour disappears as soon as the external reward is removed.

How Incentives Are Repaired

Return to the actual outcome. Check whether the problem is motivational. Measure unintended behaviour. Shorten the feedback delay. Make the rule understandable and reasonably fair. Use several indicators when one metric can be gamed. Reduce external control as internal capability and responsibility grow.

Sometimes the correct repair is not a stronger incentive. It is better information, a better environment or removal of the real constraint.

What Parents and Students Should Notice

What exact behaviour is being rewarded?
Does that behaviour actually produce the outcome we care about?
Is the student capable of the behaviour?
What happens when the reward disappears?
Does the incentive strengthen responsibility or create approval dependence?
What gaming or shortcut behaviour has appeared?
Can progress itself become a meaningful reward?

Incentives Change Margins, Not Whole Personalities

Economic incentive analysis is often about the margin: how a small change in reward, cost, probability or rule changes one choice relative to another.

A transport subsidy may alter whether some commuters switch mode. A bonus may change how much effort is allocated to one measured task. It does not imply that the person’s entire character or value system has changed.

Incentives usually move the local decision boundary before they change the person.

Expected Incentives Depend on Probability as Well as Size

A large penalty that is almost never enforced may affect behaviour less than a smaller penalty that is highly likely and immediate. A bonus with unclear eligibility may carry less motivational weight than its headline amount suggests.

A useful simplified structure is:

Perceived consequence × perceived probability × personal value × timing → behavioural pressure.

People do not always calculate these components accurately, but the components still help diagnose why an official incentive is weak in practice.

Thresholds Create Cliffs

When a reward changes sharply at one threshold, behaviour can bunch around that line. People just below the threshold have a strong incentive to cross it, while effort beyond the threshold may produce little additional reward.

This can be useful when a minimum standard truly matters, but it can also distort behaviour. Students may focus on one grade boundary. Firms may time activity around eligibility cut-offs. Workers may neglect unmeasured quality once the rewarded threshold is reached.

Designers should inspect what happens just below, at and just above every important incentive cliff.

Goodhart’s Law Is a Warning About Proxy Targets

When a measure becomes important for reward or punishment, people adapt to the measure itself. The correlation between the measure and the underlying goal can then weaken.

This is commonly summarised as Goodhart’s Law: when a measure becomes a target, it can stop being a good measure.

The repair is not to abandon measurement. It is to use multiple indicators, audit the real outcome and keep some information channel that the target cannot easily control.

Multitask Jobs Create Incentive Distortion

Many jobs contain several valuable tasks, but only some are easy to measure. Strong rewards attached to one measurable component can pull effort away from important but less measurable work.

A teacher may be rewarded for examination outcomes while pastoral care, long-term curiosity and collaboration are harder to quantify. A salesperson may be rewarded for revenue while relationship quality and suitability are less visible.

The stronger the reward on one metric, the more carefully inspect what valuable work the metric leaves outside.

Principal–Agent Problems Make Incentives Necessary — and Dangerous

A principal delegates work to an agent but cannot perfectly observe the agent’s effort, information or goals. Incentive contracts try to align the agent’s behaviour with the principal’s objective.

The problem is that the observable proxy may not perfectly represent the principal’s true goal. Stronger incentives can therefore increase both desired effort and undesirable gaming.

Good principal–agent design combines incentives with selection, trust, monitoring, professional norms and clear accountability rather than expecting one contract to encode the whole job.

Adverse Selection Happens Before the Transaction

When people possess private information about their own risk, quality or intentions, the terms offered can change who chooses to participate.

An insurance price, admissions rule, warranty or employment contract can attract a different population after the incentive changes. The resulting outcome may therefore reflect selection into the system, not only changed behaviour after entry.

This distinction is essential: Who joined? What did participants do afterward? Those are separate causal questions.

Moral Hazard Happens After Protection or Delegation Changes the Consequence

When someone is protected from part of the downside, their behaviour can change because they no longer bear the full consequence of risk.

Insurance can change precaution. A financial guarantee can change lending behaviour. A student repeatedly rescued from missed deadlines may learn that the practical cost of delay is lower than the official rule suggests.

The answer is not to remove all protection. It is to design protection without erasing every feedback signal needed for responsible behaviour.

Tournament Incentives Change Cooperation

Rankings and winner-take-more systems reward relative performance rather than absolute contribution. They can increase effort, especially when outcomes are measurable and participants are comparable.

But they can also reduce knowledge sharing, encourage sabotage or make people avoid difficult assignments that could lower rank.

Before using rankings, ask whether the larger system needs cooperation among the same people it is asking to compete.

Relative Rewards Can Matter More Than Absolute Rewards

People often compare their treatment with peers. A pay rise can feel negative if a close peer receives much more under a rule perceived as unfair. A student may care about class rank even when their own score improved.

This does not mean relative comparison always dominates. It means incentive design should account for the social reference group as part of how the reward will be interpreted.

Incentives Can Change Who Owns the Goal

External incentives can support behaviour while the person is still learning why the behaviour matters. But if every action is externally priced, the person may learn to ask “What do I get for doing this?” rather than “Is this something I should own?”

In education and parenting, the long-run design target is often transfer of regulation:

External cue/reward → understood purpose → experienced competence → growing self-monitoring → reduced external control → internal responsibility.

Crowding Out Is Conditional, Not Automatic

External rewards do not always destroy intrinsic motivation. Effects depend on task, framing, autonomy, prior motivation and whether the reward is experienced as informational, appreciative or controlling.

A reward that recognises competence can have a different effect from one that communicates surveillance or pressure. The correct question is therefore not “extrinsic or intrinsic?” but what meaning does the incentive add to this activity for this receiver?

Incentive Effects Can Decay or Reverse Over Time

People learn the rule, adapt to it and sometimes find ways around it. A new bonus may create an initial response that weakens once it becomes expected. A temporary subsidy can create dependence if withdrawal is not planned.

Evaluation therefore needs a time dimension: immediate response, adaptation period, long-run behaviour and what happens after the incentive changes or disappears.

System-Level Incentives Can Produce Locally Rational but Globally Bad Behaviour

Each actor can respond rationally to their local incentive while the combined outcome is harmful. Departments protect their own budgets, firms externalise pollution, schools optimise rankings, or users overconsume a subsidised resource.

This is why incentive design must zoom out from the individual choice to the network of interacting responses.

Locally rational action can aggregate into globally irrational system behaviour.

A High-Resolution Incentive Audit

Real outcome: What result actually matters?
Target behaviour: Which observable action is being encouraged or discouraged?
Control: Can the actor actually change that behaviour?
Margin: Which decision boundary should move?
Value: Does the reward or penalty matter to this receiver?
Probability: How likely is the consequence to occur?
Timing: How delayed is the consequence?
Threshold: Does the rule create cliffs or bunching?
Proxy: How well does the metric represent the real job?
Multitasking: What valuable work sits outside the rewarded metric?
Principal–agent: What information or goal mismatch exists between owner and actor?
Selection: Will the rule change who enters or stays in the system?
Moral hazard: Does protection remove useful consequence feedback?
Competition: Does relative reward damage needed cooperation?
Fairness: Which comparison group shapes acceptance?
Meaning: Is the signal interpreted as appreciation, price, threat, control or legitimate rule?
Intrinsic motive: What existing purpose, identity or responsibility could be supported or crowded out?
Gaming: How can the measured rule be satisfied without producing the intended outcome?
Adaptation: What changes after people learn the incentive system?
Withdrawal: What happens when the reward, subsidy or penalty disappears?
System effect: What happens when everyone responds at once?
Agency: Does the design increase long-run self-regulation or dependence?

Connect Incentives to the Wider eduKateSG Mechanism Estate

How Motivation Works — why incentives are only one input into starting, persistence and stopping.
How Cooperation Works — how local rewards can strengthen or damage collective outcomes.
How Responsibility Works — how external control should eventually transfer toward internal ownership.
How Markets Work — where prices are themselves incentive and information signals.
How Rules Work — how formal expectations become credible through enforcement, legitimacy and repeated consequence.

Continue Through eduKateSG

Evidence and Further Reading

The OECD’s Behavioural Insights and Organisations reviews why work behaviour depends on interacting factors rather than rewards alone and discusses both the power and hidden costs of incentives. Its earlier Applying Behavioural Insights to Organisations stresses diagnosing the behavioural barrier before strengthening an incentive. The World Bank’s World Development Report 2017: Governance and the Law adds the institutional layer, showing how incentives interact with power, coordination and contestability.

Frequently Asked Questions

Do rewards motivate children?

They can influence behaviour, especially when the action is clear and the reward matters to the child. But rewards do not automatically build capability, and poorly designed rewards can create dependence or shift attention toward the prize rather than the underlying purpose.

Are punishments incentives?

Yes. Penalties are negative incentives because they increase the expected cost of an action. Their effect depends on credibility, timing, fairness and how people interpret the rule.

Why do incentive systems get gamed?

Because people respond to the measurable rule. If the metric only partly represents the real job, rational adaptation can improve the number without improving the underlying outcome.

Final compression: Incentives work by changing the local payoff around behaviour, but good systems measure whether the real outcome improved—and whether motivation, fairness, capability and agency survived the attempt to change behaviour.

How Incentives Work | How Rewards, Costs and Signals Change Behaviour