Incentives: How Do People Respond to Prices and Rewards?

A One Thousand Won Discount for Orders Before Ten

Suppose the cafe you always go to has put up a sign starting today that says, "If you order before ten in the morning, we will cut one thousand won off the price." Some people leave the house ten minutes earlier. Other people come out as usual and pay the extra one thousand won. There are also people for whom one thousand won is not large enough to change their day.

Similar scenes are everywhere. A bonus is given to someone who has worked a lot. If you reduce trash, the fee is cut. There is even a classroom that gives prizes as much as you have read books.

The same question follows these scenes. When prices or rewards change, do people really change their actions? If they change, do they change in the direction hoped for by the person who made that method? What happens after the reward disappears?

We place side by side research where rewards changed actions and research where they did not. This article explains general concepts and does not advise you to do anything.

When Prices Change, What You Give Up Changes

A mechanism designed to change what you gain or lose when you take a certain action is called an incentive. Increasing what you gain is a reward, and increasing what you lose is a cost. Rewards are often compared to carrots, and costs to whips.

Incentives are counted as the fundamental driving force of economic analysis. Money is the most common one, but factors other than money also move choices (DOI: 10.1093/actrade/9780198754992.003.0002 · A chapter from an introduction to behavioral economics).

The simplest path by which incentives change behavior is by changing the value of what you give up. We revisit the concept covered in "Scarcity and Opportunity Cost: What Do You Give Up When You Choose?". The value of the best among the alternatives you gave up because of a certain choice is called opportunity cost. When a cafe discount is attached, the opportunity cost of the choice to order after ten o'clock goes up by one thousand won. People who choose the side where the value went up tend to decrease.

Up to this point, the story is that if the price goes up you choose less, and if it goes down you choose more. The problem starts from the fact that people do not move by looking at price alone.

Rewards Outside the Work and the Mind Inside It

The mindset of doing a task because of rewards given from outside the work is called extrinsic motivation. The mindset of doing a task because the work itself is fun or meaningful is called intrinsic motivation. The feeling of going to work to get a salary is the former, and a hobby you do so happily that you lose track of time is the latter.

Both mindsets can coexist within a person. In a meta-analysis gathering 40 years of research (183 samples), intrinsic motivation predicted performance whether rewards were present or not. Rewards predicted the quantity of work better, while intrinsic motivation predicted the quality of work better. The authors conclude that the two do not merely oppose each other (DOI: 10.1037/a0035661).

People are moved by rewards alone, and by the joy of the work alone. And whether one side affects the other is the question of the next section.

When Rewards Change the Meaning of Work

When an external reward comes in and inner motivation decreases, this is called motivation crowding out. The intuition behind this concept is that if you start giving a child who loves drawing some amount of money per sheet, an activity done because they enjoy it can change into one done to get paid.

Crowding out is one type of unexpected consequence that rewards produce. A result caused by an incentive that the designer did not want or expect is called an unintended consequence.

One side of the literature views that crowding out happens. In a meta-analysis combining 128 experiments, rewards given for doing, completing, or doing well on a task reduced the extent to which people voluntarily did that task in their free time. Self-reported interest also decreased in some reward conditions, while positive feedback increased both free-time choice and interest (DOI: 10.1037/0033-2909.125.6.627). Authors of a review from economics view that there is a lot of evidence showing external interventions sometimes reduce inner motivation and under other conditions increase it. They add that in some cases crowding out can be larger than the standard effect of changing prices (DOI: 10.1111/1467-6419.00150). There are also laboratory results viewing this effect from the side of reputation. The more visible an altruistic action is to others, the greater the likelihood that money will crowd out the desire to gain a good reputation, backfiring (DOI: 10.1257/aer.99.1.544). What decreases here is not the fun of the work, but a different motivation called reputation. This is a broad use by the authors.

The other side of the literature asks to what extent this result holds up. One meta-analysis gathering studies from industrial and organizational psychology journals reported that reward effects appeared only when measured by free time, whereas when measured by work performance, the two effects added together (DOI: 10.1111/j.2044-8325.1992.tb00488.x). This is a result that clashes with the previous 128-experiment meta-analysis which found that self-reported interest also decreased. In a study reviewing over 100 prior tests and adding field experiments, output, productivity, and turnover indicators fit the standard economic model better than the crowding-out model. The authors also add more speculative evidence that disappointment arising from the disappearance of a promised reward can harm quality (DOI: 10.1162/rest_a_00947).

Scholars' stances are divided as well. Based on a book blurb, Cameron and Pierce argue in their review that rewards are not widely harmful and, when carefully deployed, enhance motivation, performance, and interest (DOI: 10.5040/9798216008569 · book blurb). In a response abstract, the authors of the opposing meta-analysis state that they sorted out when rewards reduce inner motivation, when they leave it alone, and when they increase it. At the same time, they point out flaws in Cameron's meta-analysis (DOI: 10.3102/00346543071001043 · response abstract).

Results are split in real-world cases as well. In Sweden, people randomly exposed to receiving a monetary reward for their first COVID-19 vaccination were tracked using national administrative records. A pre-registered study reported no negative outcomes on subsequent vaccination, morals, trust, or safety perceptions (DOI: 10.1038/s41586-022-05512-4).

In a survey experiment with 513 adults hesitant to get vaccinated who were reminded of a reward, the result was different. About one in seven participants (in the abstract's phrasing) said they would get the shot without a reward, but refused after being told to think about a reward. Vaccine safety perceptions fell by 9 percentage points, and pro-social attitudes fell by 10 percentage points. The authors view this as an unintended consequence of rewards (DOI: 10.1002/hec.70061). The two studies have different designs. The Swedish study tracked people exposed to actual reward provision using administrative records, while the survey experiment prompted participants to think about rewards within a survey.

There are at least thirteen behavioral economics models explaining the phenomenon where altruistic actions decrease because of rewards. All of them explain that decrease as a rational choice that maximizes utility (DOI: 10.1111/joes.12606 · model review). How important this effect is in reality remains debated in both economics and psychology (DOI: 10.1111/joes.12059 · review comparing the two fields).

This is as far as the literature goes. There are results where rewards crowded out inner motivation, results where they did not, and there is no consensus on which is how common.

When People Chase the Metric

A number set to measure the outcome of work is called a performance metric. A system where pay depends on the measured outcome is called a performance-related pay system. Think of a call center where customer service agents are evaluated by the number of calls they handle. The number of calls goes up the faster you end conversations. Making the number go up and actually solving the customer's problem are not the same thing.

Shifting behavior to make the metric go up is called Goodhart's Law behavior. It becomes a problem when the metric drifts away from the outcome it was meant to measure. Because a metric only measures a part of the outcome, the unmeasured parts are left outside the reward.

Let us look at the observed side first. The Swiss case is an extreme example of Goodhart's Law behavior where the records themselves were falsified. In 2012, Switzerland changed its hospital payment system to a fixed amount per patient case. The authors estimate that 14 to 27 percent of the birth weights of low-birth-weight infants near a specific threshold were manipulated afterward. They suggest that the larger the reward, the more the manipulation increased. This is an estimate comparing administrative data and there is no experiment (DOI: 10.1002/hec.3991). In South Africa, there is a study that examined seven universities using a system that paid money per research paper through policy analysis, surveys, and interviews. The authors argue that this system can lead to a focus on quantity, an increase in predatory journals, and resentment among researchers (DOI: 10.25159/1947-9417/6688).

There is also the other side. A World Bank working paper reviewed 110 studies on public sector performance-related pay. Among these, 65 studies showed positive effects, and out of 68 rigorous studies, 46 were positive. Sixty-six of the 68 rigorous studies dealt with tasks where outcomes are easy to measure visually, such as education, healthcare, and tax collection. For areas where work is complex and outcomes are hard to measure, such as general administration, there is a lack of evidence to draw a conclusion. Although there are several observational studies pointing out Goodhart's Law behavior and unintended consequences, whether overall productivity decreased remains unclear. There are few long-term tracking studies. Therefore, the authors note two possibilities. The positive results might be an effect caused by the fact that people are being observed, or Goodhart's Law behavior might increase over time as people grow accustomed to the system (DOI: 10.1596/1813-9450-6043).

The scope of what the literature tells us stops here. Performance-related pay has accumulated many positive results in tasks where outcomes are easy to measure. How much Goodhart's Law behavior actually chips away at overall performance is not established in this literature.

Incentives Beyond Money

A mechanism that changes what you gain or lose through non-monetary things—this is called a non-monetary incentive. Words of recognition when you do a good job are an example of this.

In a retail field experiment, monetary compensation and recognition each raised performance by roughly 13 percentage points. When given together, the effects did not add up but substituted for each other. The author reads this negative interaction as evidence that financial and social recognition motives crowd each other out (DOI: 10.2308/accr-51148). What shrinks here between the two is not the joy of the work, but the motivation between the rewards. This is the author's broad usage.

In an 8-month field experiment involving 422 households in Singapore, households that received only feedback, goals, and saving tips reduced their electricity use by 16%. Households given extra monetary rewards or donations did not reduce it further (DOI: 10.1007/s10640-025-00973-3).

Defaults are another tool that is distinct from incentives. Defaults do not change what you gain or lose; they change the design of the choice window. The option pre-selected when you choose nothing—this is called a default. This is the case when a shipping method is pre-selected in an online order window. In an e-commerce field experiment (working paper), choices increased the most when eco-friendly shipping was the default. Small monetary rewards and information only increased them slightly. The authors interpret this result as matching behavior where people choose without deliberating long at the checkout counter (DOI: 10.2139/ssrn.7154300).

It Depends on Who and How It Is Given

When you gather cases where the effects diverged, four differences stand out. They are how rewards are tied to performance, who receives them, surrounding norms, and the size of the reward along with preparation. This is a summary by this article, and each study was not tested against these criteria.

First is how rewards are tied to performance. In the meta-analysis seen earlier, when rewards were tied directly to performance, the share of performance explained by intrinsic motivation shrank, and when they were tied indirectly, that share was large (DOI: 10.1037/a0035661).

Second is who receives the recipient. In a survey experiment targeting local government employees in Florida, USA, financial rewards and promotion opportunities moved employees with low a sense of mission toward public service. As the sense of mission increased, the effect of financial rewards converged to zero, and the effect of promotion opportunities became negative. The authors read this as motivation crowding out (DOI: 10.1177/00953997211025983).

Third is surrounding norms. In a preregistered analysis (preprint) using blood donation data from 28 European countries, the more positive the norms around rewards were, the higher the association between rewards and altruistic behavior was (arXiv:2104.13652v1).

Fourth is the size of the reward and the recipient's preparation. There is an experiment where a large reward showed lower performance than a small reward. In the same design, when participants chose tasks themselves or practiced first, the average performance of the large reward was higher. The authors view this difference as the result of an increase in the proportion of people who respond positively to a large reward (DOI: 10.1007/s10683-024-09841-1).

Holding a Yardstick to a Single Shop's Morning

Let us return to the café. The one-thousand-won discount raises the opportunity cost of orders placed after ten o'clock by one thousand won. If some customers shift their timing, that is the simplest picture of incentives changing prices to change behavior.

Now suppose the shop gives the clerk a bonus based on the number of orders before ten o'clock. The number of orders before ten o'clock is a performance metric, and this bonus is a performance-based pay. If the clerk rushes only the customers standing in line to boost the count and cuts back on chatting, that is gaming the metric. For a person who became a clerk because they liked talking with people, the bonus might crowd out that feeling, or it might not. As we saw earlier, this is a point where the literature is divided.

A clerk's work is a mix of parts that are easy to measure with counts and parts that are hard to measure. Because the prior literature dealt with the public sector, we cannot know which applies to the case of a café clerk.

Put to the Test in Reality

We gather four ways the literature has examined how the concept manifests in reality.

The first is rewards that shift time. Users randomly selected from the Singapore subway were given rewards for 10 weeks. The method refunded the full fare if they exited the turnstiles before the morning rush hour. During the reward period, the likelihood of trips finishing earlier increased by an average of 5.5%. The effect was not large but consistent, and the earlier travel persisted even 7 months later (DOI: 10.1287/mnsc.2017.2731).

The second is health behavior. According to an analytical article introducing a Cochrane review on financial incentives for smoking cessation, the authors of the review conclude that rewards appear to increase quit rates in both moderate-risk groups and high-risk groups such as pregnant women (DOI: 10.20344/amp7363).

The third is students' test scores. In a meta-analysis gathering 21 studies, monetary rewards given to students showed positive effects on overall achievement and mathematics, but no effect on reading and language. Results were mixed on whether the effects persisted after the rewards were removed. Programs provided alongside other academic support tended to show stronger effects (DOI: 10.1177/016146812012200304).

The fourth is an aggregate estimate of the whole. A meta-analysis (preprint) gathering 1,568 performance estimates from economic experiments finds that when publication bias is corrected, the effect of field experiments is zero or very small. Laboratory experiments remain statistically significant even after correction, but are weak. The authors note that this evidence does not align with standard economic models in which rewards boost performance (DOI: 10.31222/osf.io/wbe9k). This result also diverges in direction from the earlier finding that a single field experiment aligned better with the standard model. This article places the two results side by side.

What the literature has confirmed goes only as far as showing that rewards changed behavior in some actions, while in others they failed to change it or moved it in unexpected directions.

Where to Look Next

This article covered how prices and rewards change a single person's behavior. How the choices of many people come together to set prices in the market is covered next in "Supply and Demand: Who Sets the Price?". Cases where prices cannot tell the whole story are covered in "When Markets Fail — The Moments a Price Cannot Tell the Whole Story".

Source Papers

• "2. Motivation and incentives" (2017) — DOI: 10.1093/actrade/9780198754992.003.0002 — A chapter from the primer: Incentives are a fundamental driver of economic analysis, and while money is common, various factors steer choices, and extrinsic motivation can crowd out intrinsic motivation.
• "Intrinsic motivation and extrinsic incentives jointly predict performance: A 40-year meta-analysis" (2014) — DOI: 10.1037/a0035661 — Meta-analysis (183 samples): When rewards are tied directly to performance, the explanatory power of intrinsic motivation decreases, whereas when indirect, it is large; rewards better predict quantity while intrinsic motivation better predicts quality, and the two do not necessarily clash.
• "A meta-analytic review of experiments examining the effects of extrinsic rewards on intrinsic motivation" (1999) — DOI: 10.1037/0033-2909.125.6.627 — Meta-analysis of 128 experiments: Rewards reduced free-choice intrinsic motivation, some rewards also reduced self-reported interest, and positive feedback increased both.
• "Motivation Crowding Theory" (2001) — DOI: 10.1111/1467-6419.00150 — Review: External intervention can reduce intrinsic motivation or, under certain conditions, foster it, and in specific cases, it can surpass the price effect.
• "Doing Good or Doing Well? Image Motivation and Monetary Incentives in Behaving Prosocially" (2009) — DOI: 10.1257/aer.99.1.544 — Experiment: Monetary rewards crowd out image motivation (in the authors' broad usage), and the possibility of backfire is greater in public altruistic activities.
• "The effects of extrinsic rewards in intrinsic motivation: A meta-analysis" (1992) — DOI: 10.1111/j.2044-8325.1992.tb00488.x — Meta-analysis of industrial and organizational psychology journal studies: The overjustification effect appears only in free-time measures, whereas in performance measures, the two effects add up.
• "Do Monetary Incentives Undermine Performance on Intrinsically Enjoyable Tasks? A Field Test" (2022) — DOI: 10.1162/rest_a_00947 — Review of over 100 papers and field experiments: Output, productivity, and turnover align more with the standard model than the crowding-out model, and there is more speculative evidence that unmet compensation expectations can harm quality.
• "Rewards and Intrinsic Motivation" (2002) — DOI: 10.5040/9798216008569 — Based on the book introduction: The authors argue that rewards are not broadly harmful and, when deployed carefully, enhance motivation, performance, and interest.
• "The Pervasive Negative Effects of Rewards on Intrinsic Motivation: Response to" (2001) — DOI: 10.3102/00346543071001043 — Abstract of a response article: Authors of the opposing meta-analysis state that they distinguished when rewards decrease, leave untouched, or increase inner motivation, pointing out flaws in Cameron and Pierce's meta-analysis.
• "Financial incentives for vaccination do not have negative unintended consequences" (2023) — DOI: 10.1038/s41586-022-05512-4 — Swedish pre-registered study: People randomly exposed to the offer of monetary rewards showed no negative outcomes on subsequent vaccination, morality, trust, or safety perceptions.
• "Vaccine Incentives Harm Intrinsic Motivation: Evidence From a Priming Experiment" (2025) — DOI: 10.1002/hec.70061 — Survey experiment of 513 vaccine-hesitant adults: One in seven said they would have gotten vaccinated without a reward but refused after thinking about the reward, alongside a 9 percentage point drop in safety perception and a 10 percentage point drop in prosocial attitude, which the authors view as unintended consequences.
• "Review of behavioral economics models of the altruistic crowding-out effect from monetary incentives" (2023) — DOI: 10.1111/joes.12606 — Review: There are at least 13 models explaining the altruistic crowding-out effect, and all explain the reduction as rational utility maximization.
• "Theory and Evidence in Psychology and Economics about Motivation Crowding Out: A Possible Convergence?" (2014) — DOI: 10.1111/joes.12059 — Review: The real-world importance of crowding out is debated in economics and psychology.
• "Losing body weight for money: How provider-side financial incentives cause weight loss in Swiss low-birth-weight newborns" (2020) — DOI: 10.1002/hec.3991 — Swiss newborns: It is estimated that 14% to 27% of birth weights near the threshold were manipulated, and larger rewards increased manipulation.
• "The Unintended Consequences of Using Direct Incentives to Drive the Complex Task of Research Dissemination" (2020) — DOI: 10.25159/1947-9417/6688 — Seven South African universities: The authors argue that payouts per paper can breed a focus on quantity, predatory journals, and resentment.
• "Performance-Related Pay in the Public Sector: A Review of Theory and Evidence" (2012) — DOI: 10.1596/1813-9450-6043 — Working paper, 110 papers: 65 are positive, and among the 68 rigorous ones, 46 are positive—of which 66 involve easily measurable outcomes—while evidence for complex tasks is scarce, metric-gaming may increase over time, and the net effect is unclear.
• "Monetary Incentives, Feedback, and Recognition—Complements or Substitutes? Evidence from a Field Experiment in a Retail Services Company" (2015) — DOI: 10.2308/accr-51148 — Monetary rewards and recognition each yield an effect of about 13 percentage points, but they do not combine and are substitutes, which the author reads as a crowding out between motivations (in the broad sense).
• "Switching Off: Energy Saving Goals Outshine Incentives—Evidence from a Field Experiment" (2025) — DOI: 10.1007/s10640-025-00973-3 — 422 households in Singapore: Feedback, goals, and tips cut electricity use by 16%, while monetary or donation rewards yielded no additional savings.
• "Decide or Default? How Defaults, Information, and Financial Incentives Influence Sustainable Delivery Choices in a Field Experiment" (2026) — DOI: 10.2139/ssrn.7154300 — Working paper: Eco-friendly defaults are most effective, while small monetary rewards and information increased choices slightly.
• "The Effectiveness of Monetary and Promotion Rewards in the Public Sector and the Moderating Effect of PSM" (2021) — DOI: 10.1177/00953997211025983 — Survey experiment of Florida public servants: The higher the public service motivation, the more the monetary effect converges to zero, and the promotion effect turns negative.
• "Social Norms Offer Explanation for Inconsistent Effects of Incentives on Prosocial Behavior" (2021) — arXiv:2104.13652v1 — Preprint, 28-country blood donation: The more positive the norm surrounding rewards, the more strongly rewards are linked to altruistic behavior.
• "The roles of selection and practice in mitigating negative responses to high-powered incentives" (2024) — DOI: 10.1007/s10683-024-09841-1 — Experiment: While large rewards sometimes lower performance, when individuals self-select or practice, the average performance under large rewards is higher.
• "Temporary Incentives Change Daily Routines: Evidence from a Field Experiment on Singapore's Subways" (2018) — DOI: 10.1287/mnsc.2017.2731 — 10-week rebate: 5.5% increase in early travel, persisting even after 7 months.
• "Analysis of the Cochrane Review: Incentives for Smoking Cessation" (2016) — DOI: 10.20344/amp7363 — Analysis of Cochrane review: Rewards appear to increase smoking cessation rates in both moderate-risk groups and high-risk groups such as pregnant women.
• "Do Student-Level Incentives Increase Student Achievement? A Review of the Effect of Monetary Incentives on Test Performance" (2020) — DOI: 10.1177/016146812012200304 — Meta-analysis of 21 studies: Overall results and mathematics are positive, reading shows no effect, and persistence after withdrawal is mixed.
• "Financial Incentives and Performance: A Meta-Analysis of Economics Evidence" (2022) — DOI: 10.31222/osf.io/wbe9k — Preprint, 1,568 economics experiment performance estimates: After correction, field experiment effects are zero or very small, laboratory effects are weak, and results do not match the standard model.