Incentive-Caused Bias Mental Model… | Faster Than Normal
Psychology & Behavior
Incentive-Caused Bias
People with incentives to believe something will find ways to believe it — incentive structures shape behaviour more reliably than values, culture, or stated intentions.
Model #0142Category: Psychology & BehaviorSource: Charlie MungerDepth to apply:
Show me the incentive and I will show you the outcome. Incentive-caused bias is the tendency for human cognition to warp — unconsciously, automatically, predictably — in the direction of whatever reward or punishment structure a person operates within. It is not corruption. It is not dishonesty. It is something more dangerous than both: sincere belief, shaped by self-interest, that the person holding it genuinely cannot distinguish from objective analysis.
Charlie Munger ranked it among the most powerful of his 25 causes of human misjudgment. "Never, ever, think about something else when you should be thinking about the power of incentives," he told a Harvard audience in 1995. He wasn't being rhetorical. He was issuing a warning that most people hear, nod at, and then proceed to violate within the hour — because the bias operates beneath the threshold of awareness. The mortgage broker who originates a $500,000 loan for a borrower who cannot afford it does not think "I am destroying this family's financial future for my commission." The broker thinks "this borrower's situation is more manageable than it appears, and rates will probably come down." The incentive doesn't corrupt the reasoning. It becomes the reasoning.
The experimental evidence is robust. A 2012 study in the Journal of the American Medical Association found that physicians who received meals from pharmaceutical companies — not cash payments, not luxury trips, simply meals averaging $20 in value — were significantly more likely to prescribe the promoted brand-name drug over generics. The physicians studied would have vigorously denied that a $20 lunch influenced their clinical judgment. The prescribing data said otherwise. The gap between what people believe about their own objectivity and what incentive structures actually produce is one of the most documented findings in behavioral science.
Munger's favourite illustration was FedEx. In the early years, the company's central sorting hub in Memphis couldn't get packages transferred between planes fast enough. Teams worked shifts, and the packages moved slowly. Management tried every operational fix they could design — new procedures, better training, supervisory oversight. Nothing worked. Then someone changed the pay structure from hourly wages to per-shift completion: finish the sort and you go home with full pay. The problem vanished overnight. The same workers, the same packages, the same facility. The only variable that changed was the incentive. Munger used the story for decades because it demonstrated the point with zero ambiguity: the workers weren't lazy before. They were rational. The system was paying them to work slowly, and they obliged — not through conscious calculation, but through the invisible cognitive adjustment that incentive structures produce in every human being.
Upton Sinclair captured the mechanism in a single sentence in 1935: "It is difficult to get a man to understand something when his salary depends upon his not understanding it." The sentence survives ninety years later because it describes something everyone recognises in others but almost nobody detects in themselves. The surgeon who recommends surgery. The consultant who recommends more consulting. The auditor who approves the books of the client who pays their fees. None of these people are necessarily dishonest. Many are deeply principled. The incentive doesn't require dishonesty to distort judgment. It only requires a human brain.
The bias scales from individuals to institutions to entire economies. The 2008 financial crisis was not primarily a failure of financial engineering — it was a failure of incentive architecture. Mortgage originators earned fees per loan originated, with no retention of credit risk. Rating agencies earned fees from the banks whose securities they rated. Traders earned bonuses on annual mark-to-market gains that evaporated over longer time horizons. At every node in the system, the incentive pointed toward volume, complexity, and short-term profit. At no node did the incentive point toward the question that would have prevented $22 trillion in household wealth destruction: "What happens if housing prices fall?"
The Wells Fargo scandal of 2016 provided the domestic sequel. Retail branch employees, pressed by cross-selling quotas that determined their take-home pay and job security, opened 3.5 million fake customer accounts over a period of fourteen years. The executives who designed the quota system weren't trying to incentivize fraud. They were trying to incentivize "deep customer relationships." But the metric they chose — products per household, target of eight — was so disconnected from genuine customer need that the shortest path to the reward was fabrication. The incentive architecture produced the behavior it was designed to measure, while destroying the value it was designed to create.
Section 2
How to See It
The defining feature of incentive-caused bias is that the person exhibiting it is the last one to recognise it. External observers can map the distortion with startling clarity. The person inside the incentive structure experiences their own judgment as objective, careful, even conservative.
The diagnostic is simple in principle and difficult in practice: look at the conclusion, then look at the incentive, and ask whether the conclusion would survive if the incentive were removed. If you can't answer that question honestly — and the model predicts you can't, when you're the one inside the structure — then you need external input from someone whose incentives point in a different direction.
Finance
You're seeing Incentive-Caused Bias when an auditor signs off on aggressive accounting at a client that represents 30% of their firm's revenue. Arthur Andersen's auditors didn't wake up one morning and decide to help Enron fabricate earnings. The relationship had generated over $50 million in annual fees by 2001. Each year's approval made the next year's approval easier to justify — the confirming precedent accumulated, the economic dependency deepened, and the cognitive distortion compounded. By the time the fraud was undeniable, the auditors' judgment had been shaped by a decade of incentive exposure that no amount of professional ethics training could overcome. Arthur Andersen — an 89-year-old firm with 85,000 employees — ceased to exist within months.
Healthcare
You're seeing Incentive-Caused Bias when a healthcare system performs surgery at rates that correlate more tightly with reimbursement structures than with patient pathology. Dartmouth Atlas data has shown since the 1990s that Medicare spending per beneficiary varies by a factor of three across U.S. hospital referral regions, with no corresponding variation in patient outcomes. Regions with more surgeons perform more surgeries. Regions with more hospital beds fill more hospital beds. The physicians in high-utilisation regions aren't fraudulent — they genuinely believe their patients need the procedures. The incentive structure has shaped their clinical judgment so thoroughly that the distortion is invisible from the inside.
Technology
You're seeing Incentive-Caused Bias when a social media platform's internal research identifies harm to teenage users and the product team reframes the finding as "an engagement opportunity." Facebook's own researchers documented in 2019 that Instagram made body image issues worse for one in three teenage girls. The internal presentation was titled "We make body image issues worse for one in three teen girls." The product response was not to reduce the harmful features but to explore age-appropriate content filters that preserved engagement metrics — because the team's incentives were tied to daily active users, not to adolescent wellbeing. The researchers saw the problem clearly. The product team, operating under a different incentive structure, could not.
Legal
You're seeing Incentive-Caused Bias when a law firm billing by the hour consistently estimates that cases will require more time than fixed-fee competitors predict for identical work. A 2014 Georgetown Law study found that matters handled on contingency or fixed-fee arrangements resolved 25–40% faster than equivalent matters billed hourly. The hourly-billing attorneys weren't padding intentionally. They were conducting "thorough" analysis — an adjective that, when your revenue correlates with hours spent, takes on a meaning that has less to do with quality than with compensation structure.
Section 3
How to Use It
The primary application of understanding incentive-caused bias isn't resisting your own incentives — though that matters. It's two things: designing systems where the incentives produce the behavior you actually want, and reading other people's outputs through the lens of what they're paid to conclude.
The first application is about architecture. The second is about interpretation. Both require discipline, and the model predicts that you'll underperform on both — because your own incentives will distort your assessment of how well you're doing.
Decision filter
"Before trusting any recommendation, ask: how does the person making it get paid? If their compensation correlates with one particular conclusion, discount that conclusion proportionally — not because they're dishonest, but because they're human."
As a leader
Audit your organisation's incentive structures with the same rigour you audit its financial statements. Wells Fargo's cross-selling scandal of 2016 — where employees created 3.5 million fake accounts — was not a failure of individual ethics. It was a failure of incentive design. Employees faced aggressive sales quotas that determined their income and job security. The quota was eight products per customer, a target so disconnected from actual customer need that the only way to meet it was fabrication. The fix was structural: eliminate the cross-selling metric, redesign compensation around customer satisfaction, and remove the incentive that produced the behaviour.
As an investor
Map every recommender's incentive structure before weighting their opinion. Sell-side analysts rated "Buy" on 85% of covered stocks in the pre-Spitzer era (pre-2003), because their compensation was tied to investment banking revenue, not analytical accuracy. The Global Research Analyst Settlement of 2003 — which fined ten major banks $1.4 billion — didn't reveal that analysts were lying. It revealed that their incentive structures had made objectivity structurally impossible. Today, ask the same question of any source: venture capitalists talking their book, management teams guiding to beats, and index fund managers dismissing active management.
As a decision-maker
When designing incentives, always ask: "What behavior am I actually rewarding, and what is the easiest way to game it?" Soviet factory managers responded to tonnage quotas by producing unusably heavy goods and to unit quotas by producing unusably small ones. The planners wanted production. The incentive rewarded a metric. The gap between the two is where incentive-caused bias lives — and closing it requires anticipating the cognitive distortion that any metric-linked reward will produce.
Common misapplication: Cynicism masquerading as insight. Some people, having learned about incentive-caused bias, begin dismissing every recommendation from every compensated professional. Your doctor isn't necessarily wrong because they earn more from surgery than from watchful waiting. Your financial advisor's recommendation isn't automatically suspect because they earn a commission.
The model doesn't say incentives determine conclusions — it says they bias them, which means you adjust the weight of the recommendation, not reject it entirely. The surgeon who recommends surgery might be right. The question is whether you've sought a second opinion from someone with a different incentive structure. A fee-only financial advisor and a commission-based advisor who reach the same conclusion are providing much stronger evidence than either alone — because their incentive structures would push them in different directions if the recommendation weren't genuinely sound.
Section 4
The Mechanism
Section 5
Founders & Leaders in Action
The leaders who navigate incentive-caused bias most effectively share a common pattern: they treat incentive design as a first-order strategic problem, not an HR afterthought. They understand that compensation structures don't just reward behavior — they shape cognition.
The pattern across these four cases spans retail, technology, finance, and conglomerate management across five decades. Each leader recognised — usually after observing a catastrophic failure in another organisation — that moral exhortation is insufficient defence against incentive-caused bias. Each responded by redesigning the incentive architecture itself, rather than asking people to resist incentives through willpower. The results speak to the model's core prediction: change the structure, and the cognition follows.
Charlie MungerVice Chairman, Berkshire Hathaway, 1978–2023
Munger spent six decades studying incentive-caused bias and concluded it was the single most underestimated force in human affairs. His central insight — that incentives don't merely motivate behavior but actually reshape belief — informed every major Berkshire Hathaway investment decision from the mid-1970s onward.
When evaluating potential acquisitions, Munger systematically mapped the incentive structures of the target company's management, competitors, and regulators. His analysis of the Salomon Brothers crisis in 1991 — when Buffett had to step in as interim chairman after a Treasury bond bidding scandal — traced the fraud directly to a compensation structure that paid traders enormous bonuses on short-term profits with no clawback for long-term losses. The traders weren't uniquely immoral. They were operating inside a system that paid them to take risks they would never personally bear.
Munger's prescription was structural, not moral. He advocated for compensation designs that align the agent's incentives with the principal's outcomes over long time horizons — stock grants with multi-year vesting, deferred compensation tied to business performance, and cultural norms that treat short-term thinking as a character deficiency. At Berkshire, subsidiary CEOs are paid based on metrics specific to their own operations, not Berkshire's stock price, eliminating the incentive to game portfolio-level numbers they don't control.
His deeper insight: you don't fix incentive-caused bias by lecturing people about ethics. You fix it by redesigning the incentive. The FedEx workers didn't need a motivational seminar. They needed a pay structure that rewarded completion rather than hours. Munger applied the same logic to every system he analysed, from insurance underwriting to judicial appointments to academic tenure.
Bezos built Amazon's compensation architecture around a specific theory of incentive-caused bias: cash bonuses tied to short-term metrics produce short-term thinking. Starting in the late 1990s, Amazon's executive compensation shifted heavily toward restricted stock units vesting over four years, with minimal cash bonuses. The structure was deliberate. A VP earning 80% of their compensation in four-year RSUs has a fundamentally different cognitive orientation than a VP earning 80% in annual cash bonuses.
The design extended to customer-facing metrics. Bezos insisted that Amazon's core incentive metrics — selection, price, and delivery speed — align with customer value rather than extraction. In his 2003 shareholder letter, he explicitly warned against "proxy metrics" that substitute for actual value creation. The distinction matters: a proxy metric incentivizes optimising the proxy, not the underlying value. Bezos designed Amazon's systems to keep the two as close together as possible.
He also eliminated the annual bonus structure for most employees, replacing it with higher base salaries and equity grants. The reasoning was explicit: annual bonuses create annual thinking horizons. A VP deciding whether to invest in a two-year infrastructure project will evaluate it differently if their bonus is determined by this year's operating margin versus their net worth being tied to the company's stock price four years from now.
The contrast with contemporaries is instructive. While many tech companies incentivized growth through aggressive monetisation — pop-up ads, dark patterns, upselling — Amazon's incentive structure rewarded the behaviours that built long-term customer trust. Prime membership, introduced in 2005 at $79/year, was designed so that the company's incentive (retain subscribers) aligned with the customer's incentive (get value from the subscription). By 2024, Prime had over 200 million members globally. The alignment wasn't accidental. It was an incentive architecture choice made two decades earlier.
Walton understood incentive-caused bias instinctively, decades before behavioral economists gave it a formal name. His most consequential insight was that hourly retail workers — the people closest to the customer and most directly responsible for store performance — were incentivized to do the minimum required to retain employment. Their pay was fixed regardless of store profitability. Their behavior reflected precisely that incentive structure.
In 1971, Walton introduced profit sharing for all Walmart associates, tying a portion of each employee's compensation to store performance. The programme was radical for its era — most retailers viewed hourly workers as interchangeable and treated compensation as a cost to minimise rather than a lever to optimise.
By 1991, Walmart's profit-sharing plan held $1.8 billion in assets. Employees who had started as hourly stockers in the 1970s retired with six-figure balances. The behavioral change was measurable: stores with high profit-sharing participation showed lower shrinkage, better customer service scores, and higher same-store sales growth.
The design was not philanthropic — it was strategic. Walton wrote in Made in America (1992) that the programme's cost was more than offset by reduced turnover, reduced theft, and the discretionary effort that employees contributed when they could see the direct financial connection between their work and their wealth. He had changed the incentive, and the cognition followed. Store managers who shared in profits thought about inventory management differently than managers on fixed salaries. The same humans, the same stores — different incentive structures, different outcomes.
Dimon's approach to incentive-caused bias is best understood through contrast with Wells Fargo. Both banks operated massive retail branch networks with millions of customer accounts. Both faced pressure to demonstrate organic growth. The difference was incentive architecture.
Wells Fargo tied branch employee compensation directly to cross-selling metrics — the number of products per customer — with specific daily, weekly, and monthly quotas. By 2013, the target was eight products per household, a number so disconnected from genuine customer need that employees began opening accounts customers hadn't requested. Between 2002 and 2016, employees created approximately 3.5 million fake accounts and 500,000 unauthorized credit card applications. The fine was $3 billion. The CEO resigned. The reputational damage persists a decade later.
JPMorgan under Dimon restructured retail incentive compensation after the 2008 crisis to weight customer retention and satisfaction alongside product sales. Dimon told shareholders in his 2014 annual letter that "you get what you pay for, and if you pay for short-term metrics, you will destroy long-term value." The bank implemented clawback provisions for senior executives, tying deferred compensation to risk-adjusted returns over multi-year periods. When JPMorgan's consumer banking division grew deposits from $443 billion in 2014 to over $1 trillion by 2023, the growth was organic — driven by service quality metrics that the incentive structure had been designed to reward.
The lesson is not that Dimon is morally superior to Wells Fargo's leadership. The lesson is that incentive architecture is a design choice, and the cognitive distortions that follow from that architecture are predictable, measurable, and — if you take incentive-caused bias seriously — preventable.
Section 6
Visual Explanation
Incentive-caused bias operates as a cognitive distortion field between objective reality and the conclusions a person reaches. The distortion is proportional to the strength of the incentive — small incentives produce small distortions, large incentives produce large ones — and the person inside the field cannot detect it because the distortion applies to their self-assessment as well.
The diagram below maps how objective evidence passes through an incentive structure and emerges as a sincere but distorted conclusion. The three pathways — motivated cognition, dissonance reduction, and social validation — operate simultaneously and reinforce each other, producing a final output that the person holding it experiences as independent analysis.
Section 7
Connected Models
Incentive-caused bias sits at the intersection of individual psychology and institutional design. It rarely operates alone. Its connections to other mental models reveal how the distortion propagates — from individual cognition through institutional structures to system-wide failures. Understanding the connections tells you where to look for the early warning signs and where to install the structural correctives.
Reinforces
Confirmation Bias
Incentive-caused bias and confirmation bias form a compound distortion far more dangerous than either alone. When your salary depends on a particular conclusion, your brain selectively seeks, interprets, and remembers evidence supporting that conclusion — the exact mechanism of confirmation bias, now turbocharged by financial motivation.
Sell-side equity analysts provide the clearest demonstration. Research by Hong and Kubik (2003) showed that analysts who issued optimistic forecasts for investment banking clients were significantly more likely to be promoted or move to higher-status firms. The incentive (career advancement) activated confirmation bias (selective evidence processing) to produce systematically inflated earnings estimates. The analysts weren't lying. They were confirming — but the direction of their confirmation was shaped by what they were paid to find. Remove the investment banking incentive, and the confirmation bias has no preferred direction. Attach it, and the direction becomes predictable.
Reinforces
Principal-Agent Problem
The Principal-Agent Problem describes situations where one party (the agent) acts on behalf of another (the principal) but has divergent interests. Incentive-caused bias is the psychological mechanism that makes the divergence invisible to the agent. A corporate board hires a CEO and designs a compensation package emphasising short-term stock performance. The CEO doesn't consciously decide to sacrifice long-term value — incentive-caused bias reshapes their genuine assessment of what constitutes "good strategy" to align with what the incentive rewards.
Michael Jensen and William Meckling formalised the framework in 1976, but their treatment was primarily economic. Munger's contribution was the psychological layer: agents don't just act differently when incentives diverge — they think differently. The misalignment isn't behavioral. It's cognitive. This is why contractual solutions to the principal-agent problem so often fail: you can't contract away a bias the agent sincerely doesn't believe they have.
Section 8
One Key Quote
"It is difficult to get a man to understand something when his salary depends upon his not understanding it."
— Upton Sinclair, I, Candidate for Governor: And How I Got Licked, 1935
Section 9
Analyst's Take
Faster Than Normal — Editorial View
Incentive-caused bias is the silent architecture of most institutional failure. Not incompetence. Not malice. Not even negligence in the traditional sense. Just the predictable cognitive distortion that occurs when human brains are placed inside reward structures that point somewhere other than the truth.
I keep a running list of institutional disasters, and the pattern is monotonous in its consistency. Enron. WorldCom. The 2008 financial crisis. Wells Fargo. Boeing's 737 MAX. Purdue Pharma and the opioid crisis. In every case, the post-mortem reveals the same structural signature: incentives that rewarded a specific outcome, humans whose cognition warped to produce that outcome, and a system that mistook the warped cognition for objective analysis until the gap between belief and reality became catastrophic.
The most dangerous version of this bias operates in plain sight. Nobody hides the compensation structure of a sell-side equity analyst. Nobody conceals that mortgage brokers earn per-loan commissions. Nobody disguises that pharmaceutical sales representatives are paid on prescription volume. The incentive structure is public information. And yet the distortion persists — because the person inside the structure sincerely believes their judgment is independent of it. The transparency of the incentive doesn't neutralise the bias. It just makes the outside observer's failure to adjust for it less forgivable.
The operational framework I apply is borrowed directly from Munger: never ask a barber whether you need a haircut. Before acting on any recommendation, I map the recommender's incentive structure and adjust my confidence accordingly. This isn't cynicism — it's calibration. The surgeon who recommends surgery may well be correct. The financial advisor who recommends a complex product may genuinely believe it's optimal. The question isn't their sincerity. The question is whether their sincerity has been shaped by a compensation structure that makes one conclusion more personally profitable than another.
The most underrated application of this model is in organisational design. Most founders and executives treat compensation design as an administrative function — something HR handles after the strategy is set. This is backwards. Incentive architecture is strategy, because it determines the cognitive environment in which every subsequent decision will be made. Jeff Bezos understood this when he designed Amazon's compensation around long-term RSUs. Sam Walton understood it when he introduced profit sharing in 1971. The leaders who get incentive design right aren't just paying people differently. They're creating different cognitive environments — ones where the natural distortions of incentive-caused bias push in the direction of long-term value rather than against it.
Section 10
Test Yourself
Incentive-caused bias is easiest to see in other people's industries and hardest to see in your own. That asymmetry is itself a product of the bias: you discount the influence of your own incentives while spotting it instantly in others. These scenarios test whether you can identify the mechanism — even when the person exhibiting the bias is acting in apparent good faith and may well be reaching the correct conclusion.
Is this mental model at work here?
Scenario 1
A real estate agent tells a seller that now is a great time to list their home, citing strong comparable sales in the neighborhood. The agent earns a 3% commission on the sale.
Scenario 2
A pharmaceutical company funds a clinical trial that shows their drug is 15% more effective than the existing standard of care. The trial is pre-registered, double-blinded, and published in a peer-reviewed journal with full data transparency.
Scenario 3
A consulting firm recommends that a client undertake a 12-month digital transformation programme. The firm's proposal estimates $4.2 million in fees. An independent technology advisor, paid a flat $15,000 assessment fee, recommends three targeted integrations over 90 days at $380,000.
Scenario 4
Warren Buffett announces at Berkshire Hathaway's annual meeting that he has sold a position, explaining that the thesis no longer holds. He does not disclose what he bought with the proceeds.
Section 11
Top Resources
The essential reading on incentive-caused bias spans behavioral psychology, institutional design, and financial history. Start with Munger for the conceptual framework, then build practical pattern recognition with Levitt's empirical work and Lewis's narrative account of the 2008 crisis. Taleb provides the philosophical foundation for the structural corrective.
The foundational treatment. Munger's taxonomy of 25 cognitive tendencies places incentive-caused bias near the top and illustrates it with the FedEx case, the Salomon Brothers scandal, and decades of Berkshire investment experience. Available in full in Poor Charlie's Almanack. No single resource better integrates the psychology of incentives with practical decision-making.
Levitt's research on real estate agents, sumo wrestlers, and schoolteachers provides empirical evidence for incentive-caused bias across domains. The chapter on real estate agents — showing they sell their own homes for more than they advise clients to accept — is the most accessible demonstration of the model in print.
Taleb's argument that incentive alignment requires bearing the downside of your own recommendations. The book connects incentive-caused bias to risk management, ethics, and institutional design, arguing that systems where decision-makers don't bear consequences inevitably produce distorted judgment.
The narrative account of the 2008 financial crisis, told through the handful of investors who recognised the incentive-caused bias embedded in mortgage origination, securitisation, and rating agency compensation. Lewis traces each node of the crisis to a specific incentive misalignment, demonstrating institutional-scale distortion.
The famous daycare study showing that introducing fines for late pickups increased lateness — because the fine converted a social obligation into a market transaction. When the fine was later removed, lateness remained elevated — the social norm had been permanently displaced. Essential for understanding how incentives reshape the cognitive frame in which people evaluate their own behavior, not just its direction.
Leaders who apply this model
Playbooks and public thinking from people closely associated with this idea.
Incentive-Caused Bias — How compensation structures warp cognition, turning objective analysis into incentive-aligned conclusions without the person's awareness.
Tension
Cobra Effect
The Cobra Effect describes how incentive schemes produce the opposite of their intended outcome. During British colonial rule of India, the government offered a bounty for dead cobras. Entrepreneurs responded by breeding cobras for the bounty revenue. When the bounty was cancelled, breeders released their worthless snakes — increasing the cobra population beyond its original level.
The tension is instructive: incentive-caused bias makes designers blind to the perverse consequences of the incentive schemes they create. The British administrators didn't anticipate cobra breeding because their own incentive (demonstrate progress) biased their evaluation of the policy. The pattern repeats in every domain where policymakers design incentives from within their own incentive-shaped perspective.
Tension
[Goodhart's Law](/mental-models/goodharts-law)
"When a measure becomes a target, it ceases to be a good measure." Charles Goodhart articulated the principle in 1975, describing the downstream consequence of attaching incentives to metrics. Incentive-caused bias is the cognitive mechanism: once a metric carries compensation weight, people stop seeing the metric as a proxy for performance and start seeing the metric as performance itself.
Wells Fargo's eight-accounts-per-customer target is the canonical modern example. The metric was designed as a proxy for deep customer relationships. Once it became a compensation target, employees optimised the metric directly — through fabrication — because incentive-caused bias had redefined "serving the customer" as "hitting the number." The original relationship between metric and value evaporated the moment the incentive was attached.
Incentive-caused bias, left unchecked, creates self-reinforcing feedback loops that amplify the original distortion. A company incentivizes quarterly revenue growth. Managers pull revenue forward through channel stuffing. Reported growth meets targets. The incentive is maintained. The pulling-forward continues at larger scale. The gap between reported and sustainable performance widens each cycle.
The 2008 mortgage crisis followed this loop: volume-based origination incentives produced more loans, which generated more securitization fees, which freed more capital for lending, which drove more volume-based compensation, which accelerated origination. Each cycle confirmed the thesis that volume was good, embedding the incentive distortion deeper into institutional memory. By the time the loop reversed, the accumulated distortion was measured in trillions. Recognising the feedback loop early — before the third or fourth cycle — is the difference between a manageable correction and a systemic collapse.
Leads-to
[Loss Aversion](/mental-models/loss-aversion)
When incentive-caused bias has shaped someone's beliefs, those beliefs become part of their cognitive identity. Correcting the distortion triggers loss aversion — the pain of admitting that your previous judgment was compromised by incentives feels like losing part of your professional self-image. The combination produces entrenchment.
A portfolio manager whose bonus depended on a position that has declined 40% faces a compound problem. Incentive-caused bias shaped the original conviction. Loss aversion makes that conviction psychologically impossible to abandon. The manager doubles down — not from analysis, but from the interaction of two biases that reinforce each other.
The pattern explains why the most destructive blowups in financial history — LTCM in 1998, Amaranth in 2006, Archegos in 2021 — all featured managers who increased position sizes as losses mounted, unable to process disconfirming evidence because doing so would have required absorbing both the financial loss and the identity loss.
The model's most sobering implication is for self-assessment. You cannot audit your own incentive-caused bias, for the same reason you cannot see your own blind spot. The bias shapes the very faculty you would use to detect it. The only reliable corrective is external: structural safeguards, mandatory second opinions from differently-incentivised parties, pre-commitment to decision criteria before the incentive has time to reshape your cognition, and the intellectual humility to accept that your judgment — however objective it feels from the inside — has been shaped by what you are paid, promoted, and praised for.
I return to Munger's FedEx story because it captures the model with zero ambiguity. Same workers. Same packages. Same facility. Different incentive. Different outcome. A meaningful portion of what we attribute to competence, character, and effort is actually the shadow cast by incentive architecture. Change the incentive and you change the behavior — not because people are mercenary, but because they're human.
The simplest diagnostic I know: for any recommendation you receive, ask what the recommender earns if you say yes versus if you say no. If the delta is large, increase your scrutiny proportionally. If you can find a second opinion from someone with a different incentive structure, weight it accordingly. And if you're the one making the recommendation, ask yourself whether you'd reach the same conclusion if your compensation were structured differently. If you're not sure — and honesty demands admitting you can't be sure — build the structural safeguard anyway. Trust the process, not your objectivity.
