Chapter 11
What do good ends still owe?
On March 28, 2024, in a Manhattan courtroom, Judge Lewis Kaplan sentenced Sam Bankman-Fried to 25 years in prison. The founder of the FTX cryptocurrency exchange had been convicted on seven counts of fraud and conspiracy for misappropriating more than $8 billion of his customers' money — funneling it through secret back doors to his hedge fund, to luxury property, to political donations, to the construction of his own influence.123 What made the case a landmark in business ethics was not the size of the theft but the moral architecture around it. Bankman-Fried was the world's most famous effective altruist — a movement urging talented people to earn enormous sums precisely in order to give them away and do the most good for the most people.24 He had, by his own philosophy, the best ends in corporate history. Kaplan was unpersuaded: the altruistic persona was, "at least in part, an act," and the crime was simple — "He knew it was wrong. He knew it was criminal. He regrets that he made a very bad bet about the likelihood of getting caught."2
Ten years earlier, a company had run the inverse experiment. On February 5, 2014, CVS — the largest U.S. pharmacy chain — announced it would stop selling cigarettes and all tobacco products in its 7,600 stores, walking away from approximately $2 billion in annual revenue, about 17 cents per share, with no offsetting requirement from any law or regulator.56 "The sale of tobacco products is inconsistent with our purpose," said CEO Larry Merlo — the company was rebranding as CVS Health, and "cigarettes have no place in an environment where health care is delivered."6 Wall Street winced; some shareholders objected. But a peer-reviewed study in the American Journal of Public Health later found that the removal measurably reduced cigarette purchasing at the population level — and that smokers who had bought cigarettes only at CVS were 38 percent more likely to stop buying them altogether.78 The company that gave up the cash cow became, over the following decade, one of America's largest healthcare enterprises.
The pairing scrambles every comfortable assumption. The man armed with an explicit moral philosophy — trained to calculate the greatest good — committed one of the largest frauds in American history. The corporation, an entity Milton Friedman taught us to expect nothing from but profit-seeking, made a costly, principled refusal that saved lives and, eventually, made strategic sense. If ethical theory doesn't produce ethical behavior, what does? And what, exactly, does a business owe the society around it — nothing but honest profits, or something more? Serious people have defended every position on that spectrum for fifty years, and this chapter will not pretend the argument is over.
The central question: how do good people and good-sounding reasoning produce unethical outcomes — and what does a company actually owe? The tools: the classical moral frameworks and why knowing them is not enough; bounded ethicality and ethical fading — how decisions lose their moral content; moral disengagement — the rationalization machinery, with Bankman-Fried as its most instructive recent operator; authority, incrementalism, and the slippery slope; how organizations are built to be ethical — leadership, climate, incentives, and voice; and the corporate social responsibility debate from Friedman to stakeholder capitalism, with the evidence attached. The two cases then return for full analysis.
Four traditions supply the vocabulary of moral evaluation. Consequentialism/utilitarianism judges acts by outcomes: the right act maximizes overall good. Deontology judges by duties and rights: some acts — lying, stealing, using people merely as means — are wrong regardless of consequences. Virtue ethics asks what a person of good character would do. Justice approaches ask whether burdens and benefits are distributed fairly and by fair process (Week 3's organizational justice is its workplace descendant).
The frameworks earn their place by making disagreements precise — most hard cases are conflicts between frameworks, not failures to think. But behavioral ethics delivers the chapter's foundational bad news: framework knowledge barely predicts behavior. James Rest's four-component model explains why — ethical action requires moral awareness (noticing an issue is ethical at all), moral judgment (reasoning to the right answer), moral motivation (prioritizing ethics over competing goods), and moral character (following through under pressure) — and the chain fails most often at the first and last links, which no philosophy class trains.9 Note also consequentialism's specific occupational hazard, visible in the hook: because it licenses trading harms against benefits, it hands a self-serving reasoner an unlimited exchange rate — any means can be laundered by a sufficiently grand imagined end. Deontology exists, in practice, as the guardrail against exactly that arithmetic.
Bounded ethicality is Week 7's bounded rationality applied to morals: systematic, predictable gaps between the ethics we endorse and the ethics we enact, operating below awareness. Its central mechanism is ethical fading — Ann Tenbrunsel and David Messick's term for the process by which the moral colors of a decision drain away as it gets reframed: "Is this honest?" becomes "a business decision," "an accounting question," "a legal risk to manage."10 Once faded, the decision is processed by cost-benefit machinery with ethics simply absent from the model — which is why fraud investigations so rarely find a moment when anyone chose evil; they find spreadsheets in which the choice never appeared as ethical at all. The related findings form a family: incentive framing (calling something "a game" or attaching sanctions can *increase* violation by signaling a business frame); motivated blindness (we fail to see unethical behavior we benefit from — auditors, boards, and Bankman-Fried's admiring investors alike); and the want/should gap (we predict our ethical selves will show up tomorrow; our wanting selves make today's decision). The defenses are framing defenses: keep moral language in the room ("who is harmed if we're wrong?"), require the ethical framing explicitly in decision documents, and treat euphemism — Week 3's Wells Fargo "simulated funding" is the canonical specimen — as the smoke that means fire. Limits: fading research is strongest in the lab, and not every reframing is corrupt; the diagnostic is whether the reframe removes affected parties from view.
Albert Bandura mapped how people commit harm while preserving their self-image as good — not by abandoning moral standards but by disengaging them through eight documented mechanisms: moral justification (harm in service of a higher purpose), euphemistic labeling, advantageous comparison ("everyone in crypto did worse"), displacement and diffusion of responsibility, distortion of consequences ("customers didn't really lose"), and dehumanizing or blaming the victims.11 The mechanisms matter because they are learnable warning signs — in others and, harder, in oneself. The FTX record reads like Bandura's checklist executed in sequence: the grand justification (billions earned to save the world), the distortion (Bankman-Fried argued at sentencing that customers had not actually lost money — the judge rejected it), the diffusion (blame assigned to lieutenants, lawyers, market chaos), and the comparison to a reckless industry.12 Add moral licensing — the documented tendency for good deeds and good identities to license subsequent bad ones — and effective altruism's role becomes legible without indicting the philosophy: a maximally virtuous self-image is not a constraint on wrongdoing; for a motivated reasoner it is an asset that makes every rationalization more credible, first to the self.411 Limits: disengagement is measured mostly by self-report and inferred post hoc; it explains how wrongdoing is sustained better than it predicts who will start.
Two situational forces complete the behavioral picture. Obedience to authority: Stanley Milgram's experiments — ordinary volunteers administering what they believed were dangerous shocks because a calm experimenter said "the experiment requires that you continue" — remain, despite decades of ethical and methodological critique, the strongest demonstration that legitimate-seeming authority can override personal conscience in a majority of normal people.12 The organizational translation needs no laboratory: instructions from bosses arrive pre-legitimized, responsibility feels displaced upward, and Week 5's psychological-safety failures show what silence costs. Incrementalism: major frauds almost never begin as major frauds; they begin as a small number smoothed, a quarter borrowed from, a customer account "temporarily" tapped — each step tiny relative to the last (Week 7's anchoring, applied to conscience), each success raising the sunk moral cost of stopping (Week 7's escalation, applied to wrongdoing). The pair yields this chapter's most practical self-defense: decide your refusals in advance and in categories ("I do not sign numbers I haven't verified"), because the slope is engineered precisely so that no single step ever feels like the decision — and the person who waits to feel the bright line will be standing well past it.
If individual virtue is unreliable under organizational pressure, ethics must be engineered into the organization — and the evidence identifies the load-bearing parts. Ethical leadership: Brown, Treviño, and Harrison define and measure it as the demonstration of normatively appropriate conduct plus its active promotion — leaders who model, communicate, reward, and discipline around ethics; it predicts follower ethical behavior, willingness to report problems, and perceptions of fairness, and it works substantially through social learning: people do what the powerful visibly do, not what they say.13 Ethical climate and infrastructure: codes matter only as much as their enforcement; the components with teeth are consequence systems that punish violations *by high performers* (the Uber test, Week 8), incentive designs audited for what they actually pay for (the Wells Fargo lesson, Week 3), and protected voice channels that route around management (the Boeing lesson, Week 5 — and note that FTX fell when insiders finally talked). The integration insight: every prior week of this course has been silently teaching ethics infrastructure — goals without integrity counterweights, cultures with espoused–enacted gaps, silenced dissent, unauditable authority — because ethical collapse is almost always an organizational design failure wearing a personal morality costume. Limits: climate research leans on surveys vulnerable to social desirability, and compliance structures can themselves become ethical fading machines — converting "is it right?" into "did legal approve?"
The corporate-level debate has two poles and a large evidence file. Friedman's position (1970): the social responsibility of business is to increase its profits within the rules of the game — open competition without deception or fraud; executives spending shareholders' money on social causes are taxing owners without authority, and social problems belong to governments.14 The stakeholder position (Freeman): firms are webs of relationships with employees, customers, suppliers, and communities whose interests are the *purpose* of the enterprise, not constraints on it; Archie Carroll's pyramid operationalizes the middle ground — economic and legal responsibilities as the base, ethical responsibilities as obligatory, philanthropic as desired.1516 The evidence: Margolis and Walsh's review of 30 years of studies found the relationship between corporate social performance and financial performance to be, on balance, mildly positive and definitely not negative — doing good does not systematically cost shareholders — while warning that the obsessive search for a business case itself concedes Friedman's premise that ethics must pay to be permitted.17 Porter and Kramer's shared value argues the poles converge in practice: the durable social contributions are the ones built into strategy — which is precisely the structure of the CVS decision, an ethical refusal that was simultaneously a strategic repositioning.18 Honest limits: "stakeholder" rhetoric is cheap and often decorative (pledges without governance changes), ESG measurement remains contested and politicized, and Friedman's constraint — *without deception or fraud* — would by itself have prohibited everything Bankman-Fried did; the poles are closer than the shouting suggests.
The cases were chosen because they invert the intuitive mapping between moral theory and moral behavior. Case A features maximal ethical *theorizing* with catastrophic conduct; Case B features minimal theorizing — a purpose statement and a refusal — with measurable social benefit. Together they test this chapter's central claim: ethical outcomes are produced by awareness, framing, and organizational design far more than by the sophistication of anyone's moral philosophy.
The facts: FTX rose to a $32 billion valuation as the respectable face of crypto while, per the jury's verdict, customer deposits flowed through concealed channels to Alameda Research's trading, to real estate, to over $100 million in political contributions, and to the cultivation of Bankman-Fried's public standing; the November 2022 collapse left an $8 billion hole, hundreds of thousands of victims, and — after a month-long trial — conviction on all seven counts, a 25-year sentence, and $11 billion in forfeiture.123 As a theory test, the case operationalizes Sections 2.1–2.4 with unusual completeness. The consequentialist hazard: an "earning to give" identity supplied a limitless justification budget — every risk to customers weighed against imagined billions for humanity.4 Ethical fading: the record describes transfers processed as liquidity management and accounting questions, never as the taking of other people's money.110 Moral disengagement at sentencing: distortion of consequences (customers "did not actually lose"), rejected by the court; diffusion; comparison — with Kaplan's finding that the altruism was performance and the core a gamble on not getting caught.211 Organizational design: essentially none — no independent board, no CFO controls at the relevant entities, an inner circle bound by loyalty, no channel through which the employees who knew could safely say so until collapse made it moot. Honest cautions: the verdict establishes the fraud, not the inner narrative — whether Bankman-Fried believed his own justifications is unknowable and analytically secondary; and effective altruism is not on trial here — the mechanism (virtuous identity as rationalization asset) attaches to any noble cause, religious, patriotic, or scientific, which is exactly why it matters to you.
The facts: on February 5, 2014, CVS announced the tobacco exit — all products, all 7,600+ stores, effective that fall — disclosing to the SEC an expected $2 billion annual revenue loss (~17 cents per share) and, remarkably, maintaining its earnings guidance while promising to offset the hit; the company simultaneously renamed itself CVS Health and launched smoking-cessation programming.56 The measured consequences: the AJPH study comparing household purchasing before and after found significant population-level declines in cigarette buying, concentrated exactly where theory predicts — CVS-exclusive cigarette purchasers were 38 percent more likely to stop buying cigarettes entirely, and heavy purchasers more than twice as likely — evidence that a single retailer's refusal changed behavior rather than merely relocating it.78 As a theory test: the decision is deontic in form ("inconsistent with our purpose" — a refusal, not a calculation6) and shared-value in structure — the ethical stand *was* the strategic repositioning into healthcare, Porter and Kramer's convergence executed a decade before most firms discussed it.18 It also illustrates Carroll's pyramid navigated in the right order: the economic base protected (guidance maintained), the ethical obligation honored at real cost, the philanthropic layer ($50 million youth-prevention initiative) added on top.516 Honest cautions, each substantive: the flagship purchasing study was company-funded and partly company-authored, disclosed as such in the journal8; the $2 billion was under 2 percent of revenue — a costly signal, not an existential sacrifice; the move served an acquisition-era strategy whose later chapters (pharmacy benefit management, insurance) have drawn their own serious ethical scrutiny, so the case certifies a decision, not a company; and Friedman himself could bless it — a firm freely repositioning its brand breaks no rule in his game.14
| FTX / Bankman-Fried | CVS tobacco exit | |
| Moral reasoning visible | Elaborate: utilitarian 'earning to give' philosophy | Minimal: a purpose statement and a refusal |
| Mechanism at work | Ends-justification + fading + disengagement + zero controls | Identity-based bright line, priced in public, executed completely |
| Cost accepted | None accepted — costs hidden and transferred to customers | $2B/yr revenue (~17¢ EPS), disclosed to the SEC in advance |
| Measured outcome | $8B customer losses; 7 convictions; 25 years; $11B forfeiture | Population-level cigarette purchasing fell; CVS-only buyers 38% likelier to quit buying |
| Evidence caveat | Verdict proves acts, not inner beliefs | Key study company-funded; decision also served strategy |
Three conclusions, held carefully. First, the pairing confirms the behavioral-ethics thesis: what separated the cases was not the quality of anyone's moral philosophy but the structure of the decisions — FTX's harms were faded, diffused, and unauditable; CVS's refusal was explicit, priced, public, and irreversible, leaving no room for the machinery of Sections 2.2–2.3 to operate.51011 Second, the cases dissolve the false choice between Friedman and the stakeholder view better than any lecture: Friedman's own rule — profit *without deception or fraud* — condemns FTX absolutely, while CVS shows stakeholder benefit and shareholder strategy fused in one act; the live debate is narrower and more honest than its slogans.1418 Third, symmetry in skepticism, as always: do not canonize CVS (self-funded evidence, strategic convenience, a corporate record with other chapters) and do not demonize moral ambition because Bankman-Fried wore it — the lesson of Case A is not that idealists are frauds but that a virtuous identity is the most valuable costume a rationalization can rent, which is precisely why the people most confident of their goodness need the most external checking. Including, someday, you.
Return to the central question: how do good people and good reasoning produce bad outcomes, and what does a company owe? Four conclusions hold.
Keep the ethics visible. Most wrongdoing passes through a frame in which it never appears as ethics at all. Name the affected parties out loud, ban the euphemisms, and treat 'it's just a business decision' as the moment to slow down, not speed up.10
Audit your justifications, especially the noble ones. Higher purposes, team loyalty, and 'everyone does it' are the standard solvents of conscience; the grander the end, the harder the means deserve to be checked — by someone with the power to say no.211
Draw lines in advance; expect the slope. Authority will arrive pre-legitimized and the first step will be small. Categorical pre-commitments — the things you do not sign, say, or ship regardless — are worth more than any amount of in-the-moment judgment, because the moment is rigged.12
Build the organization that your virtue will need. Ethical leadership that rewards and punishes visibly, incentives audited for what they actually pay for, voice channels that route around power, and consequence systems that reach high performers — the whole course has been this list.13 And on what the corporation owes: at minimum Friedman's floor, honestly kept — no deception, no fraud — and at best the CVS pattern, where what the company refuses to profit from is itself the strategy.71418 The evidence says that path does not require choosing against the shareholders.17 It requires choosing on purpose.
Notes appear as superscript numbers in the text and correspond to the numbered sources above. DOIs are provided where available; classic books are cited to their original publishers.