You have been elected benevolent dictator. Your first decision: expand programs that improve the lives of people with low incomes today, or adopt policies expected to produce a slightly faster-growing economy. Which do you choose?

It is tempting to say, “both.” Sometimes that is possible, and economists are delighted when it is. But economics begins with scarcity: resources, time, attention, land, labor, and administrative capacity cannot serve every valuable purpose at once. A student chooses between friends and exam prep. A family chooses between spending now and saving. A society chooses between helping visible people now and protecting the productive capacity that will shape lives later.

A trade-off appears when gaining more of one valued thing requires accepting less of another. When a government spends on the military rather than consumer goods, that is the classic “guns versus butter” trade-off. But there are an infinite number of them, because every choice is one: studying or seeing friends; whether to go shopping, and what to buy once you are there; which college to attend, which major to pursue, which classes to take. Each is a fork with a road not taken.

This has particular bite in public policy. As Thomas Sowell put it, “there are no solutions, only trade-offs” — every solution carries a cost. Bryan Caplan pushes the same point somewhere uncomfortable: the socially optimal rate of crime is greater than zero. Why? Because the police state required to stamp out the last of it is not worth what it would cost the rest of us.

The trade-off is not always between an activity and money. A local service station choosing between more diagnostic computers and fewer mechanics, or fewer computers and more mechanics, is trading capital against labor — more technology and less human attention, or the reverse. Every recipe for producing something is a bundle of such trades.

The opportunity cost of a choice is the value of the best alternative given up. It is not necessarily money, regret, or the sum of every rejected option. It is the single best thing you could have done instead. Use a free airline voucher for a weekend trip rather than selling it for $400, and the ticket's cash price is zero — but its opportunity cost includes the $400 you surrendered. “Free” describes the cash you paid, not the value you gave up.

If economists assume something is right with people, what does that look like in practice?

In advanced economics classes, you will learn about choice models with high levels of specificity. For now, it's enough to understand the rationality assumption in terms of both dignity and modeling: The economist's first response to a puzzling choice is not “what is wrong with this person?” It is “what makes this choice sensible from where the person stands?” That posture treats people as agents rather than objects. 

Then, in terms of models: Economists usually begin with a rationality assumption defined as follows: 

1. Rational people use available and relevant information to pursue their objectives.

2. Rational people maximize their well-being.

3. Rational people respond to incentives.

4. Rational people think at the margin.

5. Rational people are a good model for real people, even if real people aren't always perfectly rational.

This set of ideas are working assumptions, not a claim that people are flawless calculators. Rational people can lack information, misjudge probabilities, act impulsively, care about others, or pursue goals you dislike. Rationality does not certify that an objective is admirable. It gives behavior enough structure to explain and predict — the point is not to slander people or to bestow sainthood on them, but to understand what they do.

Economists trust that people are doing the best they can with what they have — but we trust their actions far more than their words. Talk, after all, is cheap.

So what people tell you they value can be sincere and still predict their behavior poorly. Ask Americans in a survey whether they care about the environment, or whether they would consider going electric, and large majorities say yes. Then look at what they actually drive off the lot: for nearly fifty years running, the best-selling vehicle in the country has been a full-size pickup — the Ford F-Series — and trucks and SUVs dominate the sales charts while fully electric cars remain a modest slice. What people say and what people buy, with their own money and at real prices, point in different directions. This is the principle of revealed preference: the best evidence on what people value is what they do, not what they say. 

People respond to incentives

Gordon Tullock once proposed a grotesque thought experiment: mount a steel spike on every steering wheel, aimed straight at the driver's chest, on the theory that people drive exactly as carefully as the stakes make them. Change what a choice costs, and you change what people do.  He was not recommending anyone build the thing. He was making one idea impossible to ignore: people respond to incentives. Make a crash more personally costly, and drivers will probably slow down, leave more room, and pay closer attention.

Imagine your car had a steel spike in the center of the wheel, pointed at your chest. Would you drive more carefully?

While the example is vivid, there may well be some truth in it. It might actually be that such a scheme might save lives. As cars become safer, people drive more. As people drive more, they have more accidents. In the last 50 years, the average deaths-per-accident have dropped significantly, but the annual traffic deaths number has stayed more-or-less flat.

Why? On one theory, because people respond to incentives. People drive more recklessly today precisely because car accidents are so much safer. 

Note the following side-by-side graphs, taken from the Congressional Research Service. The first shows the declining rate of deaths-per-accident, the second shows the stubborn rate of deaths

Traffic fatality rate per crash:

Traffic deaths, annual:

The Tullock spike works the same way: it changes the driver's personal consequence of a collision. A fee on disposable bags changes the cost of taking another. A scholarship changes the benefit of earning a grade. When the terms of choice change, some choices change. And because an incentive is just a lever, one principle explains both the good and the bad — the rule that quietly steers people right, and the one that steers them somewhere absurd.

Consider a different example: DNA databases. Police who have access to DNA records can catch bad guys at higher rates. Expanding one makes identification and conviction more likely, which raises the expected cost of committing certain crimes. Incentive reasoning predicts less crime, other things equal — without requiring any offender to become a statistician.

A road map is a masterpiece of leaving things out. No trees, no weather, no traffic, no people — almost the entire world, deleted. And that deletion is the whole point: a map that showed everything would be the size of the city and just as impossible to read. An economic model does the same to a question. It throws away most of reality on purpose, so that one relationship stands out sharply enough to reason about.

Models are a main tool in the economist's kit. A model works like a map: not a true, point-for-point copy of the territory — a map that marked every blade of grass would be useless — but a deliberate, meaningful simplification that helps you get somewhere. As the statistician George Box put it, all models are wrong, but some are useful. The rationality assumption is wrong about plenty of people some of the time; it earns its keep because it is right enough, often enough, to predict how an entire market will move.

The Tullock spike is a model. It is cartoonishly unrealistic — no real car has a dagger in the wheel — and that is fine, because its job is not to be a car. Its job is to isolate one true thing: raise the personal cost of a crash and drivers get careful. So the amateur's complaint, “that's not realistic,” misses every time — every model is unrealistic. The professional asks the sharper question: does the thing you left out change the answer? 

Building a model means making its moves out loud. You fix the assumptions — the simplified world it runs in (“people pursue purposes under constraints”). You name the variables — the measurable things that can move: price, hours worked, tax rate, the odds of arrest. Then you commit to a hypothesis, a testable claim about how those variables connect, usually a causal one: raise the penalty for an action, get less of it, other things equal. 

Notice that the purpose of a model is to test it: a hypothesis has to be able to be wrong. “This policy changes society in mysterious ways” risks nothing and predicts nothing. “A five-cent bag fee cuts disposable-bag use” sticks its neck out: watch the checkout lanes, and reality can prove it false. Question, assumptions, variables, hypothesis, prediction, evidence — then revise and run it again. That loop is the engine of the whole discipline. Economists are fundamentally empiricists: we let reality weigh in on the questions we ask.

Ceteris paribus: hold all things equal

Ceteris paribus — “other things equal” — is how a model pins the world down long enough to watch one thing move. “Ceteris paribus, a higher price lowers the quantity people buy” doesn't claim price is all that matters; it holds tastes, income, and expectations still and watches price alone. You need the move because the world refuses to cooperate: ice-cream sales and sunburns climb together every summer, yet ice cream burns no one — the sun moves both. Without some way to compare like with like, correlation cheerfully impersonates cause.

Empiricism: reality gets the final vote

Empiricism is the  belief that we should follow the evidence. The rule claims that we should answer to evidence, not to eloquence. A beautiful theory the data contradict is simply wrong, however sincerely it is argued. Perfect evidence almost never exists; the goal is the explanation that survives the hardest honest attempts to kill it — run against experiments, natural experiments, historical comparisons, administrative records, prices, and the choices people actually make.

This is also what a model is for. Point one at tariffs, taxes, crime, pollution, or which occupations will be hiring next decade, and it does the same job every time: force the assumptions into the open, define the variables, and say in advance what result would prove it wrong. The demand-and-supply model we build in later chapters is the workhorse — it explains how buyers' and sellers' choices settle on a price. Disciplined simplification is the strength; a simplification whose deletions are hidden is the danger.

The Circular Flow Diagram: Our First Economic Model

Now it is time for our first model. 

The simplest map of an entire economy has just two kinds of players — households and firms — meeting in two kinds of markets.

In the product market, firms sell goods and services and households buy them. In the factor market, the roles flip: households are the sellers, because households own the resources — their labor, their land, their savings — and firms buy those resources to produce.

Follow the money and it runs in a loop. Household spending in the product market becomes firms' revenue; firms pay that revenue back out as wages, rent, interest, and profit in the factor market; that becomes household income, which gets spent again. One person's spending is the next person's income, all the way around — with goods and resources flowing the opposite direction.

Before you can think like an economist, you must learn to see like an economist.

In 1850, the French economist Frédéric Bastiat told a small story that still ranks among the most useful things an economist can carry. A shopkeeper's son breaks a window. A crowd gathers and consoles him: the accident gives work to the glazier (the glass worker), who will spend the money elsewhere. Perhaps destruction is good for business.

Bastiat's reply is wisdom that the discipline of economics takes very seriously, and it answers our question about WWII as well. The crowd counts the seen — the glazier's new work. It misses the unseen — the suit the shopkeeper would have bought with the same money. Without the broken window he would have owned a window and a suit; afterward he has only a restored window. The glazier's visible gain is matched by the tailor's invisible loss.

The broken-window fallacy is a mistake of not seeing all of the involved variables. In this case, we assume that "repairing windows will help the economy" when it won't: repair doesn't improve well-being. Rebuilding after a hurricane employs people, but the same labor and materials could otherwise have built additional homes, machines, or medicine. This matters most in politics, where visible beneficiaries gather for photographs while opportunity costs scatter across taxpayers, consumers, and future people. A good economist becomes stubborn on their behalf.

There are legion applications of this understanding. Do you want inflation to go up, or to go down? It depends: you likely want your money to have more purchasing power when you purchase, but less purchasing power when you pay others for goods. As a simple rule-of-thumb, then, inflation benefits those with large debts just as it harms lenders; deflation would invert that logic: benefitting lenders and harming debtors. Your thinking on any economic question can be improved simply by insisting on seeing the unseen.

Positive vs. Normative

We're almost ready to apply these tools to one of the most fundamental arguments in economics: income redistribution. Before we do, however, there is one more tool we need to learn to wield carefully before a brief synthesis.

Pull almost any economic question apart — “we should expand food aid,” “that will just make people work less” — and you find two different arguments wearing one coat. One is about what is true; the other is about what is right.

A positive statement describes what is, was, or will be, and can be checked against data: “SNAP raises recipients' food spending”; “a steeper phaseout cuts hours worked.” A normative statement judges what ought to be: “no child should go hungry”; “this help is worth its cost.” “Positive” does not mean favorable and “normative” does not mean irrational — a positive claim can describe a horror and still be false, and a normative claim can be carefully reasoned. The distinction just tells you which kind of argument you are in, and so what could possibly end it: evidence can settle a positive dispute, but only values can settle a normative one. That is why a complete policy argument always needs one of each — a factual premise (the program does this) and a value premise (this is worth it).

Note that economists do both: when the economist is acting as a policy advisor, her role is to talk values, application, and context. When the economist is acting as a scientist, her goal is to talk through prediction, estimation, and causal relationships. The key idea is this: you will be better at both jobs if you take them one at a time. Far too often, an ineffective policy that feels good gets too much attention and resources; and just as often, an effective-but-counterintuitive policy gets nil. While the Tullock spike is an extreme example, it's a manifestation of how economists think precisely because it's coarse and tawdry, and utterly effective.

Thinking Like An Economist

Among economists, there is a strong presumption in favor of "enlarging the pie" rather than distributing it. What does that mean, and why does it matter?

In economics, "efficiency" has a multitude of meanings; for now, a simple version would be this: an efficient outcome is one in which the size of the pie is maximized. You will note that there is nothing to suggest that a large pie is going to be split evenly among all at the table, however. Why, then, do economists focus so much on efficiency instead of equity?

One reason is practical: distribution can be difficult to model. It is relatively straightforward to ask whether a policy increases total output, lowers costs, or makes a market more efficient. It is much harder to model exactly who gains, who loses, by how much, and how much those gains and losses matter to each person. So economists often simplify the problem. First ask how large the pie will be; worry later about how it is divided. This is also a careful strategy on the part of economists: not to suggest that distribution doesn't matter, but that by separating the parts of the argument, they can be more effective at each part. 

A second reason is philosophical: data cannot tell us what a fair distribution looks like. Economics can estimate that one policy gives Person A $100 while costing Person B $80. But economics cannot tell us whether that trade is morally acceptable. Should we care more about the poor than the rich? How much inequality is too much? How much should property rights constrain redistribution? Those are questions about values. Evidence can inform them, but it cannot settle them. Economists therefore tend to be much more comfortable talking about efficiency than fairness.

There are other reasons we will finish the chapter with, too. For now, however, the purpose of the rest of this chapter is to give you a strong argument for both sides. You may disagree with any such views, of course. Our goal is simply to leave you smarter for having thought through it all.

A Crash Course in the History — and Effectiveness of — Redistribution

Redistribution can happen in a variety of ways, both private and government-mandated. People redistribute privately when they give money, time, or care — a donor may give to charity because increasing the well-being of poorer people is part of what she values. Government redistributes through taxes and spending. A progressive tax takes a larger percentage of income from higher-income groups than from lower-income groups. SNAP provides food benefits to low-income households; Medicaid provides health coverage to eligible groups. These are in-kind redistribution: assistance directed toward particular goods rather than unrestricted cash. The central disagreement is rarely whether any redistribution should exist — families, charities, insurers, and every modern government already do it. It is how much, through which institutions, for which purposes, and with what consequences.

Has contemporary, American redistribution been a success or a failure?

That depends in great measure on how you define "success," but there are still some clean data worth looking at. The first step, however, is to separate the normative and positive.

Some redistribution show signs of genuine success. Social Security helped drive old-age poverty from roughly 35 percent in 1959 to about 10 percent by the 1990s, and researchers attribute most of that fall to the growth in benefits — one of the largest measured poverty reductions in American history. The Earned Income Tax Credit, which pays only people who work, pulled large numbers of single mothers into the workforce and lifted millions of children out of poverty, with little of the work-discouraging effect a steep phaseout usually brings. And childhood access to food stamps, traced through the program's county-by-county rollout in the 1960s and '70s, shows up decades later as better adult health and higher self-sufficiency — the aid paid a return long after it was spent.

But “worked” is not one thing, and the honesty cuts both ways. As many successes as there are, there are failures in the mix as well. 

The Welfare Cliff (100%+ Marginal Tax Rates) When benefits vanish abruptly as income rises, working harder can leave a family worse off. If a worker accepts a $1-per-hour raise and immediately loses a $600-per-month childcare voucher or Medicaid eligibility, they face an effective tax rate over 100%. This "cliff" penalizes career advancement and trap families in poverty, making self-sufficiency financially irrational.

Paternalistic Inefficiencies (Deadweight Loss) In-kind aid assumes the government knows what a family needs better than they do, which often backfires. When programs like SNAP cover food but ban basic essentials like diapers, laundry detergent, or toilet paper, desperate families sell $100 of food benefits on informal markets for $50 cash. This paternalism generates massive deadweight loss: taxpayers spend $100, but the recipient gets only $50 of real value. 

Block-Grant Diversion (Bureaucratic Siphoning) Giving states broad flexibility over federal welfare money often results in funds never reaching the poor. Under programs like TANF, states have used block grants to plug state budget holes, subsidize college scholarships for middle-class students, or fund pet political projects. In extreme cases, less than 20% of federal welfare dollars reach impoverished families as direct assistance.

So the positive verdict is real but conditional: some redistribution clearly works, some does not, and design is a major difference.

Proposed Normative Approaches

Now, to the normative side: what should we actually do? That depends heavily on your values. Consider just three cases:

A Utilitarian aims to maximize total well-being. Because an extra dollar usually matters more to someone who has very little than to someone who has a great deal, transferring it from rich to poor could increase overall welfare. Note that most economic analyses rely on a foundation of utilitarianism.  Taking $1,000 from a wealthy individual causes minimal loss in well-being, but giving that same $1,000 to a struggling family dramatically expands their health, security, and utility. Redistribution is justified up to the point where the total happiness gained by the poor equals the total economic loss caused by shrinking the overall pie.

A Rawlsian follows philosopher John Rawls, who emphasized the need to take care of the poorest. Rawls asked us to imagine a "veil of ignorance," wherein we knew we would be born to a society, but didn't know what station we would occupy. A truly fair society is one that you would design assuming you could be born into it as anyone—rich, poor, disabled, or brilliant—without knowing your position in advance. Self-interest forces you to choose a system that maximizes the well-being of the least advantaged, since that disadvantaged person could easily be you. Under this difference principle, economic inequalities are only justified if they serve to make the worst-off person better off than they would be under absolute equality—meaning robust redistribution isn’t a matter of pity or charity, but the core requirement of justice. From this perspective, redistribution may be justified if it improves the lives of those at the bottom.

A Libertarian emphasizes individual liberty and property rights. Robert Nozick (a libertarian economist) taught that justice depends entirely on how wealth is acquired, not how it is distributed across society.  Since individuals own their own bodies, talent, and labor, taxing their earnings to help others is a form of forced labor—coercively taking the proceeds of your time to serve goals you did not choose. If property is acquired through honest labor or voluntary trade, the resulting distribution is inherently just, regardless of inequality. Redistribution violates fundamental self-ownership, turning state welfare into legalized theft. For libertarian, even if someone else would benefit more from the dollar, that does not necessarily justify the government taking it from its owner. 

Who is right? 

All of them, in a sense. It is good to learn to adopt the various frameworks as occasional lenses that you can put on and take off. The libertarian framework might feel extreme, for example, but the fierce defense of property rights is part of the liberal consensus mentioned in the introduction to this book that has brought so much wealth. The Rawlsian approach may similarly feel absolutist, until you consider just how many people have so little in any given society. 

On Deserts: Who deserves what? Where economics goes quiet

There is one question in this whole area that economics deliberately refuses to answer — and it is the one people most want answered: what share does each person deserve? An economist can measure how income is actually distributed, and can predict how a tax or a benefit would shift it; both are positive questions with checkable answers. But whether a surgeon should out-earn a schoolteacher tenfold, whether the top one percent holds “too much,” whether a just society equalizes incomes or instead rewards effort, risk, and talent — none of that is the unique purview of the economist. Those are questions of desert, philosophy, and fairness; normative to the core. Economists do not answer such questions. This is not a dodge; it is the discipline keeping its hands clean. An economist can say an enormous amount about what a distribution does while admitting, plainly, that whether it is fair is a moral judgment that is not the special province of regressions, statistics, or economic models. 

Winners, Losers, and the Win-Win

Most economic policies create winners and losers. Economists understand this perfectly well. But much of economics deliberately puts the question of distribution to one side.

We've taken an initial stab at why. Now, we finish the job.

One reason merits more discussion. 

Economists are sometimes caricatured as being cold utilitarians who only want to focus on economic growth: GDP per capita above all. Rather than emphasize the welfare of the poorest, they are vulcans who do only what their equations tell them. There is truth in these critiques: economists sometimes become too entranced with their own models, ignore the realities of the economic reality facing everyday people, and obsess over efficiency. 

But the caricature falls flat for a variety of reasons. One is that there are plenty of economists who really do obsess with inequality. Similarly, no economist I have ever met is obsessed with economic growth for its own sake. The story goes that the economist would choose per capita GDP over any tradeoff: literacy, infant mortality, or self-surveys of satisfaction. My experience has been nearly the opposite: in every single case, economists favor economic growth because it leads to all the other things that we should really care about. 

In short, economists focus on making the pie bigger  because has historically solved a remarkable number of distributional problems.

Compare countries—or compare the same country across time—and income per person tends to move together with many of the things we ultimately care about. As societies become richer, extreme poverty falls. Literacy rises. Infant and maternal mortality fall. Life expectancy rises. Children stay in school longer. People gain access to cleaner water, better housing, sanitation, medicine, and technology.