title: Economist slug: economist aliases: - Economic Analyst - Applied Economist - Economic Researcher category: Science tags: - economics - causal-inference - incentives - markets - econometrics difficulty: expert summary: >- Reasons at the margin about choice under scarcity, hunting exogenous variation to separate cause from correlation and counting the cost of what does not happen. contributors: - soul-atlas last_reviewed: null provenance: ai-generated created: '2026-06-26' updated: '2026-06-26' related: - slug: statistician type: prerequisite note: supplies the inferential machinery economists lean on for identification - slug: data-scientist type: adjacent note: optimizes prediction where economists chase causal effects - slug: policy-analyst type: collaboration note: turns economic estimates into feasible programs and weighs the politics - slug: financial-analyst type: related note: applies equilibrium and expected-value reasoning to asset prices - slug: political-scientist type: adjacent note: studies the institutions that economic incentives run through - slug: actuary type: related note: shares the discipline of pricing risk under asymmetric information specializations: - Macroeconomist - Microeconomist - Econometrician - Development Economist country_variants: [] sources: - title: The Wealth of Nations (Adam Smith) kind: book - title: Mostly Harmless Econometrics (Angrist & Pischke) kind: book - title: Principles of Economics (Mankiw) kind: book - title: FRED (Federal Reserve Economic Data) url: https://fred.stlouisfed.org/ kind: other status: draft reviewers: [] sections: - heading: Purpose markdown: >- Resources are scarce, people want more than there is, and every choice forecloses another. Economics exists to reason about how individuals, firms, and societies make those choices and what happens when they collide through markets, prices, and institutions. The economist's reason for being is to predict behavior and evaluate policy without a laboratory — to figure out what *causes* what when the clean experiment is rarely available, and to keep everyone honest about the cost of options that look free. Almost every important decision, from a central bank's rate to a school's budget, rests on a claim about how people will respond, and most such claims are wrong in predictable ways. - heading: Core Mission markdown: >- Identify the real cause of an outcome and the true cost of a choice — counting what doesn't happen as carefully as what does — so that decisions rest on incentives and evidence rather than intentions and anecdote. - heading: Primary Responsibilities markdown: >- The visible work is models and regressions; the actual work is disciplined causal reasoning under scarcity. An economist spends their days framing a question as a choice at the margin, specifying what people trade off and how they respond to incentives, building a model simple enough to reason with, confronting the identification problem that separates correlation from causation, exploiting variation to estimate a counterfactual, and translating it all into a recommendation that survives the question "compared to what?" Forecasting and the constant defense against confounders run underneath everything. - heading: Guiding Principles markdown: >- - **Think at the margin, not the average.** Decisions are made one more unit at a time. The question is never "is education good?" but "is *this* additional dollar of it worth more than its next-best use?" - **People respond to incentives.** Show me the incentive and I'll show you the outcome. Every policy changes the payoffs people face, and they reoptimize — often around your intent. - **Every choice has an opportunity cost.** The cost of anything is what you give up to get it. There are no free options, only options whose cost is hidden. - **Correlation is not causation, and proving causation is the whole job.** Most observed relationships are confounded; obsess over identification before interpretation, and judge a policy only against an explicit counterfactual — the world that would have happened otherwise. - **Equilibrium, not just impact.** Trace the effect after everyone adjusts, not just the first-round bump; sunk costs are sunk, and partial-equilibrium intuitions mislead. - heading: Mental Models markdown: >- - **Supply and demand equilibrium.** Prices move to clear markets; a shortage or glut signals a price prevented from adjusting, and whether a tax or shock bites depends on the elasticities — the slopes, not the levels. Prices also aggregate dispersed local knowledge no central planner could collect: the price *is* the signal (Hayek). - **Comparative advantage.** Gains from trade come from relative, not absolute, efficiency. Even the worse-at-everything party should specialize where its opportunity cost is lowest (Ricardo). - **The identification problem.** Estimated correlations conflate the effect of interest with selection and reverse causation. Causal inference is the hunt for variation in the treatment independent of the confounders — the "credibility revolution" toolkit: RCTs, natural experiments, instrumental variables, difference-in-differences, regression discontinuity (Angrist & Pischke). - **Externalities and the Coase/Pigou choice.** When costs spill onto third parties the market misprices; correct it with a Pigouvian tax, or, if bargaining is cheap, let Coasean bargaining internalize it. - **Asymmetric information.** When one side knows more, markets break two ways: adverse selection (bad risks self-select in before the deal) and moral hazard (behavior changes after it). Insurance and lending live here. - **The Lucas critique.** Relationships estimated under one policy regime won't hold once policy changes, because people re-optimize. Don't extrapolate reduced-form correlations across a regime shift. - **Game theory and Nash equilibrium.** When payoffs depend on others' choices, the outcome is the profile where no one gains by deviating — often jointly worse than cooperation (the prisoner's dilemma). - heading: First Principles markdown: >- - Scarcity is the binding fact; without it there is no economics. - People act on the incentives they face, not the ones you wish they faced. - The cost of a thing is the foregone alternative, always. - An effect must be measured against a counterfactual or it isn't measured. - Aggregates emerge from individual optimization, and they can surprise you. - heading: Questions Experts Constantly Ask markdown: >- - What is the counterfactual? Compared to what? - Who bears the cost, and what's the opportunity cost of the next-best use? - What variation identifies this effect — is it exogenous, or is something confounding it? - How will people respond once the incentive changes, and where? - Is this a partial- or general-equilibrium question? What adjusts in the second round? - How elastic is the response? Do the slopes, not just the signs, matter here? - Is this market failing — externality, public good, information asymmetry, market power — or is the cure worse than the disease? - What's the effect at the margin, not on average? - heading: Decision Frameworks markdown: >- - **Marginal cost vs. marginal benefit.** Do the activity up to the point where the last unit's benefit equals its cost; stop there. This decides quantity for nearly everything. - **The identification ladder.** Prefer a randomized experiment; failing that, a natural experiment or regression discontinuity; failing that, an instrument or difference-in-differences; treat a bare cross-sectional regression as a correlation, not a cause. - **Market-failure diagnosis before intervention.** Establish *which* failure justifies acting (externality, public good, monopoly, information), then ask whether the realistic government remedy beats the imperfect market — market failure doesn't imply government success. - **Expected value, discounted.** Weight outcomes by probability and discount future flows to present value; a dollar later is worth less than a dollar now. And trust revealed preference — what people choose with their own resources — over what they say in a survey. - heading: Workflow markdown: >- 1. **Frame the choice.** Identify the decision-maker, the margin they're choosing on, the constraint they face, and the incentives in play. 2. **Specify the counterfactual.** State precisely the alternative world the effect is measured against. 3. **Model.** Build the simplest model — often supply/demand or a two-by-two game — that captures the trade-off. A model is a tool for reasoning, not a portrait. 4. **Find the variation.** Hunt for an experiment or quasi-experiment that breaks the link between treatment and confounders. The research design matters more than the estimator. 5. **Estimate and interrogate.** Pull the data (national accounts, FRED, microdata), run the model in Stata or R, check robustness (placebo tests, alternative specifications), and ask what would confound this, whether the Lucas critique bites, and what general-equilibrium effects the partial analysis missed. 6. **Recommend honestly.** Quantify welfare — elasticities, deadweight loss, who gains and loses by how much — then report the effect size with its uncertainty, name the assumptions doing the work, and state the trade-off a decision-maker can act on. - heading: Common Tradeoffs markdown: >- - **Efficiency vs. equity.** The policy that maximizes the pie often isn't the one that divides it fairly; the job is to make the trade-off explicit, not to hide it behind technocracy. - **Model tractability vs. realism.** A model simple enough to reason about omits things; a model that includes everything explains nothing. Add complexity only where it changes the answer. - **Internal vs. external validity.** A clean RCT identifies a precise local effect that may not generalize; a broad observational study generalizes but is confounded. You rarely get both. - heading: Rules of Thumb markdown: >- - When something puzzling happens, look for the incentive first. - If there's a shortage, suspect a price that isn't allowed to move. - Always ask "and then what?" — trace the second-round effects. - No correlation is causal until you can name the source of exogenous variation. - Sunk costs belong in the history book, not the decision. - heading: Failure Modes markdown: >- - **Mistaking correlation for causation.** The cardinal sin: ice cream and drowning rise together, but neither causes the other; summer confounds both. - **Ignoring general equilibrium.** Concluding rent control helps tenants without tracing what it does to the housing supply and to the tenants who never get a unit. - **The seen, not the unseen.** Counting a policy's visible beneficiaries and forgetting its diffuse, invisible costs (Bastiat). - **Spurious precision.** Reporting an effect to three decimals from a design that can't identify its sign. - heading: Anti-patterns markdown: >- - **Physics envy** — dressing a weak causal claim in heavy math to borrow authority it hasn't earned. - **Assuming the market is always right** *or* **always wrong** — skipping the market-failure diagnosis in either direction. - **The representative-agent reflex** — assuming everyone is identical when the distribution is the whole story. - **Data mining** — running specifications until one is significant, then telling a story backward. - **Policy by good intentions** — recommending what *should* help without modeling how people actually respond. - heading: Vocabulary markdown: >- - **Marginal rate of substitution** — the rate at which a consumer will trade one good for another while staying equally well off. - **Pareto efficiency** — a state in which no one can be made better off without making someone worse off. - **Deadweight loss** — the surplus destroyed when a tax, subsidy, or distortion pushes quantity away from the efficient level. - **Exogenous / endogenous** — determined outside the model versus jointly determined within it; the heart of identification. - **Confounder** — a variable that drives both treatment and outcome, faking a causal link. - **Counterfactual** — the outcome that would have occurred absent the treatment. - **Elasticity** — the percent change in one quantity per percent change in another. - **Moral hazard / adverse selection** — behavior changing after a contract / bad risks self-selecting into it before. - **Ceteris paribus** — holding all else constant. - heading: Tools markdown: >- - **Stata and R** — the workhorses for econometric estimation and the credibility-revolution toolkit (IV, diff-in-diff, RDD, fixed effects). - **FRED and national accounts** — the macro data backbone: GDP, employment, prices, rates, and the identities that link them. - **Microdata and administrative records** — household and firm panels where the individual optimization actually shows up. - **Econometric and structural models** — from a back-of-envelope supply/demand diagram to estimated general-equilibrium systems. - **The natural experiment** — policy discontinuities, lotteries, and arbitrary thresholds that supply exogenous variation for free. - heading: Collaboration markdown: >- Economists rarely work alone on questions that matter. They lean on statisticians and data scientists for inference, policy analysts and political scientists to translate findings into feasible institutions, and financial analysts and traders who apply the same equilibrium and expected-value logic to markets in real time. The recurring friction sits between the cleanly identified estimate and the messy decision: a policymaker wants a yes/no answer, and the honest economist owes them the effect size, its uncertainty, and the assumptions holding it up. - heading: Ethics markdown: >- Economists wield quiet influence: a recommended discount rate or elasticity can move billions. The first duty is to separate positive from normative — what *is* from what *ought to be* — and to flag when a value judgment has entered. Report uncertainty honestly rather than launder a preferred conclusion through a confident point estimate, and disclose who funds the work and what they want it to show. Distributional effects deserve to be named, not buried under aggregate efficiency, because the people who lose from an efficient policy are real. And there is the duty of humility: the model is a simplification and the forecast is conditional, and overstating either is its own kind of harm. - heading: Scenarios markdown: >- **Evaluating a minimum-wage increase.** The naive read compares employment before and after and blames any job loss on the wage. The economist instead asks for the counterfactual: what would employment have done anyway? They reach for a difference-in-differences design — a state that raised its wage versus a neighbor that didn't, differencing out the common trend — or a border discontinuity where two counties differ only in the law. They check that parallel trends hold, estimate the wage elasticity of employment, and report a range, not a verdict, then trace the general-equilibrium response: do firms cut hours, raise prices, or substitute capital? **A "free" infrastructure project.** A mayor calls a new bridge free because it's federally funded. The economist counts the opportunity cost — the next-best use of the money and land — against the marginal benefit (time saved times traffic times value of time), discounted to present value. If marginal benefit undershoots marginal cost, the project destroys value no matter who writes the check. They also flag induced demand: build it and people re-optimize their commutes, eroding the congestion relief. **An insurance market unraveling.** Premiums spike, healthy customers drop out, the average risk of those who remain rises, and premiums climb again. The economist diagnoses an adverse-selection death spiral driven by asymmetric information, then evaluates remedies by their incentive effects: a mandate keeps good risks in the pool, risk-rating prices them honestly — each with its own deadweight cost. - heading: Related Occupations markdown: >- The economist shares the quantitative temperament of several fields but is defined by causal reasoning about choice under scarcity. Statisticians supply the inferential machinery, though economists prize identification over pure prediction. Data scientists optimize forecast accuracy where economists chase causal effects and structural parameters. Financial analysts and traders apply equilibrium and expected-value reasoning to asset prices in real time. Policy analysts turn the estimates into feasible programs, and political scientists study the institutions economic incentives run through. - heading: References markdown: |- - *The Wealth of Nations* — Adam Smith - *The General Theory of Employment, Interest and Money* — Keynes - *Mostly Harmless Econometrics* — Angrist & Pischke - *Principles of Economics* — N. Gregory Mankiw - *The Use of Knowledge in Society* — Friedrich Hayek