Compendium

Microeconomics Foundations — Economics Reference

34 min read

Microeconomics Foundations

1. At a glance

Microeconomics studies how individual agents — households, workers, firms, governments acting at the program level — allocate scarce resources, and how those choices aggregate into market outcomes. It is the foundation of: pricing (what should a product cost?), market design (how should an auction or matching market be structured?), policy analysis (who bears the burden of a tax?), antitrust (is this merger anticompetitive?), labor economics (what determines wages?), behavioral nudges (how do framing + defaults change choice?), and mechanism design (how do we elicit truthful preferences?).

Three foundational moves define the discipline. First, optimization under constraint: agents maximize an objective (utility, profit) subject to budgets, technology, and information. Second, equilibrium: prices and quantities adjust so individual plans are mutually consistent. Third, welfare: outcomes are evaluated against Pareto efficiency or explicit social welfare functions, with market failures (externalities, public goods, asymmetric info) flagged as cases where decentralized choice diverges from the social optimum.

Modern microeconomics 2024-26 is heavily empirical. The “credibility revolution” (Leamer 1983 critique → Angrist + Imbens + Card 2021 Nobel) means most published applied micro papers use natural experiments, IV, DiD, or RDD rather than pure structural assumption. Behavioral economics (Kahneman 2002, Thaler 2017 Nobels) has displaced strict rationality in many policy contexts. Market design (Roth + Shapley 2012, Milgrom + Wilson 2020 Nobels) has produced real working markets — kidney exchange, school choice, spectrum auctions, ad markets.

See _index for sibling subdomains; probability-fundamentals + hypothesis-testing-mle + causal-inference for empirical tooling.

2. Supply and demand

The workhorse model. Demand Q_d(P) = quantity buyers want at price P, downward-sloping (law of demand — higher P, less Q, modulo Giffen exceptions). Supply Q_s(P) = quantity sellers will provide, typically upward-sloping (rising marginal cost). Equilibrium: Pwhere Q_s(P) = Q_d(P*).

Shifts vs movements. A movement along the curve is response to a price change. A shift of the curve is response to a non-price factor — income (demand shifters: normal goods shift out as income rises; inferior shift in), prices of related goods (substitutes vs complements), tastes, expectations, number of buyers (demand); input prices, technology, taxes/subsidies, expectations, number of sellers (supply). Confusing movement with shift is the most common intro-econ error.

Comparative statics. Predict how Pand Q change when a parameter shifts. Tax on sellers shifts supply up by t; new equilibrium has higher P paid by buyers and lower P received by sellers. Incidence (who bears the tax) depends on relative elasticities — the more inelastic side bears more, regardless of which side is statutorily taxed.

Surplus. Consumer surplus CS = ∫₀^Q*(P_d(q) − P*) dq, the area between demand curve and price. Producer surplus PS = ∫₀^Q*(P* − P_s(q)) dq. Total surplus TS = CS + PS measures gains from trade. Deadweight loss (DWL) = TS lost when output deviates from competitive equilibrium.

3. Elasticity

Price elasticity of demand E_d = (dQ/Q) / (dP/P) = (dQ/dP)(P/Q). Always negative for normal goods; convention often reports |E_d|.

  • Inelastic |E_d| < 1: food staples, gasoline short-run, insulin, addictive goods, necessities with no substitutes. A tax raises revenue without much quantity distortion.
  • Unit elastic |E_d| = 1.
  • Elastic |E_d| > 1: luxury goods, individual brands within a category, goods with close substitutes, goods that are large share of budget.
  • Perfectly inelastic vertical demand (|E_d| = 0): consumer takes any price; rare in pure form but approximated by emergency medical.
  • Perfectly elastic horizontal demand (|E_d| = ∞): infinite alternatives at the going price; approximation for a single firm in perfect competition.

Cross-price elasticity E_xy = (dQ_x/Q_x)/(dP_y/P_y). Positive → substitutes (Pepsi + Coke); negative → complements (printers + ink, cars + gasoline).

Income elasticity E_I = (dQ/Q)/(dI/I). Positive normal good; >1 luxury (fine dining); 0 < E_I < 1 necessity (rice, bread); negative inferior good (bus rides, generic noodles).

Determinants of |E_d| — availability of substitutes (more → more elastic), necessity vs luxury, time horizon (more time → more elastic, as habits + capital adjust; gasoline short-run inelastic, long-run elastic via fuel-efficient cars), and share of budget (bigger share → more elastic). Definition of the market matters: demand for “Pepsi” is elastic, demand for “carbonated soft drinks” less so, demand for “any caloric beverage” still less.

Revenue and elasticity. TR = P × Q. dTR/dP = Q(1 + 1/E_d). If demand is inelastic (|E_d| < 1), raising P raises revenue. Monopolists never optimally operate on the inelastic portion of demand — they would raise price and gain.

4. Consumer theory

The behavioral foundation: agents have preferences over bundles, those preferences satisfy axioms that yield a utility representation, and agents choose the most-preferred affordable bundle.

Preference axioms (Debreu 1954, Arrow-Debreu 1954). Completeness — any two bundles can be ranked. Transitivity — A ≽ B and B ≽ C implies A ≽ C. Continuity — small changes preserve ranking. Monotonicity (more is better) and convexity (averages preferred to extremes) are common added assumptions. Together with continuity, completeness + transitivity yield a continuous utility function u : ℝⁿ₊ → ℝ representing ≽.

Budget constraint. With prices p ∈ ℝⁿ₊ and income m, the consumer faces {x : p · x ≤ m}. Choice problem: max u(x) s.t. p · x ≤ m, x ≥ 0.

Marshallian (Walrasian) demand x*(p, m) — the utility-maximizing bundle as function of prices + income. Solves the consumer problem under fixed income.

Hicksian (compensated) demand h*(p, u) — solves the dual: min p · x s.t. u(x) ≥ u̅. Hicksian demand isolates substitution effect by holding utility (not income) constant.

Slutsky equation decomposes the total response to a price change: ∂x_i / ∂p_j = ∂h_i / ∂p_j − x_j (∂x_i / ∂m) The first term is the substitution effect (always ≤ 0 for own price, by concavity of expenditure function); the second is the income effect (sign depends on whether good is normal or inferior).

Giffen goods — inferior goods so heavily consumed (potatoes in 19th-century Ireland in the apocryphal story; Jensen + Miller 2008 found empirical Giffen behavior in rice + wheat in poor Chinese households) that the negative income effect of a price rise overwhelms the substitution effect, yielding upward-sloping demand. Veblen goods — luxuries whose perceived status rises with price (conspicuous consumption, Veblen 1899), giving apparent upward demand for distinct reasons.

Revealed preference (Samuelson 1938). If bundle A is chosen when B is affordable, A is revealed preferred to B (WARP, weak axiom). SARP (strong axiom) requires transitivity of revealed preference. WARP + SARP let you recover preferences from observed choice without assuming utility a priori — a foundation for behavioral + experimental work.

Indifference curves — loci of constant utility. Marginal rate of substitution MRS_xy = −dy/dx along indifference curve = MU_x / MU_y, the rate at which consumer is willing to trade y for x. At optimum, MRS = price ratio p_x / p_y.

Common utility forms:

  • Cobb-Douglas u(x, y) = x^α y^(1-α). Constant expenditure shares — fraction α of income spent on x regardless of prices. Workhorse for tractability.
  • CES (constant elasticity of substitution) u(x, y) = (αx^ρ + (1-α)y^ρ)^(1/ρ). Elasticity of substitution σ = 1/(1-ρ). Cobb-Douglas is the σ = 1 limit; Leontief σ = 0; perfect substitutes σ = ∞.
  • Quasilinear u(x, y) = v(x) + y. No income effect on x; convenient for partial-equilibrium welfare analysis.
  • Leontief u(x, y) = min(x/a, y/b). Perfect complements — goods consumed in fixed proportions (left + right shoes).

5. Producer theory

Production function y = f(x_1, …, x_n) — max output from inputs. Marginal product MP_i = ∂f/∂x_i, typically diminishing (diminishing marginal returns — adding more of one input holding others fixed yields diminishing additional output, Turgot 1767 in agriculture).

Isoquants — loci of constant output in input space. Marginal rate of technical substitution MRTS = MP_i / MP_j, the rate at which input j substitutes for input i holding y fixed. At cost-minimizing input mix, MRTS = w_i / w_j (input price ratio).

Cost function c(w, y) = min_x w · x s.t. f(x) ≥ y. Short-run: at least one input fixed (capital); cost decomposes into fixed cost F + variable cost VC(y). Long-run: all inputs variable.

Cost curves:

  • Marginal cost MC = dC/dy.
  • Average cost AC = C/y.
  • Average variable cost AVC = VC/y.
  • Average fixed cost AFC = F/y (declines monotonically with y).
  • AC and AVC are U-shaped under standard assumptions; MC cuts both at their minima.

Profit maximization. Choose y to max P(y) · y − c(y). FOC: MR = MC. Under perfect competition firm is price-taker → MR = P, so P = MC at optimum. Under monopoly MR < P (must lower price on all units to sell more), so MC = MR < P → markup.

Returns to scale. f(λx) = λ^k f(x): k > 1 increasing returns to scale (IRS — natural monopoly territory if persistent), k = 1 constant (CRS), k < 1 decreasing (DRS — limits firm size).

Common production forms:

  • Cobb-Douglas y = A x_1^α x_2^β. α + β controls returns to scale (α + β > 1 IRS, = 1 CRS, < 1 DRS).
  • Leontief y = min(x_1/a_1, x_2/a_2). Fixed proportions; no substitution.
  • CES y = A (αx_1^ρ + (1-α)x_2^ρ)^(1/ρ). Spans the substitution-elasticity continuum.

6. Market structures

The taxonomy classifies markets by number of sellers, product differentiation, and entry conditions.

Perfect competition — many small firms, homogeneous product, free entry/exit, perfect information, price-takers. Long-run equilibrium: P = MC = AC at minimum-AC point; zero economic profit (positive accounting profit covers opportunity cost of capital). First fundamental welfare theorem: competitive equilibrium is Pareto efficient.

Monopoly — single seller, no close substitutes, blocked entry (legal, technological, network, scale). Firm chooses (P, Q) facing whole market demand. MR = P(1 + 1/E_d) < P when E_d finite. Profit max MR = MC yields Q^M < Q^competitive and P^M > P^competitive. Deadweight loss = triangle between demand and MC over (Q^M, Q^competitive). Lerner index L = (P − MC)/P = −1/E_d measures markup power.

Monopolistic competition (Chamberlin 1933, Robinson 1933). Many firms, differentiated products (brand, location, quality), free long-run entry. Each firm faces downward-sloping demand for its own variant but entry erodes profit. Long-run zero profit at point of tangency between firm demand and AC. Excess capacity result: P > MC, Q below min-AC point — but variety provides welfare offset (Dixit + Stiglitz 1977 modeled this trade-off; foundation of new trade theory + new economic geography, Krugman 2008 Nobel).

Oligopoly — few firms, strategic interaction (each firm’s optimal action depends on rivals’ actions). Game-theoretic analysis required.

  • Cournot (1838) — firms simultaneously choose quantities, market clears at sum. Reaction functions q_i*(q_{-i}). Nash equilibrium: each firm best-responds to others. With n symmetric firms + linear demand: P^Cournot lies between monopoly and competition; (P − MC)/P = 1/(n · |E_d|). As n → ∞, approaches competition.
  • Bertrand (1883) — firms simultaneously choose prices, consumers buy from lowest. With homogeneous product + identical constant MC, unique NE: P = MC and zero profit even with two firms (Bertrand paradox). Resolutions: capacity constraints (Edgeworth 1925, Kreps + Scheinkman 1983 showed Cournot can be derived from Bertrand + capacity choice), product differentiation, repeated interaction.
  • Stackelberg (1934) — sequential quantities. Leader moves first, anticipates follower’s reaction. Leader gets larger share + higher profit than under Cournot; total industry quantity higher (closer to competition).
  • Cartel — explicit collusion to act as joint monopolist. OPEC (founded 1960), historical sugar + diamond cartels. Unstable: each member has incentive to overproduce. Sustained by repeated interaction + trigger strategies (folk theorem) + external enforcement.

Monopsony — single buyer. Often labor markets in concentrated local markets (Robinson 1933 coined the term). Firm faces upward-sloping labor supply; marginal cost of labor MCL > wage; profit max MRP = MCL yields wage below competitive level and employment below competitive. Empirically important: Card-Krueger 1994 minimum-wage natural experiment (NJ vs PA fast food after NJ raised min wage) found no employment loss + sometimes gains — consistent with monopsony power; consistent with Manning 2003 “Monopsony in Motion.”

Two-sided markets / platforms (Rochet + Tirole 2003, Caillaud + Jullien 2003, Evans 2003). Platform serves two groups whose values depend on each other’s participation (cross-side network effects). Examples: Apple App Store (developers + users), Uber (drivers + riders), Amazon Marketplace (sellers + buyers), Visa (merchants + cardholders), newspaper (readers + advertisers), OS (developers + users). Optimal pricing often skewed — one side priced below cost (sometimes negative — subsidies) to attract that side, recovered from the other. Implications for antitrust: standard market-power tests + relevant-market definition need adaptation; Amex case (US Supreme Court 2018) explicitly recognized two-sided market analysis.

7. Game theory

Strategic interaction is the engine of oligopoly, mechanism design, bargaining, voting, and most modern micro. Full treatment elsewhere; key concepts here.

Normal form game — set of players N, strategy sets S_i, payoff functions u_i : S → ℝ. Extensive form — game tree with information sets, capturing sequence + information structure.

Nash equilibrium (Nash 1950, 1951 — Nobel 1994). Strategy profile (s_1*, …, s_n*) where each s_iis best response to s_{-i}. Every finite game has at least one NE (possibly in mixed strategies).

Mixed strategies — randomization over pure strategies. Necessary when no pure NE exists (matching pennies; rock-paper-scissors). Interpretation: deliberate randomization, or as a Bayesian distribution over opponent types (Harsanyi 1973).

Subgame perfect equilibrium (Selten 1965, 1975 — Nobel 1994). Refinement for extensive games: strategy profile that constitutes NE in every subgame. Eliminates non-credible threats. Solved by backward induction in finite games.

Repeated games + folk theorem (Friedman 1971, Fudenberg + Maskin 1986). In infinitely repeated games with sufficient patience (discount factor δ near 1), almost any feasible + individually rational payoff vector is sustainable as a SPE. Cooperation in PD can be sustained via trigger strategies, tit-for-tat, grim trigger. Cartel stability falls out as application.

Canonical 2×2 games:

  • Prisoner’s dilemma — dominant strategy is defect, but (cooperate, cooperate) Pareto-dominates (defect, defect). Public goods provision, arms races, cartel cheating.
  • Stag hunt — two pure NE: (stag, stag) Pareto-superior + (hare, hare) risk-dominant. Coordination problem with payoff-vs-risk trade-off (Skyrms 2004).
  • Battle of the sexes — two pure NE corresponding to coordination on either party’s preferred outcome + a mixed NE.
  • Chicken / hawk-dove — two pure NE (each player swerves while other doesn’t); models brinkmanship + bargaining.
  • Matching pennies — zero-sum, no pure NE, unique mixed NE.

Bayesian games + signaling (Harsanyi 1967-68 — Nobel 1994). Players have private information (types) drawn from a known distribution. Bayes-Nash equilibrium = NE in the expanded game with types. Signaling games (Spence 1973) — informed party moves first, sends signal that reveals information when separating equilibrium exists. Screening — uninformed party moves first, offers menu that induces self-selection.

Mechanism design — reverse game theory. Given desired outcomes + private information, design rules so equilibrium yields the outcome. Revelation principle (Myerson 1979, Gibbard 1973) — without loss, restrict to direct mechanisms in which truthful reporting is equilibrium. Vickrey-Clarke-Groves (VCG) mechanisms (Vickrey 1961, Clarke 1971, Groves 1973) — truthful + efficient in quasilinear environments. Myerson’s optimal auction (1981) — revenue-maximizing single-item auction under independent private values. Hurwicz + Maskin + Myerson shared the 2007 Nobel for mechanism-design foundations.

8. Welfare and market failure

First welfare theorem. Every competitive equilibrium with complete markets, no externalities, and price-taking behavior is Pareto efficient. The invisible hand formalized (Smith 1776 → Arrow + Debreu 1954).

Second welfare theorem. Every Pareto-efficient allocation can be supported as a competitive equilibrium given suitable lump-sum redistribution of initial endowments. Separates efficiency from distribution — in principle, society can pursue any equitable allocation through transfers + markets. In practice, lump-sum transfers are rare; real-world taxation distorts.

Externalities (Pigou 1920). A cost or benefit imposed on a third party not reflected in market price. Negative externality (pollution, congestion, antibiotic resistance) → overproduction relative to social optimum. Positive externality (vaccination, R&D spillovers, basic research) → underproduction. Pigouvian tax/subsidy — set tax = marginal external damage to internalize. Carbon pricing (Sweden 1991, EU ETS 2005, California 2013, Canada 2019) is the flagship application.

Coase theorem (Coase 1960 — Nobel 1991). If property rights are well-defined and transaction costs are zero, parties will bargain to an efficient allocation regardless of who holds the initial right. The distribution of rents depends on initial assignment, but the level of the externality does not. In practice transaction costs are rarely zero — Coase himself emphasized this — so the theorem is mostly a benchmark for identifying when bargaining might substitute for regulation (small numbers, well-defined rights, low transaction cost).

Public goods — non-rival (one person’s use doesn’t reduce another’s) + non-excludable (cannot prevent free use). National defense, basic research, clean air, lighthouses (Coase 1974 disputed canonical lighthouse example), open-source software, public-health pathogen surveillance. Free-rider problem: each agent’s privately optimal contribution is zero; total provision below social optimum. Samuelson condition (1954): efficient provision when Σ_i MRS_i = MRT (sum of marginal willingness to pay across all consumers = marginal cost of production). Solutions: government provision financed by taxation, subscription (clubs), Lindahl pricing, voluntary contribution mechanisms with matching.

Commons / common-pool resources — rival but non-excludable (fisheries, groundwater aquifers, atmospheric carbon, common grazing land). Hardin 1968 “Tragedy of the Commons” — open access leads to overuse + degradation. Ostrom 1990 + 2009 Nobel — empirical fieldwork on lobster fisheries, irrigation, forests showed community management with locally crafted rules can sustain commons without state ownership or privatization; identified design principles (clear boundaries, congruent rules, collective choice, monitoring, graduated sanctions, conflict resolution, recognition by external authorities, nested enterprises). Foundational for environmental governance, water + fishery management, climate cooperation.

Information asymmetry. Akerlof + Spence + Stiglitz shared 2001 Nobel.

  • Adverse selection — pre-contract hidden information. Akerlof 1970 “Market for Lemons” — used-car market may unravel when sellers know quality + buyers don’t; only the worst cars remain. Insurance markets: high-risk buyers self-select into coverage; market may unravel without mandates or pooling. Solutions: signaling, screening, warranties, mandates.
  • Moral hazard — post-contract hidden action. Insurance reduces care taking; principal-agent in firms (manager effort, CEO compensation); credit markets (debtor risk-taking with limited liability). Solutions: monitoring, incentive contracts (Holmström 1979, Hart + Holmström 2016 Nobel), deductibles + copays.
  • Signaling (Spence 1973) — informed party takes costly action that reveals type. Education-as-signal: even if school adds no human capital, college degree separates high-ability from low-ability if the cost of completing it is lower for high-ability. Separating equilibrium possible when signaling cost is correlated with type.
  • Screening (Rothschild + Stiglitz 1976, Stiglitz 1975). Uninformed party offers menu; agents self-select. Insurance menus (high deductible / low premium vs low deductible / high premium); airline price discrimination (advance-purchase, Saturday-night-stay restrictions).

Property rights + transaction costs (Coase 1960, 1991 Nobel; Williamson 2009 Nobel + Hart-Holmström 2016 Nobel). Firms exist because internal hierarchy economizes on transaction costs that markets cannot — bounded rationality, opportunism, asset specificity. The boundary of the firm is where marginal transaction cost in market = marginal cost of internal organization. Incomplete contracts (Hart + Moore 1990) — when contracts cannot specify all contingencies, ownership of residual rights matters.

9. Empirical microeconomics 2024-26

Modern applied micro is dominated by causal identification. The mantra: correlation is not causation; we need a credible source of exogenous variation. See causal-inference for full toolkit.

RCTs — Banerjee + Duflo + Kremer 2019 Nobel for randomized field experiments in development economics. J-PAL (MIT) + IPA + 3ie networks. Examples: deworming + school attendance (Miguel + Kremer 2004), microcredit (Banerjee et al. 2015 six-country RCT — modest effects), conditional cash transfers (Mexico PROGRESA/Oportunidades), graduation programs (Banerjee et al. 2015 Science).

Quasi-experimental — Angrist + Imbens + Card 2021 Nobel. Card + Krueger 1994 NJ-PA min-wage natural experiment. Angrist + Krueger 1991 quarter-of-birth IV for returns to schooling. Imbens + Angrist 1994 LATE theorem — IV identifies local average treatment effect for compliers, not the population ATE. Regression discontinuity (Thistlethwaite + Campbell 1960 → Lee + Lemieux 2010 modern). Difference-in-differences (Card 1990 Mariel boatlift) → Goodman-Bacon 2021 + Callaway + Sant’Anna 2021 + de Chaisemartin + d’Haultfœuille 2020 on staggered-treatment DiD pitfalls.

Structural estimation — Berry-Levinsohn-Pakes 1995 BLP demand estimation for differentiated products (cars). Solves endogeneity of price + observed/unobserved characteristics via random-coefficients logit + instruments (BLP instruments based on rival products’ characteristics). Used in merger simulation (DOJ Horizontal Merger Guidelines reference); generalized to many discrete-choice settings.

Text-as-data + ML in econometrics — Gentzkow + Shapiro + Taddy 2019 “Text as Data” review. Athey + Imbens + Wager + Chernozhukov on ML for heterogeneous treatment effects (causal forests, double ML). Conformal inference (Vovk; Lei + Wasserman 2014) gives distribution-free coverage. Use cases: heterogeneous treatment effects of charter schools, criminal-justice interventions, ML estimation of demand from scanner data.

10. Behavioral economics

Departure from strict rationality, grounded in psychology + experiments.

Prospect theory (Kahneman + Tversky 1979, 1992 cumulative — Kahneman 2002 Nobel; Tversky died 1996 before the prize). Four components:

  1. Reference-dependence — utility is over gains + losses relative to a reference point, not over wealth levels.
  2. Loss aversion — losses loom larger than equivalent gains (λ ≈ 2 in early estimates).
  3. Diminishing sensitivity — concave over gains, convex over losses (S-shaped value function).
  4. Probability weighting — overweighting of small probabilities (why lottery + insurance coexist) + underweighting of large.

Heuristics + biases (Tversky + Kahneman 1974). Availability (judge probability by ease of recall — overestimate plane crashes, underestimate diabetes). Anchoring (estimates pulled toward initial value — judges’ sentences influenced by rolled dice in lab). Representativeness (judge by similarity to stereotype, ignoring base rates — Linda problem). Confirmation (seek confirming evidence). Status-quo + default effects (Madrian + Shea 2001 on 401(k) auto-enrollment, Johnson + Goldstein 2003 on organ donation defaults). Framing (preferences reverse with reframing — Asian disease problem).

Intertemporal choice (Thaler 2017 Nobel; Laibson 1997 on hyperbolic discounting). Present bias — quasi-hyperbolic (β, δ) discounting weights immediate consumption disproportionately; leads to procrastination + savings shortfall + dieting failure. Mental accounting — money treated as non-fungible across labeled categories (Thaler 1985, 1999). Nudge (Thaler + Sunstein 2008) — libertarian paternalism using defaults + framing to improve outcomes while preserving choice. UK Behavioural Insights Team + US OIRA + dozens of nudge units worldwide.

Bounded rationality (Simon 1957, 1978 Nobel) — agents satisfice rather than optimize under cognitive constraint. Foundational for behavioral + organizational economics.

Endowment effect (Kahneman + Knetsch + Thaler 1990 mugs experiment) — willingness to accept >> willingness to pay for an item you own; consistent with loss aversion. Sunk cost fallacy — past unrecoverable costs influence forward-looking decisions; strictly irrational under standard theory but pervasive.

Experimental economics (Vernon Smith 2002 Nobel). Controlled lab markets — double-auction experiments converge to competitive prediction under minimal conditions (Smith 1962). Public-goods games show partial cooperation with conditional cooperators + free-riders. Ultimatum game (Güth + Schmittberger + Schwarze 1982) — fairness concerns reject low offers, refuting pure self-interest. Field experiments (Levitt + List 2009 critique on external validity) bridge to RCT tradition.

11. Pricing and market design

Auctions (Vickrey 1961 — Nobel 1996; Milgrom + Wilson 2020 Nobel for auction theory + design).

  • First-price sealed-bid — highest bid wins, pays own bid. Bidders shade below value; equilibrium bid increases in value with shading factor depending on N and distribution.
  • Second-price sealed-bid / Vickrey — highest bid wins, pays second-highest. Truthful bidding is weakly dominant strategy. Revenue equivalence with first-price under independent private values + risk-neutrality + symmetric bidders (Vickrey 1961, Myerson 1981, Riley + Samuelson 1981).
  • English ascending — open bidding rises until one bidder remains. Strategically equivalent to second-price under IPV.
  • Dutch descending — price falls until first bidder accepts. Strategically equivalent to first-price.
  • FCC spectrum auctions (1994 onward) — simultaneous multi-round auction (SMR) designed by Milgrom + Wilson + McAfee for selling licenses with complementarities. Incentive auction 2017 — two-sided clock auction repurposing broadcast TV spectrum to wireless; raised $19.8B.
  • Ad auctions — Google AdWords / Search Ads use Generalized Second Price (GSP, Edelman + Ostrovsky + Schwarz 2007 + Varian 2007) initially; later moved partially to VCG-style + first-price for display (2019 Google switch). Reserve prices + bid shading; quality-adjusted bids.
  • Double auctions — buyers + sellers submit bids/asks; market clears at crossing price. Stock exchange opening + closing auctions, electricity day-ahead markets.

Matching theory (Roth + Shapley 2012 Nobel). Markets where prices alone cannot clear because money is absent or rationed.

  • Deferred-acceptance algorithm (Gale + Shapley 1962) — solves stable-marriage problem. Each side proposes/accepts iteratively until no instability remains. Yields stable matching; proposer-optimal among stable matchings.
  • NRMP — National Resident Matching Program (US medical residency) used variants since 1952; Roth 1984 analyzed + Roth + Peranson 1999 redesigned in response to couples’ constraints.
  • School choice — Boston (Boston mechanism, manipulable) replaced with deferred-acceptance variants (Abdulkadiroğlu + Pathak + Roth 2005, 2009); NYC public high school 2003-04 redesign.
  • Kidney exchange (Roth + Sönmez + Ünver 2004, 2007) — chains + cycles let incompatible donor-patient pairs swap. National Kidney Registry + UNOS systems; thousands of additional transplants per year.

Two-sided platforms + multi-sided pricing (Rochet + Tirole 2003, Armstrong 2006). Pricing structure (which side bears cost) shaped by cross-side network effects + elasticities; one side often subsidized.

Price discrimination:

  • First-degree (perfect) — charge each buyer their reservation value. Captures full consumer surplus; rare in pure form (custom B2B contracts, haggling at car dealers). Some online retailers attempt via personalized pricing.
  • Second-degree — non-linear pricing + versioning. Buyers self-select. Quantity discounts, two-part tariffs (Disney+ tiers, AWS storage), versioning (basic vs premium software, hardback vs paperback timing).
  • Third-degree — segment markets by observable characteristic. Student + senior discounts, geographic pricing, hardback-vs-paperback timing, peak-vs-off-peak pricing. Welfare ambiguous — can raise + lower total surplus depending on demand curvature + market expansion.

Bundling + tying. Bundling (offering goods only together — cable TV channels) extracts more surplus when valuations are negatively correlated (Stigler 1963, Adams + Yellen 1976). Tying (sale of A conditional on buying B) — Microsoft Windows + IE (US 1998, EU 2004), Apple App Store + IAP (Epic v. Apple 2020-24). Antitrust analyzes whether tying leverages market power from one market to another.

Dynamic pricing + revenue management. Airlines (since 1980s deregulation), hotels, Uber surge, Amazon dynamic prices. Inventory + capacity constrained; demand uncertain. Tools: yield management algorithms, machine learning demand estimation, reinforcement learning.

Network effects + lock-in. Direct network effects — value grows with user base (telephones, social networks). Indirect / cross-side — more users attract more developers attract more users (operating systems, video-game consoles). Switching costs + lock-in — proprietary file formats, training, contracts; create market power even ex post (Klemperer 1995). Standards wars — VHS vs Betamax (1976-88), Blu-ray vs HD-DVD (2006-08), HTTPS vs HTTP transition. Tipping + winner-take-most outcomes common.

12. Industrial organization

Concentration. Herfindahl-Hirschman Index HHI = Σ_i s_i² where s_i is firm i’s market share in percentage points. HHI ≤ 1500 unconcentrated, 1500-2500 moderate, > 2500 highly concentrated (DOJ + FTC 2023 Merger Guidelines updated thresholds 2023). ΔHHI from a merger flags scrutiny.

Entry + barriers. Capital requirements, regulation (licensing, certification), network effects, learning curves, switching costs, brand loyalty, IP, strategic deterrence (excess capacity, limit pricing, product proliferation).

Antitrust:

  • US — Sherman Act 1890 (§1 conspiracies, §2 monopolization), Clayton Act 1914 (mergers + price discrimination + tying + exclusive dealing), FTC Act 1914 (§5 unfair methods). Enforced by DOJ Antitrust Division + FTC. Key cases: Standard Oil 1911, AT&T 1982 breakup, Microsoft 1998-2001, Google Search (DOJ 2020 filed → liability ruling Aug 2024; remedies 2025-26), Google Ad Tech (DOJ + states 2023 filed → liability ruling Apr 2025), Apple (DOJ + states Mar 2024 filed; Epic v. Apple 9th Cir. 2023 + cert denied Jan 2024), Meta (FTC 2020 ongoing), Amazon (FTC + 17 states Sep 2023 ongoing).
  • EU — TFEU Article 101 (anticompetitive agreements) + Article 102 (abuse of dominance) + EU Merger Regulation. DG COMP enforcement + national competition authorities. Digital Markets Act (in force 2023, gatekeeper designations 2023-24) imposes ex-ante obligations on Apple, Meta, Google, Amazon, Microsoft, ByteDance, Booking.com. Major cases: Microsoft 2004 + 2007 + 2013, Google Shopping 2017, Google Android 2018, Google AdSense 2019, Intel 2009 (annulled 2022 + retried), Apple Ireland tax 2016 (ECJ 2024 affirmed €13B recovery), Apple App Store 2024 (€1.8B fine).
  • Modern enforcement themes — algorithmic collusion (RealPage litigation 2022+, Topkins criminal case 2015), platform self-preferencing, killer acquisitions (Cunningham + Ederer + Ma 2021 on pharma), labor-market antitrust (no-poach + non-compete; FTC rule 2024 struck down in district court 2024, on appeal).

Vertical relationships. Double marginalization (Spengler 1950) — successive monopolists each apply markup; vertical integration eliminates the second markup, lowering price + raising surplus. Resale price maintenance (RPM) — manufacturer sets minimum retail price; per se illegal under US law until Leegin 2007 (rule of reason). Exclusive dealing — distributor sells only one brand; can be efficient (relationship-specific investment) or anticompetitive (foreclosure).

R&D + innovation. Patents (statutory monopoly 20 years from filing) trade ex-post deadweight loss for ex-ante invention incentive. Cumulative innovation (Scotchmer 1991) — when later inventions build on earlier, optimal patent breadth + length differ from one-shot case. Patent thickets + anti-commons (Heller 1998) — overlapping rights block use. Prize mechanisms as alternative (Kremer + Williams 2010, Williams 2013 on genome).

13. Labor economics

Labor supply. Worker chooses hours h to max u(c, ℓ) s.t. budget where ℓ = T − h is leisure. Wage w determines slope; income + substitution effects on hours can offset. Backward-bending labor supply at high w empirically observed in some contexts.

Labor demand. Firm hires L until VMP_L = MRP_L = w (value of marginal product = wage). Slopes downward under diminishing marginal product.

Compensating differentials (Rosen 1986). Riskier or less pleasant jobs pay more in equilibrium to compensate; used to estimate value of statistical life (VSL ≈ $10M in US 2020s) from wage-risk premia.

Human capital (Becker 1964; Becker 1992 Nobel; Mincer 1974). Schooling + on-the-job training raise productivity → wages. Mincer equation: log wage = α + β · years_education + γ · experience + δ · experience² + ε. Returns to schooling ≈ 8-10% per year typical estimate; IV estimates (compulsory schooling laws — Angrist + Krueger 1991) similar or higher than OLS.

Minimum wage. Standard supply-demand prediction: above-equilibrium minimum reduces employment. Card + Krueger 1994 NJ-PA natural experiment — NJ raised min wage 1992, comparison with PA fast-food employment showed no negative effect + possible positive. Generated long debate (Neumark + Wascher critiques; Cengiz + Dube + Lindner + Zipperer 2019 + Dube 2019 reviews); current consensus is small-to-zero disemployment for moderate increases. Theoretical reconciliation: monopsony power, search frictions, efficiency wages.

Search + matching (Mortensen + Pissarides + Diamond 2010 Nobel). Workers + firms search; meetings are stochastic via a matching function. Nash bargaining over surplus splits the gains. Yields wage dispersion among identical workers (Mortensen 2003 “Why Are Similar Workers Paid Differently?”), positive unemployment in equilibrium, and Beveridge curve (negative relation between vacancies + unemployment).

Discrimination. Taste-based (Becker 1957) — employers/coworkers/customers have disutility from a group; market competition can erode discrimination over time (but slowly, with frictions). Statistical (Phelps 1972, Arrow 1973) — group membership used as proxy for unobserved productivity under information asymmetry; can be self-reinforcing (Coate + Loury 1993). Audit + correspondence studies (Bertrand + Mullainathan 2004 “Are Emily + Greg More Employable than Lakisha + Jamal?”) consistently find substantial discrimination in hiring callbacks by race + name.

Gig + platform labor. Uber, DoorDash, Instacart, TaskRabbit. Classification debates: independent contractor vs employee. California AB5 2019 codified ABC test; Prop 22 2020 carved out gig drivers; California Supreme Court upheld Prop 22 Jul 2024. UK Supreme Court Uber BV v. Aslam 2021 ruled drivers are workers (intermediate UK category). EU Platform Work Directive adopted Oct 2024, implementation 2026 — presumption of employment + algorithmic management transparency. AI Act 2024 + algorithmic transparency.

AI + automation (Acemoglu + Restrepo 2018, 2020, 2022; Autor 2015, 2019, 2024). Two effects: displacement (tasks automated, workers exit) + reinstatement (new tasks created). Recent productivity work — Brynjolfsson + Li + Raymond 2023 on customer support (+14% productivity, larger gains for novices); Noy + Zhang 2023 ChatGPT writing tasks (+37% productivity); Peng et al. 2023 GitHub Copilot (+55% on coding tasks). Acemoglu + Autor 2024 on AI macroeconomic effects — modest aggregate impact under conservative assumptions, larger if AI both substitutes + complements heterogeneously.

14. Public economics

Optimal taxation. Ramsey 1927 commodity tax — inverse-elasticity rule for raising revenue with minimum DWL. Mirrlees 1971 (Nobel 1996) optimal income tax — trade-off between redistribution + incentive cost when ability is unobserved; optimal top marginal rate ≈ 1/(1 + a·e) where a is Pareto parameter + e is elasticity (Diamond + Saez 2011 — optimal top rate 50-70%+ under reasonable parameters). Optimal capital tax (Atkinson + Stiglitz 1976) under separability — zero capital tax with optimal nonlinear income tax; relaxed by subsequent work (Saez 2002, Piketty + Saez 2013, Stantcheva 2017+).

Tax incidence. Statutory (who legally pays) differs from economic (who bears burden after price adjustment). Determined by relative elasticities — more inelastic side bears more. Payroll tax split between employer + employee is statutory; economic incidence largely on workers (less elastic labor supply).

Social insurance. Unemployment insurance (Baily 1978 + Chetty 2006, 2008 — moral hazard vs liquidity vs insurance value), Social Security (DI + retirement), Medicare + Medicaid + ACA (Finkelstein et al. 2008+ Oregon Medicaid experiment), workers’ comp.

Education + healthcare provision. Public good + externality + redistribution + asymmetric information rationales for government role. School-finance equalization, school-choice debates, voucher experiments (Sweden, Chile, US Milwaukee + DC), Medicare for All vs targeted expansions, NHS structure.

Cost-benefit analysis. Discount future benefits + costs (social discount rate debate — Stern 2006 used 1.4%, Nordhaus used 4-5%, IPCC + OMB Circular A-4 ranges). DALY (disability-adjusted life year) + QALY (quality-adjusted life year) for health interventions (NICE UK uses ~£20-30K/QALY threshold; institutionalized in many national health-tech assessments).

15. Recent topics 2024-26

AI economic impacts. Brynjolfsson + McAfee + Mollick + Acemoglu + Autor + Goldin + Korinek + Trammell debates. Productivity: positive in narrow tasks; aggregate effect contested. Displacement: occupation-level studies (Eloundou + Manning + Mishkin + Rock 2023 GPTs occupations) suggest substantial exposure for knowledge work. Power concentration: gains may accrue to AI-firm + complementary-capital owners; labor share concerns.

Crypto + DeFi + CBDC. Crypto winter 2022-23 + FTX collapse Nov 2022 + Binance + Tornado Cash enforcement 2023 → ETF approval Jan 2024 + revived activity 2024-26. CBDC — Chinese e-CNY in pilot (multiple cities, billions of yuan); ECB digital euro preparation phase Nov 2023 → 2025-26 decision; US Project Hamilton + FedNow 2023 (latter is not CBDC, faster payments). DeFi + tokenized real-world assets (BlackRock BUIDL 2024).

Inflation + monetary policy. Post-COVID supply chain disruption + fiscal expansion + commodity shocks (Russia-Ukraine war Feb 2022) → US CPI peaked 9.1% June 2022; Fed hiked from 0-0.25% to 5.25-5.50% by Jul 2023, held through Aug 2024, eased 100 bp Sep-Dec 2024, paused 2025. ECB + BoE similar trajectory. Phillips curve debates revived — flat or steep? Bernanke + Blanchard 2023 inflation-shock decomposition.

Industrial policy renaissance. US CHIPS + Science Act Aug 2022 (369B clean energy + IRA tax credits driving battery + EV + solar siting). EU Green Deal Industrial Plan Feb 2023 + Net-Zero Industry Act Jun 2024 + Critical Raw Materials Act May 2024. China made-in-China 2025 + dual-circulation. Debates: efficiency cost (Krugman, Furman) vs security + climate rationale (Rodrik, Tooze, Mazzucato); Acemoglu + Robinson on institutional risk.

Climate economics. Nordhaus 2018 Nobel DICE integrated assessment model. Stern Review 2006 — low discount rate → urgent action. Weitzman 2009 dismal theorem — fat tails of catastrophic damage dominate cost-benefit. Social cost of carbon updates — EPA 2023 raised to ~50). Carbon border adjustments — EU CBAM transitional Oct 2023, definitive Jan 2026. See _index (when written).

AI alignment + governance as economic problem — institutional design for AGI development (Korinek + Trammell 2024, Hadfield, Bostrom + Ord), liability allocation (EU AI Act 2024 + AI Liability Directive), antitrust on frontier compute + model markets.

Inequality. Piketty 2014 “Capital in the Twenty-First Century” + Piketty + Saez + Zucman work (DINA distributional national accounts), Chetty + Hendren on intergenerational mobility (Opportunity Atlas), Auten + Splinter 2024 critique of top-share trend magnitudes. Wealth tax proposals (Saez + Zucman); cross-country comparisons (WID).

16. Tools and software

Statistical software: Stata (most-cited in applied micro), R (fixest, did, synthdid, causalweight, MatchIt, AER), Python (pandas, statsmodels, linearmodels, pyfixest, econml, causalml, doubleml), Matlab (legacy in structural + macro), Julia (Econometrics.jl, FixedEffectModels.jl).

Datasets: NBER (working papers + data archives), IPUMS (microdata harmonization: CPS, Census, ACS, international), UNIDO INDSTAT, World Bank WDI + Enterprise Surveys, Penn World Table (cross-country GDP + capital + labor), ICPSR (consortium), IPEDS (US higher-ed), Compustat + CRSP (firm + stock — paid), Orbis + Bureau van Dijk, EU-LFS + EU-SILC, BLS CPS + ATUS, BEA NIPA + IO tables, Eurostat, Lindahl + LISA registry data (Sweden).

Microsimulation: Tax-Calculator (US, Open Source Policy Center), OECD TaxBEN, EUROMOD (EU tax-benefit), TPC + ITEP + JCT models, CBO microsim.

Survey + administrative microdata access: FSRDC (US Federal Statistical Research Data Centers), HMDA, IRS SOI, ATP (Nordic), German IAB SIAB + LIAB.

17. Common pitfalls

  • Nominal vs real — failing to deflate nominal series by appropriate price index; comparing wages across decades in nominal terms.
  • Accounting vs economic profit — economic profit subtracts opportunity cost of all factors (including owner’s labor + equity); positive accounting profit can mean zero economic profit.
  • Sunk cost — past unrecoverable costs are irrelevant for forward-looking decisions; pervasive fallacy.
  • Marginal vs average — pricing + production decisions hinge on marginal, not average; averages mislead.
  • Partial vs general equilibrium — partial-equilibrium answer can reverse in GE (tariff incidence, tax incidence with multiple markets).
  • Assuming perfect competition where market power exists — labor monopsony, hospital markets, online platforms.
  • Cherry-picking studies — single-study reliance vs systematic review (preregistration + meta-analysis preferred).
  • Cargo-cult RCT — randomization without thinking about external validity, scaling, general-equilibrium effects (Acemoglu critique; Deaton 2010, 2019 Nobel lecture).
  • p-hacking + specification searching — addressed by preregistration (AEA RCT Registry mandatory since 2018), multiple-hypothesis correction, replication archives (AEA Data Editor since 2019).
  • Confusing correlation with causation without identification strategy.

18. Cross-references

19. Citations

Graduate texts: Mas-Colell + Whinston + Green “Microeconomic Theory” 1995 (the standard); Kreps “Microeconomic Foundations I + II” 2013 + 2023; Jehle + Reny “Advanced Microeconomic Theory” 3rd ed 2011.

Undergraduate / intermediate: Varian “Intermediate Microeconomics” 9th ed 2014; Varian “Microeconomic Analysis” 3rd ed 1992 (intermediate-grad bridge); Mankiw “Principles of Microeconomics” 9th ed 2020; Acemoglu + Laibson + List “Microeconomics” 3rd ed 2024; Goolsbee + Levitt + Syverson “Microeconomics” 3rd ed 2019.

Specialty: Tirole “The Theory of Industrial Organization” 1988 (IO classic; Tirole 2014 Nobel); Belleflamme + Peitz “Industrial Organization” 2nd ed 2015; Cabral “Introduction to Industrial Organization” 2nd ed 2017; Roth “Who Gets What — and Why” 2015 (matching); Milgrom “Putting Auction Theory to Work” 2004; Kahneman “Thinking, Fast and Slow” 2011; Thaler + Sunstein “Nudge” 2008/2021; Thaler “Misbehaving” 2015; Banerjee + Duflo “Poor Economics” 2011 + “Good Economics for Hard Times” 2019; Angrist + Pischke “Mostly Harmless Econometrics” 2009 + “Mastering ‘Metrics” 2014; Cunningham “Causal Inference: The Mixtape” 2021.

Nobel laureates referenced (selected): Samuelson 1970; Arrow 1972; Hayek 1974; Simon 1978; Stigler 1982; Debreu 1983; Buchanan 1986; Coase 1991; Becker 1992; Nash + Selten + Harsanyi 1994; Lucas 1995; Mirrlees + Vickrey 1996; Akerlof + Spence + Stiglitz 2001; Kahneman + Smith 2002; Hurwicz + Maskin + Myerson 2007; Krugman 2008; Ostrom + Williamson 2009; Diamond + Mortensen + Pissarides 2010; Roth + Shapley 2012; Tirole 2014; Deaton 2015; Hart + Holmström 2016; Thaler 2017; Nordhaus + Romer 2018; Banerjee + Duflo + Kremer 2019; Milgrom + Wilson 2020; Card + Angrist + Imbens 2021; Bernanke + Diamond + Dybvig 2022; Goldin 2023; Acemoglu + Johnson + Robinson 2024.

Foundational papers cited: Akerlof 1970 QJE; Spence 1973 QJE; Stiglitz + Rothschild 1976 QJE; Kahneman + Tversky 1979 Econometrica; Vickrey 1961 J Finance; Myerson 1981 Math Op Res; Gale + Shapley 1962 AMM; Hardin 1968 Science; Samuelson 1954 REStat; Coase 1937 Economica + 1960 JLE; Card + Krueger 1994 AER; Banerjee + Duflo et al. 2015 Science; Roth + Sönmez + Ünver 2004 QJE; Rochet + Tirole 2003 JEEA; Berry + Levinsohn + Pakes 1995 Econometrica; Mortensen + Pissarides 1994 RES.

Graph View