Nine weeks of material — from supply and demand through oligopoly and price discrimination — condensed into working notes, with an interactive calculator wherever the underlying math is concrete enough to play with.
Everything downstream in this course builds on two curves crossing. What matters isn't just where they cross, but how sensitive each side is to price — that sensitivity is elasticity, and it quietly drives almost every later result about taxes, monopoly pricing, and price discrimination.
Demand curves slope down. The extensive margin is people leaving the market entirely as price rises; the intensive margin is existing buyers simply buying less.
Normal goods: demand rises with income. Inferior goods: demand falls as income rises.
Substitutes (coffee & tea): price of A up, demand for B up. Complements (burgers & buns): price of A up, demand for B down.
Demand shifts with tastes, income, and prices of other goods. Supply shifts with input prices and technology — distinct from a movement along a curve, which is a response to that good's own price.
In the short run, existing producers make more. In the long run, new producers enter.
A pure sensitivity measure: how many percent does quantity move for each 1% move in price? It's unit-free, which is why it — not slope — is the right way to compare how "responsive" wildly different goods are.
Elastic (ε < −1): quantity swings a lot. Inelastic (−1 < ε < 0): quantity barely moves. Unit elastic: ε = −1 exactly.
Narrower product definition (a specific brand vs. "food"), a longer time horizon (people can adjust), and a narrower market definition all push elasticity toward more negative (more elastic).
Enter a linear demand curve (QD = ad − bdP) and linear supply curve (QS = as + bsP). Setting QD = QS and solving for P is exactly the "set demand equal to supply" step from the sheet.
Enter two price/quantity points from a demand (or supply) curve to compute the arc elasticity between them, and see how it's classified.
Consumer and producer surplus turn "the market makes people better off" into an actual number, and give a precise way to measure how much a tax, a price ceiling, or a monopoly costs society beyond the money that changes hands.
CS = Willingness to pay − Price actually paid, summed across all units bought. Geometrically, the area under the demand curve and above the market price.
The gap between the market price and the minimum price a producer would accept — the area above the supply curve and below market price.
Competitive equilibrium maximizes total surplus, but only under conditions that often fail in practice: perfect information, perfect competition (no monopolies), and no externalities.
Byproducts of a transaction that affect people who weren't part of it. Negative: pollution from a factory. Positive: knowledge spillovers from education.
Ceiling: a legal maximum price (rent control) — binds only if set below equilibrium. Floor: a legal minimum price (minimum wage) — binds only if set above equilibrium.
The side of the market with the less elastic response bears more of the tax — intuitively, whoever finds it harder to walk away from the deal ends up eating more of the cost, regardless of who the tax is legally levied on.
Enter demand and supply elasticities (demand is typically negative) at the pre-tax equilibrium, and the size of a per-unit tax, to see how the burden splits.
The more inelastic side always pays the larger share — check what happens as εS approaches 0 (perfectly inelastic supply, like land) versus a very large εS.
The consumer's problem is a constrained optimization: pick the best affordable bundle. Graphically that's the highest indifference curve touching the budget line; mathematically it's the same tangency condition rearranged into something you can check with real numbers.
Preferences are complete (any two bundles can be ranked) and transitive (prefer 1 to 2, and 2 to 3, implies prefer 1 to 3) — together these define rationality. "More is weakly better" is a separate, third assumption.
The more of a good you already have, the less an additional unit is worth to you — the reason indifference curves bow inward.
— the rate at which the market lets you trade good X for good Y.
— the rate at which you are personally willing to trade X for Y, holding satisfaction constant. A steeper indifference curve means you're more eager for X relative to Y.
The "bang for the buck" form on the right is the practical version: at the optimal bundle, the last dollar spent on X buys exactly as much extra satisfaction as the last dollar spent on Y. If one side is bigger, you're not done reallocating yet — shift spending toward whichever good gives more utility per dollar.
Enter the marginal utility and price of each good at your current bundle to see whether you're at the optimum, and if not, which way to reallocate.
A price-taking firm treats price as fixed and just picks quantity. A monopolist has to accept that selling more means pushing price down on every unit — which is exactly why its marginal revenue curve falls below its demand curve, and why it never produces the competitive quantity.
Not a price taker; the only producer of the good; faces the entire (downward-sloping) market demand curve itself.
Demand Q(P) asks "at this price, how many can I sell?" Inverse demand P(Q) asks "to sell this many, what's the highest price I can charge?" Profit-maximization is naturally solved using the inverse form.
Marginal revenue is the new unit's price, minus the revenue given up because that new unit forces the price down on every unit already being sold — the second term is always negative for a downward-sloping demand curve, which is why MR sits below the demand (price) curve.
The markup rule: since εD is negative, this pushes price above marginal cost, and by more the less elastic (closer to zero) demand is. Highly elastic demand (large negative ε) means the markup shrinks toward zero — the monopolist starts behaving like a price taker.
Enter a linear inverse demand curve, P(Q) = a − bQ, and a constant marginal cost. Solves MR = MC directly.
A firm's short-run supply decision reduces to a single rule — produce where price equals marginal cost — but getting there requires keeping several different cost concepts straight, since only some of them belong in the decision at all.
Explicit costs (actual cash paid out) plus opportunity costs (the return given up on the next-best use of your resources) — the true cost, which is larger than what shows up on a P&L.
Costs that can't be recovered regardless of what you do next. The defining rule: sunk costs should never affect a forward-looking decision.
Fixed (F): the cost of capital, unchangeable in the short run. Variable (VC(q)): labor, supplies — adjustable in the short run and the only thing that scales with output.
Short run: produce where P = MC(q), but only if P ≥ AVC (otherwise shut down — fixed costs are sunk and irrelevant). Long run: produce where P = MC, but only if P ≥ ATC (otherwise exit — every cost is avoidable).
AFC always shrinks toward zero as q grows (fixed costs get spread over more units) — that's the entire reason ATC and AVC converge at high output. MC crosses both AVC and ATC exactly at their minimum points, since pulling the average down requires the marginal unit to cost less than the average, and pushing it up requires the opposite.
Model variable cost as VC(q) = v₁q + v₂q² (covers constant or rising marginal cost). Enter a market price to find the profit-maximizing quantity and check the shutdown condition.
Show every combination of labor (L) and capital (K) that produces the same output q = F(L,K).
C = wL + rK — every combination of L and K costing the same total. Slope = −r/w.
Marginal Rate of Technical Substitution: — the production-side analog of MRS. Cost-minimizing input choice requires MRTS = r/w, i.e. .
The production-side version of the bang-for-the-buck rule: at the cost-minimizing input mix, a dollar spent on labor buys exactly as much output as a dollar spent on capital.
With only a few firms, each one's best move depends on what the others do — the defining feature of oligopoly, and the reason the same market structure produces very different outcomes depending on whether firms compete on quantity, price, or timing.
Every firm is making its best choice, given what every other firm is doing — nobody can unilaterally improve by deviating.
Firms simultaneously choose quantity; price is whatever clears the market given total output. Cournot output sits between the monopoly and perfectly competitive outcomes.
Sequential version of Cournot: one firm commits to a quantity first, the other reacts. The first mover produces more and earns more than the follower.
Firms compete on price instead. With identical products, the low-price firm takes the whole market — driving price down to marginal cost with just two firms, a much more competitive outcome than Cournot.
Firms cooperate to jointly maximize profit as if they were a single monopolist, then split the proceeds — requires limited entry and some way to police the agreement (cheating is always individually tempting).
Linear inverse demand P = a − bQ, identical constant marginal cost c, and N identical firms choosing quantity simultaneously.
Try increasing N: quantity rises, price falls toward marginal cost, and profit per firm shrinks — Cournot converges to the competitive outcome as N grows.
Same demand and cost setup, but Firm 1 (the leader) commits to its quantity first, and Firm 2 (the follower) best-responds.
People maximize expected utility, not expected income — which is the whole reason insurance markets can exist: a risk-averse person will happily pay more than the expected payout to make an uncertain loss certain.
A risk-averse person prefers a guaranteed amount over a risky bet with the same expected value — they'd rather have $50 for certain than a coin flip between $0 and $100.
The extra expected income someone must receive to be just as well off facing risk as they would be with a guaranteed (certain) amount.
Diversification: hold bets that do well precisely when your other bets do poorly. Insurance: a contract that pays out when something bad happens.
People often know more about their own risk than the insurer does — riskier people disproportionately buy insurance, driving up costs for everyone, while the least-risky opt out.
Being insured changes your behavior for the worse, since you no longer bear the full cost of a bad outcome.
Expected value is the probability-weighted average payoff — what you'd earn on average over many repetitions. Variance measures how spread out the actual outcomes are around that average, which is the "risk" a risk-averse person cares about beyond the average itself.
Add a row for every possible outcome, with its probability and payoff. Probabilities should sum to 1.
Charging everyone the same price leaves money on the table whenever willingness to pay varies — price discrimination is the general strategy of capturing more of that variation, and it requires market power plus some way to stop resale.
Charge each unit at exactly that consumer's willingness to pay — captures the maximum possible profit. In practice approximated by a subscription fee plus marginal-cost pricing, or selling one unit at a time.
Offer product variations (different bundles, quantities, or quality tiers) and let consumers self-select — no need to observe anyone's type directly.
Group consumers by an observable characteristic (student, senior, region) and charge each group a different price.
Market power (can't be a competitive market) and no resale — otherwise low-price buyers would just resell to high-price ones and undo the discrimination.
Selling two or more products together works best when values for the products are negatively correlated across consumers — it smooths out the variation in what any one bundle-buyer is willing to pay.
Solving MR = MC separately in each market (using the same inverse-demand-to-marginal-revenue logic as the monopoly section) gives a clean takeaway: price is set lower in the more elastic market, and higher in the less elastic one — each market's markup follows the same P = MC/(1+1/ε) rule from before, applied market by market.
Enter linear demand in each market (Qi = ai − biP) and a shared marginal cost. Each market is solved independently at MR = MC.