Chapter 2 gave us the supply-and-demand model: curves, equilibrium, shifts, and interventions. But that model tells us only the direction of price and quantity changes, not their magnitudes. When demand increases, how much does price rise? When the government imposes a tax, who actually bears the burden — buyers or sellers? To answer these questions, we need a measure of responsiveness: elasticity.
This chapter also introduces the welfare framework — consumer surplus, producer surplus, and deadweight loss — that allows us to evaluate whether a market outcome is efficient and measure the cost of interventions. Together, elasticity and surplus analysis give us the tools to make quantitative judgments about markets and policies, not just qualitative ones.
Saying "quantity demanded falls when price rises" is qualitative. A business owner needs to know: by how much? If I raise my price by 10%, will I lose 5% of my customers or 50%? The answer determines whether the price increase is profitable or ruinous. Elasticity provides the answer.
What this says: Elasticity measures how sensitive buyers are to price changes, in percentage terms. If elasticity is -2, a 1% price increase causes a 2% drop in quantity demanded.
Why it matters: Unlike slope, elasticity is unit-free — you can compare the price sensitivity of coffee, cars, and concert tickets on the same scale. It answers the business question: "If I raise my price, will I lose a lot of customers or just a few?"
See Full Mode for the derivation.By the law of demand, $\varepsilon_d$ is typically negative (quantity moves opposite to price). Convention varies — some texts take the absolute value. We keep the negative sign and use $|\varepsilon_d|$ when comparing magnitudes.
Why use percentages? Because they make elasticity unit-free and comparable across goods. A \$1 price increase means very different things for a \$1 cup of coffee and a \$10,000 car. But a 10% price increase is a meaningful comparison regardless of the unit.
| $|\varepsilon_d|$ | Term | Meaning | Example |
|---|---|---|---|
| $> 1$ | Elastic | Quantity responds more than proportionally | Restaurant meals, vacation travel |
| $= 1$ | Unit elastic | Quantity responds proportionally | The revenue-maximizing point |
| $< 1$ | Inelastic | Quantity responds less than proportionally | Gasoline (short run), insulin |
| $= 0$ | Perfectly inelastic | Quantity does not respond (vertical curve) | Life-saving medication with no substitute |
| $= \infty$ | Perfectly elastic | Any price increase kills demand (horizontal curve) | Wheat from one farmer in a competitive market |
For a continuous demand function $Q_d = a - bP$, the derivative $dQ_d/dP = -b$, so:
What this says: For a straight-line demand curve, elasticity depends on where you are on the curve. Even though the slope is constant, the ratio P/Q changes — so elasticity varies from point to point.
Why it matters: This is why "steep = inelastic" is wrong. At the top of a linear demand curve (high price, low quantity), demand is elastic; at the bottom (low price, high quantity), it is inelastic. The midpoint is unit elastic.
See Full Mode for the derivation.Notice something important: even though the slope $-b$ is constant along a linear demand curve, the elasticity is not constant. It depends on the ratio $P/Q$, which changes as you move along the curve. At high prices (where $P$ is large and $Q$ is small), $P/Q$ is large, making $|\varepsilon_d|$ large — demand is elastic. At low prices (where $P$ is small and $Q$ is large), $P/Q$ is small, making $|\varepsilon_d|$ small — demand is inelastic. At the midpoint of the demand curve, $|\varepsilon_d| = 1$.
This is a subtlety that trips up many students: a steep demand curve is not the same as an inelastic one, and a flat curve is not the same as an elastic one. Slope and elasticity are different concepts. Slope ($\Delta Q/\Delta P$) uses absolute changes; elasticity uses percentage changes.
Figure 3.1. Elasticity varies along a linear demand curve even though the slope is constant. The upper portion is elastic ($|\varepsilon_d| > 1$), the midpoint is unit elastic ($|\varepsilon_d| = 1$), and the lower portion is inelastic ($|\varepsilon_d| < 1$). Hover over any point on the curve to see the exact elasticity.
When we don't have a continuous function but only two discrete data points $(P_1, Q_1)$ and $(P_2, Q_2)$, computing elasticity faces an asymmetry problem: using $(P_1, Q_1)$ as the base gives a different answer than using $(P_2, Q_2)$. The midpoint (arc) method resolves this by using the average of the two points as the base:
What this says: When you only have two data points (not a smooth curve), the midpoint method uses the average of the two prices and quantities as the base. This gives the same answer regardless of which direction you measure the change.
Why it matters: Without the midpoint method, going from point A to B gives a different elasticity than going from B to A. The arc formula eliminates this asymmetry, making it the standard approach for real-world data with discrete observations.
See Full Mode for the derivation.The arc elasticity gives the same answer regardless of which direction you compute the change — from point 1 to point 2 or from point 2 to point 1.
Using $Q_d = 100 - 20P$:
Point elasticity at $P = 3$, $Q = 40$:
$\varepsilon_d = -20 \cdot \frac{3}{40} = -1.5$ — elastic. A 1% price increase would reduce quantity demanded by 1.5%.
Point elasticity at $P = 1$, $Q = 80$:
$\varepsilon_d = -20 \cdot \frac{1}{80} = -0.25$ — inelastic. A 1% price increase would reduce quantity by only 0.25%.
Arc elasticity between $(P_1 = 2, Q_1 = 60)$ and $(P_2 = 3, Q_2 = 40)$:
$\varepsilon_d^{arc} = \frac{40 - 60}{3 - 2} \cdot \frac{2 + 3}{60 + 40} = \frac{-20}{1} \cdot \frac{5}{100} = -1.0$ — unit elastic over this range.
What makes demand for some goods elastic and others inelastic? Five factors matter:
1. Availability of close substitutes. This is the most important determinant. If many alternatives exist, consumers easily switch when the price rises — demand is elastic. If few or no substitutes exist, consumers are stuck — demand is inelastic.
The key insight: elasticity depends on how narrowly you define the market. Demand for "beverages" is very inelastic. Demand for "coffee" is somewhat inelastic. Demand for "Starbucks coffee" is quite elastic. Demand for "a tall latte at the Starbucks on 5th and Main" is extremely elastic.
2. Necessities vs. luxuries. Necessities — insulin for diabetics, basic food staples, heating fuel in winter — have inelastic demand. Luxuries — vacation travel, fine dining, designer clothing — have elastic demand.
3. Time horizon. Demand is more elastic in the long run than the short run. Short-run gasoline demand is very inelastic ($|\varepsilon_d| \approx 0.2$); long-run demand is more elastic ($|\varepsilon_d| \approx 0.7$).
4. Share of budget. Goods that account for a large share of the consumer's budget have more elastic demand.
5. How broadly or narrowly the market is defined. Narrower markets have more elastic demand. "Food" is inelastic. "Organic heirloom tomatoes from the farmers' market" is very elastic.
The elasticity concept extends beyond own-price demand.
| $\varepsilon_I$ | Classification | Examples |
|---|---|---|
| $> 1$ | Luxury (income-elastic normal good) | Organic food, international travel, private education |
| \$1 < \varepsilon_I < 1$ | Necessity (income-inelastic normal good) | Basic groceries, utilities, clothing staples |
| $< 0$ | Inferior good | Instant noodles, bus tickets, generic store brands |
As income rises, the budget share of necessities falls (Engel's law) and the share of luxuries rises.
$\varepsilon_{xy} > 0$: goods are substitutes. $\varepsilon_{xy} < 0$: goods are complements. $\varepsilon_{xy} = 0$: goods are unrelated.
Cross-price elasticities matter enormously in antitrust economics. Regulators use them to define markets: if two products have high cross-price elasticity (strong substitutes), they are in the same market.
Supply elasticity is typically positive. It depends on spare capacity, input availability, and the time horizon.
Total revenue is $TR = P \times Q$. When price changes, two forces work in opposite directions: a higher price means more revenue per unit (price effect), but fewer units sold (quantity effect). Which force wins depends on elasticity.
Taking the derivative:
What this says: When you raise the price, two things happen: you earn more per unit sold (price effect), but you sell fewer units (quantity effect). Whether total revenue goes up or down depends on which effect is stronger — and that is exactly what elasticity measures.
Why it matters: If demand is elastic, the quantity drop dominates and a price increase hurts revenue. If demand is inelastic, the higher price per unit dominates and revenue rises. Revenue is maximized where elasticity equals -1 (unit elastic).
See Full Mode for the derivation.Since $\varepsilon_d < 0$, the sign of $dTR/dP$ depends on whether $|\varepsilon_d|$ is greater or less than 1:
| If demand is... | $|\varepsilon_d|$ | Price rise → TR... | Price fall → TR... |
|---|---|---|---|
| Elastic | $> 1$ | Falls (quantity effect dominates) | Rises |
| Unit elastic | $= 1$ | Unchanged | Unchanged |
| Inelastic | $< 1$ | Rises (price effect dominates) | Falls |
Using $Q_d = 100 - 20P$: $TR = P(100 - 20P) = 100P - 20P^2$.
To find the maximum: $dTR/dP = 100 - 40P = 0 \implies P = 2.50$.
At $P = 2.50$: $Q = 50$, $TR_{max} = 125$. Elasticity: $\varepsilon_d = -20 \times (2.50/50) = -1.0$. Unit elastic — revenue is maximized where $|\varepsilon_d| = 1$.
Figure 3.2. Move the price slider. Left: the demand curve with the current price highlighted. Right: the total revenue curve — an inverted parabola peaking at $P = 2.50$ where demand is unit elastic.
Elasticity tells us how much quantities respond to prices. Surplus analysis tells us how much benefit buyers and sellers receive from market transactions — and how much is lost when markets are distorted.
What this says: Consumer surplus is the total "bonus" buyers get from paying less than they were willing to pay. Graphically, it is the triangle between the demand curve and the market price line.
Why it matters: It measures the net benefit buyers receive from participating in the market. When prices fall, consumer surplus grows — buyers capture more value.
See Full Mode for the derivation.What this says: Producer surplus is the total "bonus" sellers get from receiving more than the minimum price at which they would have been willing to sell. Graphically, it is the triangle between the market price line and the supply curve.
Why it matters: It measures the net benefit sellers receive from participating in the market. When prices rise, producer surplus grows — sellers capture more value.
See Full Mode for the derivation.A fundamental result: total surplus is maximized at the competitive equilibrium quantity. Any deviation from $Q^*$ — whether from taxes, price controls, monopoly, or quotas — reduces total surplus. The lost surplus is called deadweight loss.
Using $Q_d = 100 - 20P$ and $Q_s = 20P - 10$. Equilibrium: $P^* = 2.75$, $Q^* = 45$.
$CS = \frac{1}{2}(5.00 - 2.75)(45) = 50.63$
$PS = \frac{1}{2}(2.75 - 0.50)(45) = 50.63$
$TS = 50.63 + 50.63 = 101.25$
Figure 3.3. Drag the price away from equilibrium (\$1.75) to see how CS and PS change. A deadweight loss triangle appears whenever the price deviates from the equilibrium — these are mutually beneficial trades that no longer occur.
A question that surprises most people: when the government imposes a tax on sellers, do sellers actually bear the burden? The answer: not necessarily. Tax incidence — who truly pays — depends on the relative elasticities of supply and demand, not on who legally remits the tax.
A per-unit tax of $t$ imposed on sellers drives a wedge between the price buyers pay ($P_B$) and the price sellers receive ($P_S$): $P_B = P_S + t$.
What this says: A per-unit tax creates a gap (wedge) between the price buyers pay and the price sellers receive. The market still clears, but at a new, lower quantity where buyers' willingness to pay at the higher price matches sellers' willingness to sell at the lower price.
Why it matters: The tax drives a wedge between buyer and seller prices, reducing the number of transactions. Some trades that would have been mutually beneficial no longer happen.
See Full Mode for the derivation.What this says: The side of the market that is more inelastic (less able to adjust) bears more of the tax burden. If demand is very inelastic and supply is elastic, buyers bear most of the tax — and vice versa.
Why it matters: It does not matter whether the law says "sellers pay the tax" or "buyers pay the tax." The economic burden depends entirely on who has fewer alternatives. Taxing insulin sellers still hits patients, because patients cannot stop buying insulin.
See Full Mode for the derivation.The rule: the more inelastic side bears more of the tax. The party with fewer alternatives cannot easily escape the tax by adjusting behavior. They are "stuck" — and the tax burden falls on them.
A $t = 0.50$ per-cup tax on lemonade sellers (with $Q_d = 100 - 20P$, $Q_s = 20P - 10$):
$P_B = 2.75 + 0.5(0.50) = 3.00$ | $P_S = 2.75 - 0.5(0.50) = 2.50$
$Q_{new} = 100 - 20(3.00) = 40$
Buyers bear \$1.25 of the \$1.50 tax (50%). Sellers bear the other \$1.25 (50%). The even split occurs because $b = d = 20$ — equal absolute slopes.
Figure 3.4. A fixed \$1.00 tax. Change the demand slope to see the burden shift: steeper (more inelastic) demand means buyers bear more of the tax because they cannot easily reduce consumption. Flatter (more elastic) demand means sellers bear more.
You just learned tax incidence: who actually bears a tax depends on elasticities, not on who writes the check. The surplus framework measures total welfare — but it is silent on how the pie is distributed.
The surplus framework tells you the size of the pie and the cost of policies that shrink it. Tax incidence shows that the economic burden of a tax falls on the less elastic side of the market, regardless of the legal assignment. A payroll tax nominally paid by employers is largely borne by workers if labor supply is inelastic. Consumer surplus plus producer surplus measures total welfare — and total surplus is maximized at the competitive equilibrium. This is the efficiency benchmark: any deviation (taxes, price controls, quotas) reduces the pie.
But "maximize the pie" quietly assumes the slicing doesn't matter. Total surplus treats a dollar to a billionaire identically to a dollar to someone in poverty. That violates most people's moral intuitions — and it's not a minor aesthetic objection. If the marginal utility of income is declining (a reasonable assumption backed by extensive evidence), then a dollar transferred from a rich person to a poor person increases aggregate well-being even if total surplus stays the same. The efficiency framework can't see this. Worse, the clean separation of efficiency and equity — "maximize the pie, then redistribute" — is practically impossible. Every real redistribution tool (income taxes, transfers, minimum wages) also changes incentives and shrinks the pie. You cannot slice without affecting size.
Welfare economics tried to address this through social welfare functions — ways of aggregating individual utilities that encode values about distribution. A utilitarian SWF sums total utility (favoring some redistribution due to diminishing marginal utility). A Rawlsian SWF maximizes the welfare of the worst-off (favoring extensive redistribution). But the choice of SWF is a normative judgment — economics can formalize the tradeoffs, but it cannot tell you which values are correct.
The efficiency framework is necessary but not sufficient for thinking about inequality. It tells you the cost of redistribution — every tax creates deadweight loss, every price control distorts quantities — but it cannot tell you whether that cost is worth paying. That is a moral and political question that economics can inform but not resolve. Be skeptical of anyone who uses "efficiency" as a conversation-stopper. Efficiency is a tool for measuring costs, not a philosophy for deciding what matters.
How large is the efficiency cost of redistribution in practice? The answer depends on behavioral elasticities that you don't have tools to estimate yet. Come back at Chapter 4 (§4.1, §4.4) where externalities and public goods provide efficiency-based arguments for some redistribution. And at Chapter 16 (§16.7), optimal tax theory gives precise, quantitative answers: the Ramsey rule and Mirrlees framework tell you exactly how much efficiency you sacrifice for a given redistribution target.
Dan Riffle, then a top aide to Alexandria Ocasio-Cortez, popularized the slogan in 2019. It went viral — bumper stickers, protest signs, Twitter bios. The claim: the existence of any billionaire proves the system is rigged. But does the welfare theorem actually say that?
AdvancedElizabeth Warren proposed a 2% annual tax on wealth above \$10 million. Wealth concentration has returned to Gilded Age levels — but is a wealth tax the right mechanism, or would it do more damage than the inequality it targets?
AdvancedDWL is not a transfer from one group to another. Tax revenue is a transfer (from private parties to the government). But DWL is a net loss — it goes to nobody. It is the cost of inefficiency.
What this says: Deadweight loss is the area of the triangle formed by the tax wedge and the lost transactions. It equals half the tax times the reduction in quantity traded.
Why it matters: This is value destroyed, not transferred. Tax revenue goes to the government (a transfer), but deadweight loss goes to nobody. It represents trades that would have made both buyer and seller better off, but no longer happen because the tax makes them unprofitable.
See Full Mode for the derivation.where $\Delta Q = Q^*_{no\,tax} - Q^*_{tax}$ is the reduction in quantity caused by the tax.
From Example 3.4: $t = 0.50$, $\Delta Q = 45 - 40 = 5$.
$DWL = \frac{1}{2}(0.50)(5) = 1.25$
Verification: $TS_{original} = 101.25$. With tax: $CS = 40.00$, $PS = 40.00$, Revenue $= 20.00$, so $TS = 100.00$. The \$1.25 difference is the deadweight loss.
For linear supply and demand, $\Delta Q$ is proportional to $t$. Since $DWL = \frac{1}{2} t \cdot \Delta Q$ and $\Delta Q \propto t$:
What this says: Deadweight loss grows with the square of the tax rate. Double the tax, quadruple the waste.
Why it matters: This is one of the most important results in public finance. It means small taxes are relatively cheap in efficiency terms, but large taxes are devastating. The policy implication: it is far better to spread taxes thinly across many goods than to pile a heavy tax on a single good.
See Full Mode for the derivation.Doubling the tax rate quadruples the deadweight loss. This has a profound implication: it is more efficient to spread taxes across many goods at low rates than to concentrate them on a few goods at high rates.
Figure 3.5. Drag the tax slider from \$1 to \$1. Watch the DWL triangle (yellow) grow with the square of the tax rate. At $t = 1$, DWL = \$1.00. At $t = 2$, DWL = \$10.00 — four times as much. The purple rectangle is tax revenue, which eventually shrinks as high taxes destroy too many transactions.
DWL is larger when supply and demand are more elastic. Elastic markets are responsive — the tax eliminates many transactions. Inelastic markets are unresponsive — the tax barely changes behavior, so few transactions are lost.
This creates a tension: the most efficient taxes (smallest DWL) fall on goods with inelastic demand — but these are also the taxes where buyers bear the largest burden. Efficiency and equity can conflict.
Figure 3.6. The same tax applied to an elastic market (left, $b = 40$) and an inelastic market (right, $b = 5$). The elastic market loses far more transactions and has much larger DWL. Drag the tax slider to compare.
You just proved that total surplus is maximized at the competitive equilibrium — any tax or price control creates deadweight loss. The market looks like the gold standard. But look closely at the conditions required.
Total surplus — the sum of consumer surplus and producer surplus — is maximized when the market reaches competitive equilibrium. Every unit where the buyer's willingness to pay exceeds the seller's cost gets produced and traded. No central planner is needed: the price adjusts until quantity supplied equals quantity demanded, and at that point, every value-creating transaction occurs. A tax drives a wedge between the price buyers pay and the price sellers receive, preventing some mutually beneficial trades. The resulting deadweight loss triangle is a precise measure of efficiency lost. By this metric, the unfettered competitive market gets it exactly right.
But the result depends on conditions that are analytically convenient and empirically rare. Total surplus maximization requires no externalities (all costs and benefits are captured in market prices), no market power (all agents are price takers), complete information (buyers and sellers know quality and alternatives), and no public goods. These aren't minor caveats — they are the rule, not the exception. Pollution is an externality that the market ignores. Monopolists restrict output below the efficient level. Patients can't evaluate whether they need surgery. The "efficiency" of competitive equilibrium is a theorem about a world that rarely exists in full.
The mainstream treats this result as a benchmark, not a description of reality. "Markets are efficient unless there's a market failure" is the standard framing — and the next chapter catalogs the failures (externalities, public goods, information asymmetry, market power). The strength of the benchmark is that it tells you precisely where to look for problems: whenever one of the conditions fails, surplus is not maximized, and there is a potential case for intervention.
The surplus framework is the right tool for evaluating whether a specific market is efficient. The competitive equilibrium result is genuinely powerful — markets coordinate millions of decentralized decisions without any central authority, and they do it remarkably well in many settings. But "remarkably well" is not "perfectly," and the conditions for the optimality result are demanding. The reader should hold both truths simultaneously: markets are an extraordinary coordination mechanism, and they fail systematically whenever the textbook conditions don't hold.
How common are market failures? Are they rare exceptions to a generally efficient system, or are they pervasive enough to undermine the benchmark? Come back at Chapter 4 (§4.1–§4.6) for the systematic catalog of market failures. And at Chapter 11 (§11.6–§11.7), the formal welfare theorems give you the precise mathematical conditions under which the result holds — and prove just how demanding those conditions are.
Dan Riffle, then a top aide to Alexandria Ocasio-Cortez, popularized the slogan in 2019. It went viral — bumper stickers, protest signs, Twitter bios. The claim: the existence of any billionaire proves the system is rigged. But does the welfare theorem actually say that?
AdvancedBernie Sanders made healthcare a centerpiece of his 2016 campaign. Americans spend 17% of GDP on healthcare and get worse outcomes than countries that spend half as much. Arrow explained why in 1963.
IntroThe city council, looking for revenue, imposes a \$1.50 per-cup tax on lemonade vendors.
Recall from Chapter 2: $Q_d = 100 - 20P$, $Q_s = 20P - 10$, equilibrium at $P^* = 2.75$, $Q^* = 45$.
Before tax: Revenue = \$1.75 \times 45 = \\$123.75$/day. CS = \$10.63, PS = \$10.63, TS = \$101.25.
After tax ($t = 0.50$): Buyers pay \$1.00; Maya receives \$1.50; she sells 40 cups.
Maya's revenue: \$1.50 \times 40 = \\$100.00$/day (down from \$123.75).
CS = \$10.00 (fell by \$10.63). PS = \$10.00 (fell by \$10.63). Tax revenue = \$10.00. DWL = \$1.25.
Maya's daily revenue of \$100.00 is now below her opportunity cost of \$120/day from the bookstore job (Chapter 1). The tax pushed her from barely viable to clearly unprofitable. The five cups that go unsold each day represent transactions that would have created value for both buyer and seller. The \$1.25 of deadweight loss is the total value those five transactions would have created.
| Label | Equation | Description |
|---|---|---|
| Eq. 3.1 | $\varepsilon_d = (\Delta Q_d / \Delta P)(P/Q)$ | Price elasticity of demand |
| Eq. 3.2 | $\varepsilon_d = -b \cdot P/Q$ | Point elasticity for linear demand |
| Eq. 3.3 | $\varepsilon_d^{arc} = \frac{Q_2-Q_1}{P_2-P_1} \cdot \frac{P_1+P_2}{Q_1+Q_2}$ | Arc (midpoint) elasticity |
| Eq. 3.4 | $\varepsilon_I = (\Delta Q_d / \Delta I)(I/Q_d)$ | Income elasticity of demand |
| Eq. 3.5 | $\varepsilon_{xy} = (\Delta Q_x / \Delta P_y)(P_y/Q_x)$ | Cross-price elasticity |
| Eq. 3.6 | $\varepsilon_s = (\Delta Q_s / \Delta P)(P/Q_s)$ | Price elasticity of supply |
| Eq. 3.7 | $TR = P \times Q$ | Total revenue |
| Eq. 3.8 | $dTR/dP = Q(1 + \varepsilon_d)$ | TR response to price change |
| Eq. 3.9 | $CS = \int_0^{Q^*} D(Q)\,dQ - P^* Q^*$ | Consumer surplus (general) |
| Eq. 3.10 | $CS = \frac{1}{2}(P_{max} - P^*)Q^*$ | Consumer surplus (linear demand) |
| Eq. 3.11 | $PS = P^* Q^* - \int_0^{Q^*} S(Q)\,dQ$ | Producer surplus (general) |
| Eq. 3.12 | $PS = \frac{1}{2}(P^* - P_{min})Q^*$ | Producer surplus (linear supply) |
| Eq. 3.13 | $TS = CS + PS$ | Total surplus |
| Eq. 3.14 | $Q_d(P_B) = Q_s(P_B - t)$ | Tax equilibrium condition |
| Eq. 3.15 | Buyer's share $= \varepsilon_s / (\varepsilon_s + |\varepsilon_d|)$ | Tax incidence — buyers |
| Eq. 3.16 | Seller's share $= |\varepsilon_d| / (\varepsilon_s + |\varepsilon_d|)$ | Tax incidence — sellers |
| Eq. 3.17 | $DWL = \frac{1}{2} t \cdot \Delta Q$ | Deadweight loss from per-unit tax |
| Eq. 3.18 | $DWL \propto t^2$ | DWL grows with square of tax rate |