Session 3 · Demand and Elasticity
Session 2 ended holding a function: the Marshallian demand \(x^{*}(p_x, p_y, M)\), solved out of one consumer’s problem.
One consumer is not a market. Today we add them up, and then we measure how sharply the total responds. 📐
Elasticity is that measure. For anyone coming from mathematics, it is a normalized derivative.
We finish with the tool that session 4 needs to talk about market power. 🔧
Fix \(p_y\) and \(M\), let \(p_x\) vary, and last session’s solution traces a curve.
For Cobb-Douglas, \(x^{*} = \dfrac{\alpha M}{p_x}\), so \(\dfrac{\partial x^{*}}{\partial p_x} < 0\). Downward sloping, and derived, not assumed.
Economists plot it with \(p_x\) on the vertical axis, which is the inverse function. A historical accident you simply have to live with. 🙃
At each price, ask every consumer how much they want and add the quantities.
\[Q(p) = \sum_{i=1}^{n} x_i^{*}(p, \cdot)\]
It is a sum across quantities at a given price, so on a diagram you add horizontally, not vertically.
Each individual demand slopes down, so the sum does too. Market demand inherits the property from the consumer’s problem. 🔑
Now a second route to the same curve, one that makes surplus visible. Ana is thinking about buying a concert ticket.
Value \(v\): what the ticket is worth to Ana, measured in euros. It is a property of Ana and the good, not of the market.
Ana’s value is one number. It does not move when the price moves. What the price decides is only whether buying is worth it. 🎯
Ana buys when her surplus is not negative:
\[\text{surplus} = v - p \;\ge\; 0\]
So she buys for any \(p \le v\), and the highest price at which she still buys is exactly \(v\).
Reservation price of a buyer: the highest price at which they are still willing to buy. It equals their value \(v\).
Value is what the good gives you. Price is what the market asks. The reservation price is where the two just meet. ⚖️
Ana values it at 8 euros, Bruno at 6, Carla at 4, Diogo at 2.
At a price of 5, whoever values it above 5 buys: Ana and Bruno. Lower the price and Carla joins, then Diogo.
Sort the reservation prices from highest to lowest and you have just drawn the demand curve again, from the other direction.
Each step down is one more buyer coming in. With many buyers the steps shrink and the staircase becomes a downward sloping line.
Demand is, at bottom, the sorted list of willingness to pay. Nothing more mysterious than that. 📋
This matters more than it looks. Take \(P = 10 - Q\).
Horizontally, at a given price: how much is bought? \(Q = 10 - P\). This is the quantity demanded.
Vertically, at a given quantity: what is the value of the last unit sold? \(P = 10 - Q\). This is the marginal willingness to pay.
The vertical reading is the one that gives us surplus, and in a moment elasticity. Keep it. 🔑
Consumer surplus (CS): what buyers value above what they pay, summed over every buyer who trades.
For one buyer, \(v - p\). For the market, the area below the demand curve and above the price.
Ana, who valued it at 8, pays 5 and walks away 3 euros better off. Diogo, who valued it at 2, does not trade and gets nothing. 🎟️
The vertical reading gives the value of the last unit, \(P(q)\). Sum value over units, subtract what was paid:
\[\text{CS} = \int_{0}^{Q} P(q)\,dq \;-\; p\,Q\]
With \(P = 10 - q\) and \(p = 5\), so \(Q = 5\): \(\int_0^5 (10 - q)\,dq - 25 = 37.5 - 25 = 12.5\). The triangle, without the geometry. ✅
Do not read too much into the number. Adding euros across people is a choice, not a theorem, and session 2 already warned you that utility itself is not comparable. ⚠️
Two multiple choice questions and one exercise. ✍️
A buyer’s reservation price:
A. Falls whenever the market price falls.
B. Is the price the buyer would most like to pay.
C. Equals the value of the good to that buyer.
D. Is the same for every buyer in the market.
✅ C. The value is one number and does not move with the price. The reservation price is that value read as a price: the highest price at which buying is still worth it. What the buyer would like to pay is zero.
Market demand is obtained from individual demands by:
A. Adding quantities at each given price.
B. Adding prices at each given quantity.
C. Averaging the individual demand curves.
D. Taking the demand of the consumer with the highest income.
✅ A. At a price, each consumer names a quantity, and the market quantity is their sum. That is why the aggregation is horizontal. Adding prices vertically is a different operation entirely, and it belongs to public goods.
Two consumers. Ana’s demand is \(q_A = 12 - 2p\) and Bruno’s is \(q_B = 8 - p\), both for \(p\) low enough that the quantity is positive.
a) Find market demand for \(p \le 4\).
b) Find market demand for \(4 < p \le 6\), and explain the kink.
c) At \(p = 3\), compute consumer surplus for Ana.
d) At \(p = 3\), compute total consumer surplus.
a) Both buy, so \(Q = (12 - 2p) + (8 - p) = 20 - 3p\).
b) Bruno drops out at \(p = 8\), Ana at \(p = 6\). For \(4 < p \le 6\) both are still in, so \(Q = 20 - 3p\) still. The kink is at \(p = 6\), above which only Bruno remains: \(Q = 8 - p\).
c) Ana’s inverse demand is \(p = 6 - q/2\), choking at 6. At \(p = 3\) she buys 6, so \(\text{CS}_A = \tfrac{1}{2}\times 6 \times (6 - 3) = 9\) euros.
d) Bruno buys 5, and his choke price is 8: \(\text{CS}_B = \tfrac{1}{2}\times 5 \times (8 - 3) = 12.5\). Total \(= 21.5\) euros. ✅
Price elasticity of demand: the percentage change in quantity over the percentage change in price.
\[ \varepsilon = \frac{dQ}{dP}\cdot\frac{P}{Q} \]
It is the derivative \(dQ/dP\) normalized by \(P/Q\), so that it does not depend on units. The sign is negative; we work with \(|\varepsilon|\).
Equivalently \(\varepsilon = d\ln Q / d\ln P\). That form is worth remembering: it is why elasticities are estimated in logs. 📊
\(|\varepsilon| > 1\): elastic. Quantity responds more than proportionally.
\(|\varepsilon| < 1\): inelastic. Quantity responds less than proportionally.
\(|\varepsilon| = 1\): unit elastic.
📏 Perfectly inelastic (\(\varepsilon = 0\)): quantity is fixed, the curve is vertical. Example: a life-saving drug.
➖ Perfectly elastic (\(|\varepsilon| \to \infty\)): at any price above some value, demand vanishes. The curve is horizontal.
The second one is not a curiosity. It is exactly what a single firm faces under perfect competition, which is why that firm is a price taker in session 4. 🔁
Take \(Q = a - bP\). Then \(dQ/dP = -b\) and
\[ \varepsilon = -b\cdot\frac{P}{Q} = -b\cdot\frac{P}{a - bP}. \]
At the top of the curve (high P) it is elastic; at the bottom (low P) it is inelastic; in the middle it is unit elastic.
So “this good is elastic” is a statement about a point, not about a curve. The slope is constant; the elasticity is not. ⚠️
Between two observed points, the percentage change depends on which one you start from. The midpoint (arc) formula avoids that:
\[\varepsilon_{arc} = \frac{\Delta Q}{(Q_1 + Q_2)/2} \Big/ \frac{\Delta P}{(P_1 + P_2)/2}\]
It is a discrete approximation to the same object. We use the point definition throughout, because you already have calculus and the arc formula is just what you do when you only have two data rows. 📐
🔀 Substitutes. The more of them, the more elastic. This is the dominant factor.
👛 Share of the budget. Salt is inelastic partly because nobody notices the price.
⏳ Time. Given a year, people switch heating systems; given a day, they do not. Long run elasticities are always larger.
🎯 How the market is defined. Demand for “coffee” is inelastic; for “coffee from this one brand”, very elastic. The narrower the definition, the more elastic.
One functional form is worth memorizing:
\[Q = A\,P^{-\eta} \quad \Longrightarrow \quad \varepsilon = \frac{dQ}{dP}\frac{P}{Q} = -\eta \quad \text{(constant)}\]
Take logs: \(\ln Q = \ln A - \eta \ln P\). The elasticity is the slope of a log-log regression.
Which is exactly how elasticities are estimated in practice, and why you will see demand models written in logs. 📊
The same idea on the sellers’ side:
Price elasticity of supply: the percentage change in quantity supplied over the percentage change in price.
It depends above all on time: in the short run it is more inelastic, in the long run more elastic. ⏳
The reason is capacity. Overnight a firm can only run its existing plant harder; over years it can build another one. We derive that curve properly next session. 🏭
Two multiple choice questions and one exercise. ✍️
Along a linear demand \(Q = a - bP\), elasticity in absolute value:
A. Is constant along the curve, and equal to \(b\).
B. Rises as the price falls.
C. Is undefined except at the midpoint.
D. Falls as the price falls.
✅ D. \(|\varepsilon| = bP/(a - bP)\) is increasing in \(P\), so lower prices sit in the inelastic region. Option A confuses the slope, which is constant, with the elasticity, which is not.
Demand is \(Q = A P^{-2}\). A 1 percent rise in the price changes quantity by roughly:
A. 2 percent up.
B. 2 percent down.
C. 0.5 percent down.
D. It depends on the price you start from.
✅ B. Constant elasticity \(\varepsilon = -2\), so quantity falls by about twice the percentage rise in the price, from any starting point. Option D would be right for a linear demand and is wrong here: that is the whole point of the constant elasticity form.
Demand \(Q = 100 - 4P\), with \(P\) in euros.
a) Find \(Q\) at \(P = 10\), and the elasticity there.
b) Say whether demand is elastic or inelastic at that point.
c) At which price is demand unit elastic?
d) At which price is the choke point, where quantity hits zero?
a) \(Q = 100 - 40 = 60\), and \(\varepsilon = \dfrac{dQ}{dP}\dfrac{P}{Q} = -4 \times \dfrac{10}{60} \approx -0.67\).
b) \(|\varepsilon| = 0.67 < 1\): inelastic at that point.
c) \(|\varepsilon| = 1\) requires \(4P = 100 - 4P\), so \(P = 12.5\) euros and \(Q = 50\). Note it is the midpoint of a linear demand, always.
d) \(Q = 0\) at \(P = 25\) euros, where \(|\varepsilon| \to \infty\). At the other end, \(P = 0\), elasticity is zero. The whole range is travelled along one straight line. ✅
Revenue is \(R = P\cdot Q\). Differentiating with respect to the price,
\[ \frac{dR}{dP} = Q + P\frac{dQ}{dP} = Q\,(1 + \varepsilon). \]
If \(|\varepsilon| < 1\): raising the price increases revenue. If \(|\varepsilon| > 1\): it reduces it. Revenue is maximized when \(|\varepsilon| = 1\). 🎯
This is the first genuinely useful thing elasticity buys you: it tells a firm which way to move the price. 💶
Now differentiate with respect to the quantity instead:
\[ MR = \frac{dR}{dQ} = P + Q\frac{dP}{dQ} = P\left(1 + \frac{1}{\varepsilon}\right). \]
Two readings. Selling one more unit brings in the price, minus what you gave up by cutting the price on everything else.
And \(MR < P\) whenever demand slopes down. Remember this formula: it is the centre of market power next session. 🔁
How does quantity change when income \(M\) changes?
\[ \varepsilon_M = \frac{dQ}{dM}\cdot\frac{M}{Q}. \]
\(\varepsilon_M > 0\): a normal good. \(\varepsilon_M < 0\): an inferior good. \(\varepsilon_M > 1\): a luxury good. 💎
How does the quantity of one good change when the price of another good \(Y\) changes?
\[ \varepsilon_{XY} = \frac{dQ_X}{dP_Y}\cdot\frac{P_Y}{Q_X}. \]
\(\varepsilon_{XY} > 0\): substitutes (tea and coffee). \(\varepsilon_{XY} < 0\): complements (a car and fuel). ⛽
Session 2 solved Cobb-Douglas: \(x^{*} = \dfrac{\alpha M}{p_x}\). Read the three elasticities straight off it.
\(\varepsilon = -1\), exactly unit elastic everywhere. \(\varepsilon_M = 1\), so normal and never a luxury. \(\varepsilon_{XY} = 0\), so neither substitute nor complement.
That is three restrictions, not three findings. Cobb-Douglas is convenient and narrow, and a demand system estimated on real data will reject all three. 🧠
Which is the honest reason to know where a functional form comes from: it decides your answers before you see the data. ⚠️
For anyone heading into finance, elasticity is sensitivity analysis under another name.
How much does demand for the firm’s product change if a competitor cuts its price by 5 percent? That is cross-price elasticity.
How far does demand fall in a recession? That is income elasticity. The same formulas, applied to revenue models. 💼
Two multiple choice questions and one exercise. ✍️
A firm currently sells where demand is inelastic. To raise revenue it should:
A. Cut the price, because quantity will respond strongly.
B. Raise the price, because quantity falls less than proportionally.
C. Leave the price alone, since revenue is already at a maximum.
D. Cut the price, because revenue always rises when the price falls.
✅ B. With \(|\varepsilon| < 1\) we have \(dR/dP = Q(1 + \varepsilon) > 0\). Revenue keeps rising with the price until the unit elastic point, where \(dR/dP = 0\).
The cross-price elasticity between two goods is negative. The goods are:
A. Complements.
B. Substitutes.
C. Independent of each other.
D. Both inferior goods.
✅ A. The price of one rises and demand for the other falls, so they are consumed together: complements. Note that income elasticity, not cross-price elasticity, is what classifies a good as inferior.
Demand for a product is estimated as \(\ln Q = 6 - 1.5 \ln P + 0.8 \ln M + 0.4 \ln P_Y\).
a) State the three elasticities.
b) Is the good normal, inferior or a luxury?
c) Is \(Y\) a substitute or a complement?
d) The firm raises its price by 4 percent. What happens to revenue, roughly?
a) In a log-log form the coefficients are the elasticities: \(\varepsilon = -1.5\), \(\varepsilon_M = 0.8\), \(\varepsilon_{XY} = 0.4\).
b) \(0 < \varepsilon_M < 1\): normal, but not a luxury. Demand grows with income, more slowly than income does.
c) \(\varepsilon_{XY} > 0\): a substitute. Y becoming dearer sends buyers this way.
d) \(|\varepsilon| = 1.5 > 1\), so revenue falls. Quantity drops about \(1.5 \times 4 = 6\) percent against a 4 percent price rise, for a net change near \(-2\) percent. ✅
➕ Market demand is individual demands added horizontally. It slopes down because each consumer’s does.
💙 Consumer surplus is value above price: the area under demand and above the price, or an integral if you prefer.
📐 Elasticity is a normalized derivative, \(d\ln Q/d\ln P\). Along a linear demand it runs from elastic at the top to inelastic at the bottom.
💰 \(dR/dP = Q(1 + \varepsilon)\) tells a firm which way to move the price, and \(MR = P(1 + 1/\varepsilon)\) is the same fact on the quantity side.
👛 Income and cross-price elasticities classify goods: normal or inferior, substitute or complement.
Production, Costs and Market Structures.
We have the buyers. Next we build the seller: where costs come from, why supply is marginal cost, and what changes when one firm faces the whole demand curve. 🏭
See you next week. 🙌