Week 4 · Dr. Boaz Zik

Theory of the Firm, Partial Equilibrium & General Equilibrium

Theory of the Firm, Partial Equilibrium & General Equilibrium

Part of: Micro-Economics II | Key concepts: Theory of the Firm, Cost Function, Isoquant, RTS, Partial Equilibrium, Consumer Surplus, Firm Surplus, Edgeworth Box, Pareto Efficiency, First Welfare Theorem, Walras' Law


What This Presentation Is About

This presentation builds the supply side of the market from first principles. It has three major sections:

  1. Theory of the Firm — how firms choose output to maximise profit, and how the cost function is derived from the firm's production technology (one input, then two inputs).
  2. Partial Equilibrium — how supply and demand interact in a single market, why competitive equilibria are efficient, and what government subsidies do to welfare.
  3. General Equilibrium — moving beyond a single market to analyse an exchange economy with multiple agents, using the Edgeworth box, the concept of Pareto efficiency, the contract curve, and the First Welfare Theorem.

Part 1 — Theory of the Firm

The Firm's Problem

Definition: Firm

A firm transforms inputs (labour and capital) into outputs with the objective of profit maximisation. Π(q)=R(q)−C(q)\Pi(q) = R(q) - C(q) where qq is the quantity produced, R(q)R(q) is revenue, and C(q)C(q) is cost.

The firm chooses q∗q^* to solve:

max⁡q  Π(q)=R(q)−C(q)\max_q \; \Pi(q) = R(q) - C(q)

We assume Π(q)\Pi(q) is concave, meaning Π′′(q)<0\Pi''(q) < 0 — the marginal profit Π′(q)\Pi'(q) is decreasing. This guarantees the FOC gives a maximum.

First-Order Condition: Differentiate and set to zero:

R′(q∗)=C′(q∗)R'(q^*) = C'(q^*)
Marginal Revenue=Marginal Cost\boxed{\text{Marginal Revenue} = \text{Marginal Cost}}
Why MR = MC?

If MR > MC, producing one more unit increases profit — so you should produce more. If MR < MC, producing one more unit costs more than it earns — so you should produce less. The optimum is where these are equal. The concavity of Π\Pi guarantees that to the left of q∗q^* we have MR > MC (keep going) and to the right MR < MC (stop).

Example

R(q)=100q−q2R(q) = 100q - q^2, C(q)=q3C(q) = q^3. R′(q)=100−2qR'(q) = 100 - 2q, C′(q)=3q2C'(q) = 3q^2. FOC: 100−2q=3q2⇒3q2+2q−100=0100 - 2q = 3q^2 \Rightarrow 3q^2 + 2q - 100 = 0. Solutions: q1,2=−2±4+12006q_{1,2} = \frac{-2 \pm \sqrt{4 + 1200}}{6}. Taking the positive root: q∗≈5.45q^* \approx 5.45.


Cost Function — One Input

When the firm uses only labour ll with wage ww, and the technology is f(l)f(l):

  • To produce qq units, the firm must hire f−1(q)f^{-1}(q) units of labour.
  • The cost function is:
C(q)=w⋅f−1(q)\boxed{C(q) = w \cdot f^{-1}(q)}

How to find f−1(q)f^{-1}(q): Start from q=f(l)q = f(l) and isolate ll.

Examples
  • f(l)=lf(l) = \sqrt{l} → l=q2l = q^2 → C(q)=w⋅q2C(q) = w \cdot q^2
  • f(l)=l2f(l) = \frac{l}{2} → l=2ql = 2q → C(q)=2wqC(q) = 2wq

Diminishing Marginal Productivity: We assume f′′(l)<0f''(l) < 0, meaning each additional worker adds less output than the previous one. This is economically intuitive: a kitchen with 10 cooks is not 10 times as productive as a kitchen with 1 cook.

Why does this make C(q)C(q) convex?

If f(l)f(l) is concave, then f−1(q)f^{-1}(q) is convex. Since C(q)=w⋅f−1(q)C(q) = w \cdot f^{-1}(q), the cost function is convex, meaning C′′(q)>0C''(q) > 0. In plain terms: it costs more to produce each extra unit as output rises, because you need disproportionately more labour.

Optimal Profit Function Properties

For a price-taking firm with R(q)=p⋅qR(q) = p \cdot q, define:

Π∗(w,p)=p⋅q∗(w,p)−w⋅f−1(q∗(w,p))\Pi^*(w, p) = p \cdot q^*(w,p) - w \cdot f^{-1}(q^*(w,p))

This function has three key properties:

Property Statement Intuition
Homogeneous degree 1 Π∗(tw,tp)=t⋅Π∗(w,p)\Pi^*(tw, tp) = t \cdot \Pi^*(w,p) for t>0t>0 Doubling all prices doubles profit — the optimal quantity doesn't change
Increasing in pp p^>p~⇒Π∗(p^,w)>Π∗(p~,w)\hat{p} > \tilde{p} \Rightarrow \Pi^*(\hat{p},w) > \Pi^*(\tilde{p},w) Higher output price → higher revenue for any quantity produced
Decreasing in ww Higher ww → higher costs → lower profit A wage rise squeezes margins
Proof sketch (increasing in pp)

Under p^\hat{p}, the firm could still produce q~∗\tilde{q}^* — but it won't, because it can do better by optimising again. So: Π∗(p^,w)≥p^⋅q~∗−w⋅f−1(q~∗)>p~⋅q~∗−w⋅f−1(q~∗)=Π∗(p~,w)\Pi^*(\hat{p},w) \geq \hat{p} \cdot \tilde{q}^* - w \cdot f^{-1}(\tilde{q}^*) > \tilde{p} \cdot \tilde{q}^* - w \cdot f^{-1}(\tilde{q}^*) = \Pi^*(\tilde{p},w)


Cost Function — Multiple Inputs

In reality, firms use multiple inputs. With labour LL and capital KK at prices plp_l and pkp_k, and production function f(l,k)f(l,k), the firm's problem is:

min⁡l,k  pl⋅l+pk⋅ksubject tof(l,k)=q0\min_{l,k} \; p_l \cdot l + p_k \cdot k \quad \text{subject to} \quad f(l,k) = q_0

This is the dual of the consumer's expenditure minimisation problem from Micro 1.

Key Concepts

Marginal Physical Product (MP)
  • MPl(l,k)=∂f∂l(l,k)MP_l(l,k) = \frac{\partial f}{\partial l}(l,k) — extra output from one more unit of labour, holding capital fixed
  • MPk(l,k)=∂f∂k(l,k)MP_k(l,k) = \frac{\partial f}{\partial k}(l,k) — extra output from one more unit of capital, holding labour fixed
Isoquant

An isoquant records all combinations (l,k)(l,k) that produce the same quantity q0q_0: {(l,k):f(l,k)=q0}\{(l,k) : f(l,k) = q_0\} Higher isoquants correspond to higher output.

Rate of Technical Substitution (RTS)

The RTS is the rate at which labour can be substituted for capital while keeping output constant (the slope of the isoquant): RTS(l,k)=MPlMPk(l,k)\boxed{RTS(l,k) = \frac{MP_l}{MP_k}(l,k)} Intuition: if you increase labour by one unit (gaining MPlMP_l in output), how many units of capital must you give up to stay on the isoquant? You need to reduce output by MPlMP_l, and each unit of capital you drop saves MPkMP_k in output — so you drop MPl/MPkMP_l / MP_k units of capital.

Assumption — Diminishing RTS: As ll increases along an isoquant, the RTS decreases. This means the isoquant is convex. Economically: when you have lots of labour relative to capital, substituting one more unit of labour for capital creates less and less gain — you'd need to give up less and less capital to compensate.

Diversity in production pays

Diminishing RTS implies that a blend of inputs produces more than an extreme bundle. If (4,2)(4,2) and (2,4)(2,4) both lie on the isoquant f=8f=8, then (3,3)(3,3) lies on a higher isoquant: f(3,3)=9>8f(3,3) = 9 > 8 under f(l,k)=l⋅kf(l,k) = l \cdot k.

The Optimality Condition

Isocost Line

An isocost is the set of all (l,k)(l,k) bundles with the same total cost c0c_0: pl⋅l+pk⋅k=c0p_l \cdot l + p_k \cdot k = c_0 The slope of an isocost equals −plpk-\frac{p_l}{p_k}.

The firm minimises cost on the isoquant by finding the isocost with the lowest value that still touches the isoquant. At an interior solution, the isocost is tangent to the isoquant:

RTS(l∗,k∗)=plpk\boxed{RTS(l^*, k^*) = \frac{p_l}{p_k}}

Why this is right: The RTS tells you how much capital you can sacrifice for one more unit of labour while staying on the isoquant. The price ratio tells you how much capital one unit of labour costs. If RTS > price ratio, you can swap labour for capital, produce the same output, and pay less — so that can't be optimal. Similarly if RTS < price ratio. Only when they're equal can't you do better.

RTS vs pl/pkp_l/p_k Implication Firm's action
RTS>pl/pkRTS > p_l/p_k Labour is "cheap" relative to its productivity gain Increase labour (move right along isoquant)
RTS<pl/pkRTS < p_l/p_k Capital is "cheap" relative to its productivity gain Decrease labour (move left along isoquant)
RTS=pl/pkRTS = p_l/p_k Cost is minimised Optimal point

Interior Solution: Cobb-Douglas

For f(l,k)=lαkβf(l,k) = l^\alpha k^\beta:

MPl=αlα−1kβ,MPk=βlαkβ−1MP_l = \alpha l^{\alpha-1} k^\beta, \quad MP_k = \beta l^\alpha k^{\beta-1}
RTS=αβ⋅klRTS = \frac{\alpha}{\beta} \cdot \frac{k}{l}

Setting RTS=pl/pkRTS = p_l/p_k gives the cost-minimising ratio k/l=βplαpkk/l = \frac{\beta p_l}{\alpha p_k}.

Worked Example (Cobb-Douglas)

f(l,k)=l0.5k0.5f(l,k) = l^{0.5} k^{0.5}, pl=1p_l = 1, pk=4p_k = 4. RTS=k/lRTS = k/l. Set k/l=1/4k/l = 1/4, so l=4kl = 4k. Substitute into f=q0f = q_0: k0.5(4k)0.5=q0⇒2k=q0⇒k(q0)=q0/2k^{0.5}(4k)^{0.5} = q_0 \Rightarrow 2k = q_0 \Rightarrow k(q_0) = q_0/2, l(q0)=2q0l(q_0) = 2q_0. Cost: C(q0)=1⋅2q0+4⋅q0/2=2q0+2q0=4q0C(q_0) = 1 \cdot 2q_0 + 4 \cdot q_0/2 = 2q_0 + 2q_0 = 4q_0.

Corner Solutions

When the RTS is either always above or always below the price ratio on the isoquant, the optimal bundle uses only one input:

Scenario Corner solution
RTS>pl/pkRTS > p_l/p_k everywhere on isoquant Use only labour: k(q0)=0k(q_0) = 0, l(q0)l(q_0) from f(l,0)=q0f(l,0) = q_0
RTS<pl/pkRTS < p_l/p_k everywhere on isoquant Use only capital: l(q0)=0l(q_0) = 0, k(q0)k(q_0) from f(0,k)=q0f(0,k) = q_0
Corner Solution Example

f(k,l)=k+lf(k,l) = k + l, pl=10p_l = 10, pk=20p_k = 20. MPl=MPk=1⇒RTS=1MP_l = MP_k = 1 \Rightarrow RTS = 1. Price ratio =10/20=1/2= 10/20 = 1/2. Since RTS=1>1/2RTS = 1 > 1/2: corner solution with k=0k = 0. l(q0)=q0l(q_0) = q_0, so C(q0)=10q0C(q_0) = 10 q_0.

Quasi-Linear Production: f(k,l)=g(l)+kf(k,l) = g(l) + k

This special case has MPl=g′(l)MP_l = g'(l) and MPk=1MP_k = 1, so RTS=g′(l)RTS = g'(l).

The optimal labour l∗l^* satisfies g′(l∗)=pl/pkg'(l^*) = p_l / p_k, independent of q0q_0.

  • If q0≤g(l∗)q_0 \leq g(l^*): firm only uses labour. l(q0)=g−1(q0)l(q_0) = g^{-1}(q_0), k(q0)=0k(q_0) = 0.
  • If q0>g(l∗)q_0 > g(l^*): firm uses l∗l^* units of labour plus capital for the rest. l(q0)=l∗l(q_0) = l^*, k(q0)=q0−g(l∗)k(q_0) = q_0 - g(l^*).
C(q)={pl⋅g−1(q)if q≤g(l∗)pl⋅l∗+pk⋅(q−g(l∗))if q>g(l∗)C(q) = \begin{cases} p_l \cdot g^{-1}(q) & \text{if } q \leq g(l^*) \\ p_l \cdot l^* + p_k \cdot (q - g(l^*)) & \text{if } q > g(l^*) \end{cases}
Quasi-Linear Example

f(k,l)=ln⁡(1+l)+kf(k,l) = \ln(1+l) + k, pl=2p_l = 2, pk=6p_k = 6. g′(l)=11+lg'(l) = \frac{1}{1+l}. Set =pl/pk=1/3= p_l/p_k = 1/3: 1+l∗=3⇒l∗=21+l^* = 3 \Rightarrow l^* = 2. g(l∗)=ln⁡(3)g(l^*) = \ln(3).

  • If q0≤ln⁡(3)q_0 \leq \ln(3): l(q0)=eq0−1l(q_0) = e^{q_0} - 1, C=2(eq0−1)C = 2(e^{q_0}-1).
  • If q0>ln⁡(3)q_0 > \ln(3): l=2l = 2, k=q0−ln⁡(3)k = q_0 - \ln(3), C=4+6(q0−ln⁡(3))C = 4 + 6(q_0 - \ln(3)).

Part 2 — Partial Equilibrium

What Is Partial Equilibrium?

Partial equilibrium analyses a single market (good XX) holding all other prices and incomes fixed. This is an approximation — it works well when the market for XX is not too interlinked with other markets.

Partial Equilibrium (Definition)

A partial equilibrium is a price pep^e such that the quantity supplied by profit-maximising firms equals the quantity demanded by utility-maximising consumers: XD(pe)=XS(pe)X^D(p^e) = X^S(p^e)

The Demand Side

Each consumer maximises utility subject to a budget constraint (from Micro 1). We aggregate to get the market demand XD(p)X^D(p), which is decreasing in pp.

  • Demand curve: pairs (p,x)(p, x) on XD(p)X^D(p).
  • Inverse demand: PD(x)P^D(x) — price at which consumers demand exactly xx units.

The Supply Side

Each firm is a price taker (takes pp as given). With R(x)=p⋅xR(x) = p \cdot x:

p=C′(x)⇒xS(p)=(C′)−1(p)=PS(x)p = C'(x) \quad \Rightarrow \quad x^S(p) = (C')^{-1}(p) = P^S(x)

Firms supply until price equals marginal cost. The supply curve is thus the marginal cost curve.

Finding Equilibrium

Equilibrium is where PD(xe)=PS(xe)P^D(x^e) = P^S(x^e): the demand and supply curves cross.

Efficiency in Partial Equilibrium

To connect equilibrium to welfare, we make two special assumptions:

  1. Quasi-linear consumer: Utility U(x,y)=v(x)+yU(x,y) = v(x) + y where v′′(x)<0v''(x) < 0.

    • Optimal demand: v′(x)=pxv'(x) = p_x, so XD(p)=v′−1(p)X^D(p) = v'^{-1}(p).
    • Inverse demand: PD(x)=v′(x)P^D(x) = v'(x) — the demand curve equals marginal utility.
  2. Representative firm: C(0)=0C(0) = 0, C′′(x)≥0C''(x) \geq 0.

    • Supply: PS(x)=C′(x)P^S(x) = C'(x) — the supply curve equals marginal cost.

Under these assumptions, social welfare is:

W(x)=v(x)−C(x)W(x) = v(x) - C(x)

The socially optimal x∗x^* satisfies v′(x∗)=C′(x∗)v'(x^*) = C'(x^*), which is exactly the competitive equilibrium condition. This proves:

First Welfare Theorem (Partial Equilibrium Version)

Under quasi-linear preferences, the competitive equilibrium quantity maximises social welfare.

Consumer Surplus and Firm Surplus

Welfare is distributed between consumers and firms:

CS=∫0x1PD(x)−p1  dx=v(x1)−p1x1CS = \int_0^{x_1} P^D(x) - p_1 \; dx = v(x_1) - p_1 x_1
FS=∫0x1p1−PS(x)  dx=p1x1−C(x1)FS = \int_0^{x_1} p_1 - P^S(x) \; dx = p_1 x_1 - C(x_1)
CS+FS=v(x1)−C(x1)=W(x1)\boxed{CS + FS = v(x_1) - C(x_1) = W(x_1)}

CS is the area between the demand curve and the price (value consumers get above what they pay). FS is the area between the price and the supply curve (revenue minus cost).

Common mistake

CS and FS are not always accurate welfare measures. They only equal social welfare under the quasi-linear preference assumption. In general, income effects mean that CS can overstate or understate true welfare gains.

Government Intervention: Subsidies

A subsidy Δ\Delta per unit means consumers pay p−Δp - \Delta instead of pp. This shifts the demand curve up by Δ\Delta (or equivalently, demand at price pp rises to XD(p−Δ)X^D(p - \Delta)).

Effect: Equilibrium quantity rises. CS and FS both increase. But the government spends Δ⋅xnewe\Delta \cdot x^e_{\text{new}} — and this cost exceeds the combined gain in CS + FS, creating a deadweight loss.

Numerical Subsidy Example

Demand: XD(p)=100−pX^D(p) = 100 - p. Cost: C(x)=x2/2C(x) = x^2/2.

Without subsidy: Equil: 100−x=x⇒x∗=50100 - x = x \Rightarrow x^* = 50, p∗=50p^* = 50. CS=∫050(100−x−50)dx=1250CS = \int_0^{50}(100-x-50)dx = 1250. FS=1250FS = 1250. Total welfare = 25002500.

With subsidy Δ=20\Delta = 20: Demand becomes XD(p)=120−pX^D(p) = 120 - p. Equil: 120−x=x⇒x∗=60120 - x = x \Rightarrow x^* = 60, p∗=60p^* = 60 (firm price); consumer pays 4040. CS=∫060(120−x−60)dx=1800CS = \int_0^{60}(120-x-60)dx = 1800. FS=1800FS = 1800. Gains = 36003600. Government cost = 20×60=120020 \times 60 = 1200. Net welfare = 3600−1200=2400<25003600 - 1200 = 2400 < 2500. Deadweight loss = 100.

Intuition for the welfare loss

At the competitive equilibrium, PD=PSP^D = P^S — the last unit produced is worth exactly what it costs. A subsidy induces production of units where PD<PSP^D < P^S, i.e., units that cost more to produce than consumers value them at. The government pays the difference, creating waste.

Limitations of Partial Equilibrium

Partial equilibrium ignores price interactions across markets. If goods XX and YY are substitutes or complements, a price change in XX shifts the demand curve for YY, affecting welfare in ways not captured by partial equilibrium. For the full picture, we need general equilibrium.


Part 3 — General Equilibrium

Why General Equilibrium?

Partial equilibrium treated the prices of other goods as fixed. In reality:

  • Substitutes and complements create cross-market spillovers.
  • Income effects from one market affect demand in others.
  • Government policy in market XX creates ripple effects across the economy.

General equilibrium models all markets simultaneously.

The Exchange Economy

We simplify by removing production: agents trade their endowments.

Exchange Economy
  • Two agents (1 and 2) with utility functions U1(x1,y1)U_1(x_1, y_1) and U2(x2,y2)U_2(x_2, y_2).
  • Two goods XX and YY.
  • Agent ii's initial endowment: (ωxi,ωyi)(\omega^i_x, \omega^i_y).
  • An allocation is feasible if: x1+x2=ωx1+ωx2x_1 + x_2 = \omega^1_x + \omega^2_x and y1+y2=ωy1+ωy2y_1 + y_2 = \omega^1_y + \omega^2_y.
  • The initial allocation (ωx1,ωy1),(ωx2,ωy2)(\omega^1_x, \omega^1_y), (\omega^2_x, \omega^2_y) is called the endowment.

The Edgeworth Box

The Edgeworth box is a graphical tool for visualising all feasible allocations simultaneously.

  • Agent 1's perspective: bottom-left origin O1O_1; axes measure x1x_1 (right) and y1y_1 (up).
  • Agent 2's perspective: top-right origin O2O_2; axes measure x2x_2 (left) and y2y_2 (down) — agent 2's diagram is "flipped" and overlaid.
  • Box dimensions: width =ωx1+ωx2= \omega^1_x + \omega^2_x, height =ωy1+ωy2= \omega^1_y + \omega^2_y.
  • Each point in the box represents a unique feasible allocation.

Agent 1's indifference curves look normal (convex to O1O_1). Agent 2's are convex to O2O_2 (flipped upside-down).

Pareto Efficiency

Pareto Efficiency

An allocation is Pareto efficient if there is no feasible alternative allocation that makes every agent at least as well off and at least one agent strictly better off. Equivalently: it is impossible to make someone better off without making someone else worse off.

Pareto efficiency ≠ fairness

There can be many Pareto efficient allocations, including very unequal ones. Pareto efficiency is a necessary condition for a good outcome, not a sufficient one. It says nothing about distributional equity.

Condition for interior Pareto efficiency:

An interior allocation is Pareto efficient if and only if the indifference curves of both agents are tangent at that point:

MRS1=MRS2\boxed{MRS_1 = MRS_2}

If MRS1≠MRS2MRS_1 \neq MRS_2 at an allocation, there exists a trade that makes both better off:

  • If MRS1<MRS2MRS_1 < MRS_2: take 1 unit of XX from agent 1, give it to agent 2, compensate agent 1 with MRS1MRS_1 units of YY from agent 2. Agent 1 is indifferent; agent 2 gains because agent 2 valued that unit of XX at MRS2>MRS1MRS_2 > MRS_1 units of YY.
  • If MRS1>MRS2MRS_1 > MRS_2: the reverse trade improves both.

The Contract Curve

The contract curve (or Pareto set) is the set of all Pareto efficient allocations in the Edgeworth box. It is found by:

MRS1(x1,y1)=MRS2(x2,y2)MRS_1(x_1, y_1) = MRS_2(x_2, y_2)
x1+x2=X,y1+y2=Yx_1 + x_2 = X, \quad y_1 + y_2 = Y

Three equations, four unknowns → infinitely many solutions (a curve).

Contract Curve Example 1

U1=(x1)2y1U_1 = (x_1)^2 y_1, U2=(x2)2y2U_2 = (x_2)^2 y_2, X=30X = 30, Y=40Y = 40. MRS1=2y1x1MRS_1 = \frac{2y_1}{x_1}, MRS2=2y2x2MRS_2 = \frac{2y_2}{x_2}. Tangency: y1x1=y2x2\frac{y_1}{x_1} = \frac{y_2}{x_2}. Resource constraints: x2=30−x1x_2 = 30 - x_1, y2=40−y1y_2 = 40 - y_1. Substituting: y1=43x1y_1 = \frac{4}{3} x_1. This is the contract curve — a straight line through both origins.

Contract Curve Example 2 (asymmetric)

U1=x1y1U_1 = x_1 y_1, U2=(x2)2y2U_2 = (x_2)^2 y_2, X=30X = 30, Y=40Y = 40. MRS1=y1/x1MRS_1 = y_1/x_1, MRS2=2y2/x2MRS_2 = 2y_2/x_2. Solving with resource constraints: y1=80x130+x1y_1 = \frac{80 x_1}{30 + x_1}. Note the egalitarian point (15,20)(15, 20) is not on this curve.

Equilibrium in the Exchange Economy

Agents face prices (px,py)(p_x, p_y). Agent ii's income is Ii=pxωxi+pyωyiI_i = p_x \omega^i_x + p_y \omega^i_y — the market value of their endowment.

Agent ii maximises utility subject to: pxxi+pyyi=Iip_x x_i + p_y y_i = I_i.

In the Edgeworth box, each agent's budget constraint is a line through the endowment point with slope −px/py-p_x/p_y. Both agents face the same line (since the endowment is a single point in the box).

Equilibrium in an Exchange Economy

An equilibrium is an allocation {(x1∗,y1∗),(x2∗,y2∗)}\{(x^*_1, y^*_1), (x^*_2, y^*_2)\} and prices (px∗,py∗)(p^*_x, p^*_y) such that:

  1. Each agent's bundle is utility-maximising given their budget constraint.
  2. Markets clear: x1∗+x2∗=ωx1+ωx2x^*_1 + x^*_2 = \omega^1_x + \omega^2_x and y1∗+y2∗=ωy1+ωy2y^*_1 + y^*_2 = \omega^1_y + \omega^2_y.

In the Edgeworth box, equilibrium is the point where both agents' chosen bundles coincide — and their indifference curves are tangent to the budget line.

Finding the equilibrium: Adjust the price ratio px/pyp_x/p_y until supply equals demand for both goods.

Walras' Law

Walras' Law

For any prices (px,py)(p_x, p_y) and any bundles on the agents' budget constraints: px[x1(p)+x2(p)−ωx1−ωx2]+py[y1(p)+y2(p)−ωy1−ωy2]=0p_x [x_1(p) + x_2(p) - \omega^1_x - \omega^2_x] + p_y [y_1(p) + y_2(p) - \omega^1_y - \omega^2_y] = 0

Proof: Each agent spends exactly their income:

px(xi−ωxi)+py(yi−ωyi)=0p_x(x_i - \omega^i_x) + p_y(y_i - \omega^i_y) = 0

Summing over both agents gives Walras' Law.

Key implication: If the market for good XX clears, the market for good YY must automatically clear (and vice versa). So to find equilibrium, you only need to clear one market.

Relative Prices

Equilibrium prices are only determined up to a positive scalar — only the ratio px/pyp_x/p_y matters. Multiplying all prices by t>0t > 0 leaves budgets unchanged and thus leaves all demands unchanged. Convention: normalise py=1p_y = 1 and solve for pxp_x.

Equilibrium Example

U1=U2=(xi)2yiU_1 = U_2 = (x_i)^2 y_i, ω1=ω2=(15,20)\omega^1 = \omega^2 = (15, 20). Demand: solving FOC with budget constraint gives xi∗=10+403pxx^*_i = 10 + \frac{40}{3p_x}. Market clearing: x1∗+x2∗=30⇒20+803px∗=30⇒px∗=83x^*_1 + x^*_2 = 30 \Rightarrow 20 + \frac{80}{3p^*_x} = 30 \Rightarrow p^*_x = \frac{8}{3}. Equilibrium allocation: each agent keeps their endowment (15,20)(15, 20) — symmetric agents never trade!

The First Welfare Theorem

First Welfare Theorem

Every competitive equilibrium allocation is Pareto efficient. Equilibrium⇒Pareto Efficiency\text{Equilibrium} \Rightarrow \text{Pareto Efficiency}

Proof by contradiction: Suppose the equilibrium {(x1∗,y1∗),(x2∗,y2∗)}\{(x^*_1, y^*_1), (x^*_2, y^*_2)\} is NOT Pareto efficient. Then there exists a feasible allocation {(x1′,y1′),(x2′,y2′)}\{(x'_1, y'_1), (x'_2, y'_2)\} that both agents strictly prefer.

Since each agent is choosing the best bundle they can afford in equilibrium, if (xi′,yi′)≻i(xi∗,yi∗)(x'_i, y'_i) \succ_i (x^*_i, y^*_i), then agent ii must not be able to afford it:

pxxi′+pyyi′>pxωxi+pyωyip_x x'_i + p_y y'_i > p_x \omega^i_x + p_y \omega^i_y

Summing over both agents:

px(x1′+x2′)+py(y1′+y2′)>px(ωx1+ωx2)+py(ωy1+ωy2)p_x(x'_1 + x'_2) + p_y(y'_1 + y'_2) > p_x(\omega^1_x + \omega^2_x) + p_y(\omega^1_y + \omega^2_y)

But feasibility requires x1′+x2′=ωx1+ωx2x'_1 + x'_2 = \omega^1_x + \omega^2_x and y1′+y2′=ωy1+ωy2y'_1 + y'_2 = \omega^1_y + \omega^2_y. Substituting gives 0>00 > 0 — a contradiction. ∎

What the First Welfare Theorem tells us

Competitive markets, left alone, produce allocations that can't be improved without making someone worse off. This is the theoretical foundation for the efficiency case for free markets. However:

  • It says nothing about fairness — the equilibrium can be very unequal.
  • It requires complete markets, no externalities, no market power, and perfect information.
  • The Second Welfare Theorem (not covered here) says any Pareto efficient allocation can be achieved as a competitive equilibrium, provided we can redistribute initial endowments.

Summary

Topic Core Result
Firm's optimum MR=MCMR = MC
One-input cost function C(q)=w⋅f−1(q)C(q) = w \cdot f^{-1}(q); convex if ff concave
Two-input cost minimisation Interior: RTS=pl/pkRTS = p_l/p_k; corner solutions exist
Partial equilibrium PD(xe)=PS(xe)P^D(x^e) = P^S(x^e); CS+FS=CS + FS = social welfare (quasi-linear)
Subsidy Increases CS and FS but creates deadweight loss
Pareto efficiency Interior: MRS1=MRS2MRS_1 = MRS_2
Contract curve Set of all Pareto efficient allocations; MRS1=MRS2MRS_1 = MRS_2 + resource constraints
Exchange equilibrium Utility max + market clearing; unique up to price normalisation
Walras' Law One market clearing implies all markets clear
First Welfare Theorem Every competitive equilibrium is Pareto efficient