Week 6 · Dr. Boaz Zik

Public Goods, Free Riding & Altruism

Public Goods, Free Riding & Altruism

Part of: Micro-Economics II | Key concepts: Public Goods, Free Riding, Nash Equilibrium, Best Response, Under-Provision, Altruistic Utility, Market Failure, Pareto Efficiency


What This Presentation Is About

This is the final section of the cumulative Presentation 6 deck, and it continues straight on from Lec_05-Production Economy & Externalities. A public good is treated as a special kind of externality, one that hits many agents at once. That small difference breaks the usual fixes, because every agent now has to guess what everyone else will contribute.

The section does four things:

  1. Define a public good and name the two problems it causes (under-provision and free riding).
  2. Illustrate with a small numerical game (Alice and Bob's seminar work).
  3. Solve the general model twice: the social planner's optimum, then the free-market Nash equilibrium.
  4. Add altruism: what happens when agents partly care about each other's utility.
Source scope

Slides 298 to 313 of the deck. Slide 297 is only a section divider ("Externalities / Public Goods").


Part 1 — What Is a Public Good?

Public Good

A public good is a good whose aggregate quantity affects agents' utilities. Everyone's utility depends on the total amount provided, not just on what they provide themselves.

Slide examples: national security, a public swimming pool, sidewalks, an elevator in a building, seminar work (a joint group project).

Why it is "a special kind of externality"

An ordinary externality usually involves a pair of agents. A public good is an externality that affects many agents at once. That creates a coordination problem:

  • When each agent decides how much to invest in the public good, she must conjecture how much the others invest.
  • The slide's link back to the previous lecture: if two agents (Homer and Ned) suffer from Barney's smoking, and Ned wants to bargain with Barney over how much to cut cigar consumption, Ned's decision depends on what Homer is doing.

The two phenomena

Phenomenon What happens
Under-Provision (a Market Failure) The public good is provided at a level that is too low
Free Riding An agent invests less (or nothing) because other agents' investment already gives her the benefit

Part 2 — Example: Alice and Bob's Seminar Work

Setup

Two students, Alice and Bob, hand in one joint seminar paper.

  • Each chooses an investment eA,eB∈{0,1,2,3,… }e_A, e_B \in \{0, 1, 2, 3, \dots\}.
  • Each unit of investment costs the investor a disutility of 1 (in money).
  • The shared grade depends only on total investment:

g(eA+eB)={FeA+eB=0DeA+eB=1CeA+eB=2BeA+eB=3AeA+eB=4g(e_A + e_B) = \begin{cases} F & e_A + e_B = 0 \\ D & e_A + e_B = 1 \\ C & e_A + e_B = 2 \\ B & e_A + e_B = 3 \\ A & e_A + e_B = 4 \end{cases}

The grade is the public good: both students get the same one, whoever worked.

Utility from each grade (in money terms):

Grade F D C B A
Alice UAU_A 0 3 5 6 6.5
Bob UBU_B 0 2 2.9 3.5 4.1

Alice is motivated, Bob is less so. Net payoff is the grade utility minus the student's own investment cost.

Best responses (as stated on the slide)

  • Bob: if eA≥1e_A \ge 1, Bob does not invest (cost exceeds benefit). If eA=0e_A = 0, Bob invests 1.
  • Alice: if eB≥3e_B \ge 3, Alice does not invest. If eB<3e_B < 3, she invests 3−eB3 - e_B.
Stable outcome: free riding

The only stable situation is Bob invests 0, Alice invests 3, and the grade is B. Bob free rides on Alice's effort.

A Pareto improvement exists

Force Bob to invest 1 (total 4, grade A), and have Alice pay him a fee ff with 0.4<f<0.50.4 < f < 0.5.

My check of the slide's numbers

Not on the slide, my working. At the free-riding outcome, Alice nets 6−3=36 - 3 = 3 and Bob nets 3.53.5.

Outcome Alice Bob
Free riding (eA=3,eB=0e_A=3, e_B=0, grade B) 6−3=36 - 3 = 3 3.53.5
Bob 1, Alice 3, grade A, fee ff 6.5−3−f=3.5−f6.5 - 3 - f = 3.5 - f 4.1−1+f=3.1+f4.1 - 1 + f = 3.1 + f

Alice is better off iff 3.5−f>33.5 - f > 3, i.e. f<0.5f < 0.5. Bob is better off iff 3.1+f>3.53.1 + f > 3.5, i.e. f>0.4f > 0.4. This reproduces the slide's range.

Small tie in Alice's choice (my observation)

With eB=0e_B = 0, Alice nets 5−2=35 - 2 = 3 by investing 2 (grade C) and 6−3=36 - 3 = 3 by investing 3 (grade B). She is indifferent. The slide picks 3 (the rule "3−eB3 - e_B"), so I follow that, but the equilibrium is knife-edge for Alice.


Part 3 — The General Model

Model setup
  • Two agents, 1 and 2.
  • Two goods: the public good GG and money YY.
  • Quasilinear Utility for both:

U1(g,y)=V1(g)+yU2(g,y)=V2(g)+yU_1(g, y) = V_1(g) + y \qquad U_2(g, y) = V_2(g) + y

  • Note gg has no agent index: both consume the same gg.
  • Each agent can produce one unit of gg at a cost of 1 dollar (1 unit of YY).
  • Agents have money endowments I1I_1 and I2I_2, so gi+yi=Iig_i + y_i = I_i.

Because each agent produces gig_i and the total is g=g1+g2g = g_1 + g_2, each agent's utility is Vi(g1+g2)+Ii−giV_i(g_1 + g_2) + I_i - g_i.


3a. The Social Planner's Problem

The Social Welfare maximiser picks everyone's contribution and money:

max⁡g1,g2,y1,y2  V1(g1+g2)+y1+V2(g1+g2)+y2\max_{g_1, g_2, y_1, y_2} \; V_1(g_1 + g_2) + y_1 + V_2(g_1 + g_2) + y_2

subject to g1+y1=I1g_1 + y_1 = I_1 and g2+y2=I2g_2 + y_2 = I_2.

Typo on slide 304

The slide writes the second constraint as g2+y2=I1g_2 + y_2 = I_1. It should be I2I_2 (the substituted objective on the same slide correctly uses I2I_2).

Substituting the constraints into the objective:

max⁡g1,g2  V1(g1+g2)+I1−g1+V2(g1+g2)+I2−g2\max_{g_1, g_2} \; V_1(g_1 + g_2) + I_1 - g_1 + V_2(g_1 + g_2) + I_2 - g_2

First-order conditions (one per gig_i):

∂V1∂g(g1∗+g2∗)+∂V2∂g(g1∗+g2∗)=1(w.r.t. g1)\frac{\partial V_1}{\partial g}(g_1^* + g_2^*) + \frac{\partial V_2}{\partial g}(g_1^* + g_2^*) = 1 \quad \text{(w.r.t. } g_1\text{)}
∂V1∂g(g1∗+g2∗)+∂V2∂g(g1∗+g2∗)=1(w.r.t. g2)\frac{\partial V_1}{\partial g}(g_1^* + g_2^*) + \frac{\partial V_2}{\partial g}(g_1^* + g_2^*) = 1 \quad \text{(w.r.t. } g_2\text{)}

Both conditions are identical, because ViV_i depends only on the sum g1+g2g_1 + g_2. Writing g∗=g1∗+g2∗g^* = g_1^* + g_2^*:

∂V1∂g(g∗)+∂V2∂g(g∗)=1\boxed{\frac{\partial V_1}{\partial g}(g^*) + \frac{\partial V_2}{\partial g}(g^*) = 1}
Interpretation (Samuelson-style condition)

The public good is produced until its marginal social benefit (the sum of both agents' marginal benefits) equals its marginal cost (1).

Many efficient splits

Only the total g∗g^* is pinned down, so there are many (g1,g2)(g_1, g_2) with g1+g2=g∗g_1 + g_2 = g^*. The slide states: an allocation is Pareto efficient if and only if g1+g2=g∗g_1 + g_2 = g^*.


3b. Why the Condition Characterises Efficiency

The slide shows that if the condition fails, a Pareto improvement exists. Suppose

∂V1∂g(g)+∂V2∂g(g)>1.\frac{\partial V_1}{\partial g}(g) + \frac{\partial V_2}{\partial g}(g) > 1.

Ask agent 1 to produce one more unit. Agent 1's utility changes by ∂V1∂g(g)−1\frac{\partial V_1}{\partial g}(g) - 1.

Case What happens
∂V1∂g−1>0\frac{\partial V_1}{\partial g} - 1 > 0 Agent 1 gains. Agent 2 also gains (extra gg at no cost). Pareto improvement.
∂V1∂g−1<0\frac{\partial V_1}{\partial g} - 1 < 0 Agent 2 compensates agent 1 by paying 1−∂V1∂g1 - \frac{\partial V_1}{\partial g}. Agent 1 is back at his old utility.

In the second case agent 2's utility changes by

∂V2∂g(g)−(1−∂V1∂g(g))=∂V1∂g(g)+∂V2∂g(g)−1>0,\frac{\partial V_2}{\partial g}(g) - \left(1 - \frac{\partial V_1}{\partial g}(g)\right) = \frac{\partial V_1}{\partial g}(g) + \frac{\partial V_2}{\partial g}(g) - 1 > 0,

which is positive by the assumption, so agent 2 gains and agent 1 is no worse off. Pareto improvement.

My intuition

This is the same idea as the Alice/Bob fee: when the combined marginal benefit exceeds the marginal cost, there is surplus on the table. Whoever produces the extra unit can be paid out of the other agent's gain, so both end up better off. (I have kept the slide's argument for the >1>1 case only, it is the only case the slide covers.)


3c. The Free Market: Nash Equilibrium

To model the market without a planner, the slides import game theory.

Nash Equilibrium

A pair (g1e,g2e)(g_1^e, g_2^e) is a Nash equilibrium if g1eg_1^e is agent 1's optimal choice given that agent 2 plays g2eg_2^e, and g2eg_2^e is agent 2's optimal choice given that agent 1 plays g1eg_1^e.

Step 1: the stand-alone optimum. If agent 2 produces nothing, agent 1 would produce g1og_1^o solving

max⁡g1  V1(g1)+I1−g1⟹V1′(g1o)=1\max_{g_1} \; V_1(g_1) + I_1 - g_1 \quad \Longrightarrow \quad V_1'(g_1^o) = 1

(g1og_1^o is how much agent 1 wants in total if she had to provide it all herself.)

Step 2: best response. Agent 1 only cares about the total g1+g2g_1 + g_2, so:

  • If g2<g1og_2 < g_1^o: agent 1 tops the total up to her ideal, g1+g2=g1og_1 + g_2 = g_1^o.
  • If g2≥g1og_2 \ge g_1^o: agent 1's utility is decreasing in g1g_1, so she produces 0.
BR1(g2)={g1o−g2if g2<g1o0if g2≥g1oBR_1(g_2) = \begin{cases} g_1^o - g_2 & \text{if } g_2 < g_1^o \\ 0 & \text{if } g_2 \ge g_1^o \end{cases}

Analogously for agent 2:

BR2(g1)={g2o−g1if g1<g2o0if g1≥g2oBR_2(g_1) = \begin{cases} g_2^o - g_1 & \text{if } g_1 < g_2^o \\ 0 & \text{if } g_1 \ge g_2^o \end{cases}

A Nash equilibrium satisfies g1e=BR1(g2e)g_1^e = BR_1(g_2^e) and g2e=BR2(g1e)g_2^e = BR_2(g_1^e).

The equilibria, by case

Case Nash equilibrium Interpretation
g1o=g2og_1^o = g_2^o Every pair with g1+g2=g1o=g2og_1 + g_2 = g_1^o = g_2^o Agents want the same total, so any split works
g2o>g1og_2^o > g_1^o Only g1=0,  g2=g2og_1 = 0,\; g_2 = g_2^o Complete free riding by agent 1
g2o<g1og_2^o < g_1^o Only g2=0,  g1=g1og_2 = 0,\; g_1 = g_1^o Complete free riding by agent 2
My intuition

The agent who wants the public good more (higher stand-alone level gog^o) ends up providing all of it. The other agent's ideal total is already reached, so she contributes nothing. This matches the Alice/Bob result, where motivated Alice carries the load.


3d. Why the Market Under-Provides

Result: under-provision

In any equilibrium, the total g1e+g2eg_1^e + g_2^e satisfies either ∂V1∂g=1\frac{\partial V_1}{\partial g} = 1 or ∂V2∂g=1\frac{\partial V_2}{\partial g} = 1 (the one who contributes is at her own optimum). This implies, at the equilibrium quantity:

∂V1∂g(g1e+g2e)+∂V2∂g(g1e+g2e)>1.\frac{\partial V_1}{\partial g}(g_1^e + g_2^e) + \frac{\partial V_2}{\partial g}(g_1^e + g_2^e) > 1.

So the equilibrium total g1e+g2eg_1^e + g_2^e is smaller than the welfare-maximising g∗g^*.

Why? Each agent cares only about her own utility and does not internalise the effect of her choice on the other agent's welfare.

Social planner Free-market Nash equilibrium
Condition V1′(g∗)+V2′(g∗)=1V_1'(g^*) + V_2'(g^*) = 1 V1′(ge)=1V_1'(g^e) = 1 or V2′(ge)=1V_2'(g^e) = 1
What is compared to cost 1 Sum of marginal benefits One agent's marginal benefit
Total quantity g∗g^* ge<g∗g^e < g^*
Who pays Any split summing to g∗g^* Often one agent only
Pareto efficient? Yes No
My intuition

The strict ">1> 1" step also relies on each ViV_i having positive, diminishing marginal benefit (concave ViV_i), which the slide uses implicitly. The key picture: at geg^e the non-contributing agent still has a positive marginal benefit, and nobody counts it.


Part 4 — Altruism

The slide's question

What happens when agents are altruists, that is, they internalise to some extent the other agent's utility?

Altruistic Utility
  • Agent 1: V1(g1+g2)+y1+α⋅(V2(g1+g2)+y2)V_1(g_1 + g_2) + y_1 + \alpha \cdot \big(V_2(g_1 + g_2) + y_2\big) with α≤1\alpha \le 1
  • Agent 2: V2(g1+g2)+y2+β⋅(V1(g1+g2)+y1)V_2(g_1 + g_2) + y_2 + \beta \cdot \big(V_1(g_1 + g_2) + y_1\big) with β≤1\beta \le 1

α\alpha and β\beta measure how much each agent weighs the other's utility (α=β=0\alpha = \beta = 0 is the selfish case above).

Claimed results (to be shown in the problem set):

  1. With altruists, the equilibrium quantity of the public good is greater than with selfish agents.
  2. When α=1\alpha = 1 or β=1\beta = 1, the equilibrium quantity equals the socially optimal quantity.
Not derived on the slides

The deck states these results and leaves the proof to the problem set, so there is no derivation here. Do not quote a formula for the altruistic equilibrium from this lecture.

My intuition

If agent 1 sets α=1\alpha = 1, she weighs agent 2's utility fully, so her private marginal benefit from gg becomes V1′+V2′V_1' + V_2', which is the planner's marginal social benefit. This is the same kind of fix as a tax in Lec_05-Production Economy & Externalities: change the incentive so the agent counts the external effect. (My reading, not stated on the slide.)


Idea In Lec_05-Production Economy & Externalities Here
Externality One agent's choice affects another's utility or cost A public good affects many agents at once
Why markets fail Private choice ignores the external effect Each agent ignores the benefit to the others
Fix Pigouvian Tax or Coase Theorem bargaining Side-payment (Alice/Bob fee), or altruism (α,β\alpha, \beta)
Difference from the bilateral case

The Coase-style bargaining of Lec 5 works between two parties. With a public good each agent must guess everyone else's investment, so the pairwise bargain breaks down (slide 299).


Summary

  • A public good is a good whose aggregate quantity enters every agent's utility; it is an externality hitting many agents at once.
  • Two problems follow: under-provision (a Market Failure) and free riding.
  • Alice/Bob example: the stable outcome is Bob 0, Alice 3 (grade B); a fee between 0.4 and 0.5 enables a Pareto improvement with Bob investing 1.
  • Model: Ui=Vi(g)+yiU_i = V_i(g) + y_i, g=g1+g2g = g_1 + g_2, cost of 1 per unit.
  • Planner: produce until V1′(g∗)+V2′(g∗)=1V_1'(g^*) + V_2'(g^*) = 1 (sum of marginal benefits equals marginal cost). Efficient iff g1+g2=g∗g_1 + g_2 = g^*.
  • Nash equilibrium: BRi(gj)=max⁡(gio−gj,0)BR_i(g_j) = \max(g_i^o - g_j, 0). If g1o≠g2og_1^o \ne g_2^o, one agent free rides completely; total provision <g∗< g^*.
  • Altruism (α,β≤1\alpha, \beta \le 1) raises provision, and at α=1\alpha = 1 or β=1\beta = 1 restores the social optimum (proof left to the problem set).