Compare external effects, corrective prices, standards, and common-resource governance.
By the end: Calculate a social marginal cost and identify what a corrective policy would need to know.
Who is missing from the transaction?
An external effect occurs when an activity changes another person's wellbeing or production possibilities in a way the decision-maker does not fully take into account through the relevant arrangements. Pollution harming downstream users is a negative example; knowledge that other producers can reuse can be a positive one. Not every price change is such an external cost: competition lowering a rival's sales is not automatically the same as physical damage.
For one additional unit of output, marginal social cost includes the extra private resource cost and the extra external harm. If private cost is 20 and unpriced harm is 15, social marginal cost is 35. A buyer willing to pay 30 would support the transaction privately, but its benefit is below the modeled social cost. This example assumes willingness to pay is the chosen benefit measure and acknowledges its dependence on income.
Follow the cost to the person who bears it
Start by separating a payment from the use of a resource. Suppose a workshop sells one item for 48 currency units. Making it uses materials, energy, and labor with an opportunity cost of 32. Opportunity cost means the value of those inputs in their best available alternative use. The workshop records a margin of 16 before any fixed costs. Suppose its discharge also causes 9 of damage to a downstream grower, who cannot charge the workshop. The workshop's margin does not contain that loss. Adding the missing cost gives a social resource cost of 41 and a net gain of 7 under these assumed valuations.
This paper example assumes that 48 measures the buyer's benefit as well as the price, inputs are valued at opportunity cost, and damage is counted once. If the buyer would have paid 55, net social gain would instead be 14. A receipt establishes how much money changed hands, not the buyer's maximum willingness to pay or the full benefit of the item.
Now compare a second item with buyer benefit 38 and the same production cost and harm. Private gains are 6 before the omitted damage, so an unregulated trade can occur at some price between 32 and 38. But the modeled net social gain is negative 3 because benefit 38 is below combined cost 41. The relevant question is about this additional item, not whether the workshop's entire activity is beneficial. Its earlier items might produce benefits far above their social costs.
Which item covers its full cost?
In this invented workshop case, every item has social cost 41: resource cost 32 plus harm 9. Benefits of 48 and 55 exceed that threshold; benefit 38 does not, despite exceeding private cost 32. These are assumed willingness-to-pay valuations, not measured welfare or evidence that affected growers receive compensation.
Marginal means associated with one additional unit. Total damage is damage across all units; marginal damage is the change caused by an extra unit. If ten identical discharges each cause 9 of independent harm, total harm is 90. That multiplication fails if a river's ability to absorb waste becomes exhausted or damage depends on the timing of discharges. Marginal harm could then rise with output. Specifying currency units per item prevents confusing a total damage estimate with a charge appropriate for the next item.
An external effect depends on the institutional arrangement. If a grower has an enforceable claim for damage and the workshop expects to pay it, the workshop faces a different private comparison. If enforcement is unreliable, a formal right may leave much of the practical cost outside its decision. A contract between these parties could also change incentives. Rights, monitoring costs, and bargaining power therefore affect whether an effect remains external; physical proximity alone does not settle the question.
The arithmetic is a conditional welfare comparison, not causal evidence that a particular workshop causes a particular illness or crop loss. Establishing that link requires exposure and outcome evidence against a credible alternative without the discharge. Choosing willingness to pay as the benefit measure is also a normative and measurement choice. People with little income may have urgent needs but little purchasing power. A positive sum in currency units cannot by itself establish that those harmed consented or that gains reached them.
The same discharge under different claims
Institutional comparison
Arrangement
Private comparison
Practical limit
No enforceable claim
Grower's loss omitted
Harm remains external
Reliable liability
Expected damages count
Causal proof needed
Unreliable enforcement
Payment may be avoided
Formal right may fail
Contract with grower
Agreed payment counts
Monitoring costs matter
Hold the workshop's physical discharge fixed while comparing who can make its cost enter the decision. Reliable liability differs from a formal but unenforced claim; bargaining requires a workable contract. None of these arrangements alone establishes the causal damage estimate or shows that affected growers consented to a currency valuation of their loss.
Corrective policy changes the relevant comparison
A per-unit charge matching marginal external damage at the desired outcome can align private incentives with social costs under strong conditions. For our constant-damage example, a charge of 15 makes the seller face a cost of 35 per unit including the charge. The government's revenue is a transfer, while the external damage is a real harm. Do not add both as independent social costs of the same unit.
Standards, liability, tradable permits, public provision, and negotiated rules are other possibilities. Policy choice depends on information, enforcement, uncertainty, measurement costs, and distribution. If damages vary by location and time, one uniform charge may be a poor match. If bargaining is costly or affected people are numerous, a theoretical bargain need not occur. Assigning rights also affects who pays and who receives compensation.
Compare a charge, a standard, and an enforceable bargain
Consider another paper case with three possible units of output. Their buyer benefits are 60, 42, and 28 currency units, respectively. Each uses 20 of private resources and causes 15 of external damage. Assume separate buyers, no fixed costs, competitive provision, and no other effects. The first unit creates a social gain of 25, the second 7, and the third a loss of 7. Producing two units gives total benefit 102 and total social cost 70, so the modeled net gain is 32. Producing all three gives benefit 130 against cost 105, leaving only 25.
Without a charge, each unit's benefit exceeds private cost 20, so all three can be privately worthwhile. A charge of 15 per unit makes the relevant private threshold 35. The first two buyers are willing to cover that amount; the third is not. Under these assumptions, two units remain. Their production uses 40 of resources, causes 30 of harm, and generates tax revenue of 30. The revenue is not a further 30 of damage. It changes who has spending power while the resource and environmental costs still total 70.
Compare distribution with efficiency. If the transaction price is 35, buyers retain benefits of 25 and 7 above their payments, while sellers just cover their private costs and the charge. The authority receives 30 and the affected outsiders still suffer harm valued at 30. Returning revenue to those outsiders would change their financial position, but would not undo the physical harm. Whether money is adequate compensation is a further question about rights, health, and the valuation used. The charge does not itself guarantee a compensation rule.
A charge changes output, not the cost definition
The invented three-unit case yields less net gain than producing two units: 25 rather than 32. Each allocation divides total buyer benefit into resources used, external harm, and net gain. The charge induces the two-unit choice under the stated behavior; its revenue is a transfer, not another resource cost or automatic compensation.
A quantity standard allowing exactly two units could reproduce this quantity if enforcement is perfect. It need not reproduce who produces them or who receives any scarcity income. Suppose two otherwise identical plants can eliminate one discharge at costs of 4 and 13. Requiring the expensive plant alone to eliminate a discharge uses 9 more resources than requiring the cheaper one. A common emissions price can encourage the cheaper reduction first when plants know their own costs, but only if emissions and compliance can actually be measured.
A negotiated agreement faces a different budget problem. Suppose a grower would gain 11 from a filter that costs the workshop 8. There is a possible joint gain of 3. A payment of 9 from grower to workshop leaves the grower 2 better off and the workshop 1 better off. If arranging and enforcing the agreement consumes 4 of resources, that possible gain disappears. Many affected people may also wait for others to pay. The existence of a beneficial agreement on paper is not evidence that actual bargaining will reach it.
Legal rights change the starting point for such bargaining. A workshop required to avoid harm may have to fund prevention itself; a grower without that protection may have to offer payment. Those arrangements distribute income and leverage differently even if a simplified model predicts the same physical result without transaction costs. Comparing instruments therefore requires their information requirements, administrative resources, and distribution of claims, rather than assuming that every instrument delivering two units is equivalent.
Work through the mechanism
Add the omitted marginal harm
20 is the extra private resource cost and 15 the external damage of one additional unit. A tax transfer is not an additional physical damage.
One additional unit
Amount
Buyer benefit under the chosen measure
30
Private resource cost
20
External damage
15
Social marginal cost
35
Net social gain
−5
The privately attractive unit is socially costly under these stated values. Distributional weights and unmeasured effects could change the evaluation.
Read the marginal comparison before adding totals
The displayed equation and table describe one additional unit, not a whole industry's output. Read each row with its unit attached. The buyer benefit is 30 currency units for the extra item. The private resource cost is 20 for that item. External harm is 15 for that item. Adding the last two gives social marginal cost 35. Subtracting 35 from benefit 30 gives negative 5. Nothing in the table says that all earlier units had that same benefit or that every unit should be prohibited.
To see the distinction, use a paper extension with four successive units whose benefits are 55, 45, 30, and 22. Retain the displayed private cost 20 and harm 15 for each unit. Their individual net social gains are 20, 10, negative 5, and negative 13. The first two together yield 30. Adding the third reduces that total to 25, even though total benefit continues to increase. Adding the fourth reduces it to 12. Maximum total net benefit and maximum output are different objectives in this constructed case.
Positive private margins can hide social losses
For the paper extension, subtract the same social cost of 35 from each successive benefit: 55, 45, 30, and 22. The last two additions reduce aggregate net benefit even though buyers value them above private cost 20. These bars concern marginal gains without filtration; different harm or prevention assumptions change the comparison.
Private production without liability could extend through the fourth unit because even benefit 22 exceeds private cost 20. A correctly enforced charge of 15 makes the third and fourth units unattractive at a cost threshold of 35. The displayed material contains no control for adding these demand values or choosing this output. Work through them on paper beside the existing single-unit table. There is no externality slider, pollution animation, or policy simulator on this page, and the numerical extension does not change the table's assumptions.
Also compare prevention with simply reducing production. Suppose a filter removes all the assumed harm at a resource cost of 8 per item while leaving production cost 20 unchanged. The cost of filtered output becomes 28. In the four-unit extension, the first three benefits exceed 28 but the fourth does not. Their gains are 27, 17, and 2, totaling 46. The filter supports more useful output than the no-filter choice of two units, whose gain was 30. This conclusion depends on the assumed filter performance and cost, not on revenue raised by a tax.
If that filter actually costs 18 per item, filtered production costs 38 while unfiltered social cost remains 35. For the same unit and harm estimate, filtration uses more resources than the harm it avoids. Requiring it across the board can then sacrifice net benefits under the stated criterion. That calculation is not a verdict about every safety standard. A real standard may address uncertain severe damage, rights that cannot be traded away, or harms poorly represented by the chosen money measure. State those reasons separately instead of altering the arithmetic.
Finally, distinguish the sums from their inputs. Once benefit and cost figures are specified, subtraction is accounting within the example. Predicting that a charge changes output requires a behavioral model of buyers, producers, and enforcement. Estimating the actual health benefit requires causal evidence. Deciding how to weight a child's health against another person's consumption requires an explicit evaluative judgment. The table makes one comparison transparent; it does not supply those missing measurements or judgments.
Prevention changes which units are worthwhile
Currency units; benefits 55, 45, 30, and 22
Method and cost per item
Beneficial units
Total net gain
No filter: 35
2
30
Filter costs 8: total 28
3
46
Filter costs 18: total 38
2
24
At social cost 35, benefits 55 and 45 justify two items, giving gains 20 and 10. A filter that lowers cost to 28 also makes the third item worthwhile: gains 27, 17, and 2 total 46. Requiring the expensive filter raises cost to 38 and leaves gains 17 and 7. These conditional comparisons count resources and harm, not tax receipts as an extra cost.
Stocks and collective action complicate the problem
Climate change depends on accumulated emissions and long-lived physical effects, not only today's flow. Policies must consider timing, technology, distribution, and cross-border coordination. A carbon price is one instrument, not a complete model of energy infrastructure, research, or a just transition. Competing assessments can differ because they use different damage estimates or value future and unequal impacts differently.
Common-pool resources such as fisheries are difficult to exclude users from and can be depleted by use. Elinor Ostrom's research describes settings where communities developed monitoring and rules rather than relying solely on either unrestricted markets or centralized state control. This is evidence against a universal two-option account, not a guarantee that every local arrangement succeeds. Institutions have to fit the resource and enforcement problem.
Govern a resource that persists across periods
A stock is an amount at a date; a flow is an amount added or removed during a period. This distinction changes environmental reasoning. In a paper fishery, start a season with 1,000 adult fish. Suppose reproduction and natural survival together add a net 200 before harvesting. A catch of 300 leaves 900 at the season's end. Catch of 150 leaves 1,050. These are stock reconciliations under an assumed biological addition, not predictions that fish populations always add 200 regardless of their size or environmental conditions.
Each fisher may receive the sale proceeds of another catch while sharing its effect on future availability with everyone using the lake. Suppose an extra catch earns 80 after fuel and labor costs but reduces the value of other users' future catches by 100. The fisher's private gain and the group's loss can coexist. Discounting future losses or imposing access rules changes the comparison. Unrestricted access, member-managed access, and private exclusive rights are different institutions; a resource with difficult exclusion is not automatically an ungovernable resource.
Monitoring uses resources too. Suppose ten users propose a seasonal rule that avoids total harm valued at 600, but reduces their combined current fishing gains by 400. Its gross net benefit is 200. If a workable monitoring system costs 120, the remaining net benefit is 80. Equal dues of 12 per user fund that monitoring, but equal dues do not imply equal individual gains. If two users give up most of the profitable catch, they may oppose the rule even though its combined gain is positive.
Monitoring consumes part of the governance gain
In the invented ten-user fishery, avoiding harm worth 600 requires giving up current fishing gains of 400. That leaves 200 before monitoring and 80 after monitoring costs 120. Each row allocates the same avoided harm. The remaining aggregate gain does not show each user's outcome or include people excluded from membership.
The budget and authority must therefore be specified. Who records catches, can inspect boats, settles disputes, and imposes sanctions? Who can change membership or the seasonal limit? Users excluded from the decision may lose access without appearing among the ten members' benefits. Ostrom's work examines arrangements with rules and monitoring suited to particular resources. The lesson is not that informal trust always solves depletion. It is that actual governance arrangements deserve comparison with both unrestricted use and centralized regulation.
Persistent pollutants raise a related stock problem without behaving exactly like fish. Suppose an illustrative pollution stock begins at 500 tonnes, 10% is removed naturally during the year, and new emissions add 80 tonnes afterward. Removal is 50 tonnes, leaving 450 before additions and 530 afterward. Cutting emissions to 50 would stabilize this year's stock at 500 under those assumptions. Cutting emissions by 20%, to 64, would still leave 514. A smaller emissions flow is not necessarily a falling accumulated stock.
The physical calculation says nothing yet about which producers should reduce emissions or how their costs should be shared. Firms with low-cost alternatives may reduce more under a common price, while communities with high exposure may need location-specific protections. Future technology, damage uncertainty, and cross-border enforcement alter the feasible choices. Evaluate an institution by the decisions it can implement, the real resources it consumes, and the people it includes. A balanced public budget or profitable firm can coexist with depletion if important stock changes remain outside its accounts.
Exercises
Do not count revenue as damage
A unit creates 12 of external harm and is subject to a corrective tax of 12. Has the tax made the external harm 24?
Hint
Distinguish a transfer payment from a physical or welfare loss.
Worked answer
Item
Amount
Type
External harm
12
Social damage
Tax paid and received
12
Transfer
Units produced
May change
Behavioral response
Administrative costs would be a separate real cost, not the tax payment itself.
No. The harm remains 12 per produced unit under the assumption. Tax revenue transfers purchasing power to government. The tax may reduce how many harmful units are produced; it does not itself double the per-unit environmental damage.
A positive external effect
Training costs a firm 50 and benefits it by 40, but gives other firms and workers an additional benefit of 25. Is private rejection consistent with a positive modeled social return?
Hint
Sum benefits across affected parties before subtracting the resource cost.
Worked answer
Perspective
Benefit
Cost
Net
Firm only
40
50
−10
All stated parties
65
50
+15
The extra benefits are assumed and would require evidence in a real policy case.
Yes. The firm sees a private net loss of 10, while modeled social benefit is 65 against cost 50, giving a gain of 15. This suggests a coordination problem, not an automatic justification for any subsidy regardless of targeting and implementation.
Sources and further reading
The lessons and answers stand alone. These optional references identify underlying models, evidence, and perspectives. All numerical examples here are invented unless explicitly attributed; none is a forecast or a personal investment recommendation.