Knowledge Spillovers and Social Returns

Why Knowledge Can Benefit More Than Its Creator

A firm that develops a production method may improve its own output, but the method can also help other firms. Workers may carry useful experience to new employers, suppliers may learn from a customer’s process, or other producers may adapt an idea they observe. When those other firms benefit without fully compensating the creator, part of the return is a knowledge spillover.

The distinction matters for growth analysis because the creator’s decision is based on the benefits it can capture, while the total economic gain includes benefits received elsewhere. A positive spillover can therefore make the private return smaller than the social return. This does not mean every knowledge investment is worthwhile: its full social benefit still has to be compared with its opportunity cost, and the spillover may be small, uncertain, or costly to transmit.

Definition: A knowledge spillover is a benefit or cost that a knowledge-creating activity imposes on other economic agents and that is not fully reflected in a payment to or from the creator. The private return is the net benefit received by the decision-maker. The social return is the net benefit to the decision-maker plus the net effects on all others, measured using the same valuation basis.

Knowledge is nonrival in the sense that one firm’s use of an idea need not prevent another firm from using it. That property can make spillovers possible, but it does not guarantee that knowledge will spread or be useful to every recipient. A recipient may need skilled workers, equipment, or time to apply an idea. The economic question is whether the activity creates benefits for others that the decision-maker does not count.

Tracing a Spillover Through Production

To measure a spillover, separate the effect on the creator from the effect on other firms. Consider a simple one-period production setting with three firms. Firm $j$'s output, measured in dollars, is

y_j=100+4e_j+2H,\qquad H=e_1+e_2+e_3,

where $e_j$ is firm $j$'s knowledge-creating effort and $H$ is the economy-wide knowledge stock relevant to these firms. The equation is a local illustrative model: each firm’s own effort has a direct effect of 4 dollars per unit and an additional effect through the common knowledge stock of 2 dollars per unit. Assume a one-unit increase in one firm’s effort costs that firm 8 dollars, and that other firms make no simultaneous adjustment. These are one-period dollar amounts, so no discounting is required.

Worked Example: The Output Gain From One Firm’s Knowledge Effort

Firm 1 increases $e_1$ by one unit, while $e_2$ and $e_3$ stay fixed. Find the output gain for each firm, the private gross return, the social gross return, and the net return to the economy after the effort cost.

Method and assumptions. Compare output before and after the change, holding all other inputs and firms’ behavior fixed. A unit increase in $e_1$ raises both $e_1$ and the common stock $H$ by one unit.

Firm 1. Its direct term $4e_1$ rises by $4$, and its common-stock term $2H$ rises by $2$. Thus $ \Delta y_1=4+2=6 $ dollars. This is the private gross return in this model.

Firms 2 and 3. Neither firm’s own effort changes. For each, the common-stock term rises by $2(1)=2$ dollars, so $ \Delta y_2=2 $ and $ \Delta y_3=2 $.

Social gross return and net return. Add the output changes across all three firms: $ 6+2+2=10 $ dollars. Subtract the 8-dollar opportunity cost of the effort: the social net gain is $ 10-8=2 $ dollars. Firm 1’s private net gain is $6-8=-2$ dollars.

Check and common mistake. The total output change is $10$, which is also $4+2+2+2$: the direct effect on firm 1, its common-stock effect, and the common-stock effects on the other two firms. Counting only firm 1’s $6$-dollar gain and calling it the social return omits the $4$ dollars received by the other firms. Conversely, counting the 8-dollar cost as a cost to every firm would multiply the cost incorrectly; in this example, firm 1 bears it once.

This calculation illustrates the accounting principle: sum the incremental benefits and costs across affected agents. A benefit transferred from one agent to another should not be counted twice as a net social gain. In the example, the other firms’ output gains are additional production, not payments taken from firm 1.

Why the Social Return Exceeds the Private Return

The same logic can be stated with marginal products. Suppose a small increase in firm $i$'s knowledge effort changes output in every firm $j$. The social marginal product is the sum of the output changes across firms; the private marginal product includes only the change in the creator’s own output. If there are no offsetting external costs, their difference is the marginal spillover received by others.

\frac{\partial y_i}{\partial e_i} \quad\text{(private marginal product)},\qquad \sum_j \frac{\partial y_j}{\partial e_i} \quad\text{(social marginal product)}.

The sum includes the creator’s own term. It is not the creator’s term plus the sum over all firms including the creator a second time.

A short derivation makes the distinction precise. Let total output be $Y=\sum_j y_j$. Differentiating total output with respect to firm $i$'s effort gives

\frac{\partial Y}{\partial e_i} =\frac{\partial}{\partial e_i}\left(\sum_j y_j\right) =\sum_j\frac{\partial y_j}{\partial e_i} =\frac{\partial y_i}{\partial e_i} +\sum_{j\ne i}\frac{\partial y_j}{\partial e_i}.

The final sum is the marginal effect on other firms. If it is positive, and the creator cannot capture those gains through a payment or other arrangement, the social marginal benefit exceeds the private marginal benefit. A firm that chooses effort by comparing its own marginal benefit with its own marginal cost may stop while additional effort still has a positive social net benefit.

This argument depends on the model’s perspective and boundaries. If knowledge creation also causes congestion, duplication, or other external costs, those effects belong in the social calculation too. If recipients pay the creator for the benefit, the payment affects the creator’s private return, but the payment itself is generally a transfer within the economy rather than an additional social benefit. The underlying gains and resource costs still need to be counted once.

Present Values: Keeping Private and Social Benefits Separate

Knowledge investments can have benefits at different dates. To compare them, value all benefits and costs at a common date using a stated discount rate. For a benefit $B_t$ received at the end of year $t$, its present value at effective annual rate $i$ is $B_t/(1+i)^t$. The social present value includes the private benefits and external benefits, each counted once.

Worked Example: Private and Social Net Present Value

A hypothetical process-improvement project costs $ \$2{,}500 $ today. An appraisal has already converted its benefits to present values using an effective annual discount rate of $8\%$. It estimates a private present-value gain to the adopting firm of $ \$2{,}045.351474 $ and a present-value spillover gain to other firms of $ \$743.764172 $. Calculate the private and social net present values. Treat the two benefit amounts as already discounted; do not discount them again.

Method. Subtract the initial cost from the present value of the benefits received by the relevant decision-maker. For the social calculation, add the private and external benefits before subtracting the project’s resource cost.

Private NPV. The firm captures the private gain, so $ \$2{,}045.351474-\$2{,}500=-\$454.648526 $, which rounds to $-\$454.65$.

Social present value of gains. Add the unrounded present values: $ \$2{,}045.351474+\$743.764172=\$2{,}789.115646 $. Rounded to the nearest cent, this is $ \$2{,}789.12 $.

Social NPV. Subtract the initial resource cost from the unrounded total: $ \$2{,}789.115646-\$2{,}500=\$289.115646 $, which rounds to $ \$289.12 $. The project has a negative private NPV but a positive social NPV under these estimates.

Check and common mistake. The rounded private and spillover components are $ \$2{,}045.35 $ and $ \$743.76 $, whose displayed sum is $ \$2{,}789.11 $. That sum is not the correctly rounded total: rounding the unrounded sum gives $ \$2{,}789.12 $. Keep full precision until the final step. Also, using the $8\%$ rate a second time would be wrong because the supplied benefit amounts are already present values.

A positive social NPV indicates that the estimated total benefits exceed the estimated resource cost; it does not, by itself, say who receives the gains or how they should be distributed. The negative private NPV indicates why the firm might decline the project if it considers only its own benefits. This comparison is conditional on the stated estimates and discount rate, not a claim about any actual company or industry.

When Positive Spillovers Lead to Underinvestment

A positive external benefit alone does not prove that every market outcome has too little investment. The standard underinvestment conclusion requires a decision rule and additional conditions: the creator chooses effort to maximize its own net benefit; it cannot capture all external gains; marginal private and social benefits decline with effort; marginal costs are the same in the private and social calculations; and the relevant optimum is interior. Under these conditions, the private choice equates private marginal benefit and marginal cost, while the social choice equates social marginal benefit and marginal cost.

The following static example makes the comparison explicit. Effort $e$ is nonnegative and measured in units; benefits and costs are measured in dollars for one period. There is no discounting because all amounts occur in that period. Let the firm’s private gross benefit be $B_P(e)=30e-e^2$, the external benefit to other firms be $B_E(e)=12e-0.5e^2$, and the resource cost be $C(e)=10e+0.5e^2$. The marginal private benefit is $30-2e$, the marginal external benefit is $12-e$, and marginal cost is $10+e$. Both marginal benefits decline over the relevant range, and marginal cost rises.

Worked Example: Comparing Private and Socially Efficient Effort

Find the firm’s privately chosen effort, the socially efficient effort, and the social surplus at each choice. State whether the private choice underinvests in this model.

Private choice. The firm maximizes private net benefit $B_P(e)-C(e)$. Its first-order condition sets private marginal benefit equal to marginal cost: $30-2e=10+e$. Therefore $20=3e$, so $e_P=20/3$, approximately $6.6667$ units. The second derivative of private net benefit is $-3$, so this is a maximum.

Social choice. Total social benefit is $B_P(e)+B_E(e)$, so social net benefit is $42e-1.5e^2-(10e+0.5e^2)=32e-2e^2$. Its derivative is $32-4e$. Setting it equal to zero gives $e_S=8$ units; the second derivative is $-4$, confirming a maximum. Equivalently, social marginal benefit is $(30-2e)+(12-e)=42-3e$, and $42-3e=10+e$ also gives $e=8$.

Compare the choices. The marginal external benefit is $12-e$. At the private choice it is $12-20/3=16/3$, which is positive. Thus an additional unit at the private choice still creates an external gain not included in the firm’s decision. The socially efficient effort exceeds private effort by $8-20/3=4/3$ units.

Check using social surplus. At $e=8$, social net benefit is $32(8)-2(8^2)=256-128=\$128$. At $e=20/3$, it is $32(20/3)-2(400/9)=640/3-800/9=1{,}120/9$, or about $\$124.4444$. The difference is $128-1{,}120/9=32/9$, or about $\$3.5556$. The social surplus is higher at the social choice, as required.

Common mistake avoided. Equating private marginal benefit to social marginal cost, or comparing total benefits without subtracting costs, mixes different decision rules. The private condition uses private marginal benefit and marginal cost; the social condition includes the external marginal benefit and the same resource marginal cost. The computed choices and surplus comparison are consistent with that distinction.

The example establishes underinvestment only within its assumptions. If effort creates a negative externality, the social marginal benefit can be below the private marginal benefit. If recipients compensate the creator, the private incentive may reflect more of the social gain. If benefits or costs are uncertain, the relevant comparison must also account for that uncertainty. The core diagnostic remains: identify which effects the decision-maker includes, then compare its private margin with the full social margin.

Common Interpretive Traps

Calling every benefit a spillover. A gain received by the creator is private unless it reaches other agents. State who benefits and whether the creator is compensated.
Adding only the creator’s output. Social returns require summing the incremental effects on all affected agents, not just the firm that made the investment.
Double-counting. Count each output gain or resource cost once. A payment between firms is not automatically a new social benefit on top of the underlying benefit.
Assuming the social return is always higher. It is higher when the net external effect is positive. External costs can reduce or reverse the gap.
Confusing a return with a growth rate. A positive social return on an additional knowledge investment does not, by itself, establish a particular long-run growth rate. That requires a specified growth model.

Key takeaway: A knowledge creator compares its private marginal benefit with its private marginal cost. Society compares the sum of marginal benefits across affected agents with the full resource cost. When unpriced external marginal benefits are positive and the other assumptions for an interior choice hold, private effort is below the socially efficient level.

Check Your Understanding

For each problem, distinguish effects on the decision-maker from effects on others, and keep units and valuation dates consistent.

A two-firm production model has $y_1=50+3e_1+e_2+2H$, $y_2=40+2e_2+H$, and $H=e_1+e_2$. Firm 1 increases $e_1$ by one unit, with no other changes. Find each firm’s output change and the total output change. (Answer: $\Delta y_1=\$5$, $\Delta y_2=\$1$, total $=\$6$.)
A one-period knowledge activity costs its creator $\$14$, raises the creator’s output by $\$11$, and raises two other firms’ output by $\$3$ each. Find the creator’s private net return and the social net return. (Answer: private net return $=-\$3$; social net return $=\$3$.)
A project costs $\$1{,}800$ today. At a 5% effective annual rate, its private benefit is $\$1{,}050$ at the end of year 1 and $\$900$ at the end of year 2; its external benefit is $\$300$ at the end of year 2. Find private and social NPV, rounding to the nearest cent. (Answer: private NPV $=\$1{,}050/1.05+\$900/1.05^2-\$1{,}800=-\$35.37$; social NPV $=\$1{,}050/1.05+\$1{,}200/1.05^2-\$1{,}800=\$236.84$.)
In a static model, $B_P(e)=24e-e^2$, $B_E(e)=8e-0.5e^2$, and $C(e)=6e+0.5e^2$, with all amounts in dollars and $e\geq0$. Find private and socially efficient effort. (Answer: private condition $24-2e=6+e$ gives $e_P=6$; social condition $32-3e=6+e$ gives $e_S=6.5$ units.)
For the functions in question 4, calculate social net benefit at the private and social effort choices and the difference. (Answer: $B_P+B_E-C=26e-2e^2$; at $e=6$, social net benefit is $\$84$; at $e=6.5$, it is $\$84.50$; the social choice raises net benefit by $\$0.50$.)

What You'll Learn