Private Incentives to Conduct Research

Why a Firm Might Invest in Research

Research and development (R&D) is costly, and its outcome is uncertain. A firm must decide whether the profits it expects to earn if research succeeds are large enough, after accounting for the chance of failure and the cost of the research, to justify investing. This is a private decision: the firm compares its own expected returns with its own costs.

The earlier material on knowledge spillovers distinguishes private returns from social returns. A firm’s private payoff may omit benefits that other firms or consumers receive from new knowledge. The decision model here concerns the firm’s own expected payoff; it does not assume that private and social returns are equal.

Definition: Let $C$ be the research cost paid by the firm, $p$ the probability that the project succeeds, and $V$ the present value of the firm’s incremental private profit if it succeeds. Suppose failure produces no incremental profit and the firm is risk-neutral. The project’s expected net present value is $$ \operatorname{ENPV}=pV-C. $$ The firm should undertake the project if $\operatorname{ENPV}>0$, is indifferent if $\operatorname{ENPV}=0$, and should reject it if $\operatorname{ENPV}<0$.

This basic rule relies on specific assumptions: the research cost is paid whether the project succeeds or fails; the firm’s stated success probability is appropriate for the project; $V$ measures incremental profit rather than total company profit; and $V$ is already expressed as a present value. If failure has a salvage value, or the cost is contingent on success, the expected-value calculation must include that fact rather than applying the basic formula unchanged.

From Market Size to a Private Payoff

A larger potential market can increase the payoff from a successful innovation. To make that relationship precise, suppose a successful product earns an incremental contribution margin of $m$ dollars per customer per year, reaches $N$ customers each year, and does so for $T$ years. Assume the customer count and margin are constant, and let $r$ be the firm’s effective annual discount rate. If profits arrive at each year-end, the present value of successful innovation is

V=\sum_{t=1}^{T}\frac{Nm}{(1+r)^t} =Nm\sum_{t=1}^{T}\frac{1}{(1+r)^t}.

This calculation assumes the innovation earns the specified incremental profit in every year of the stated horizon, with no additional development cost after the initial research cost $C$. It also assumes that $N$ represents customers actually served, not the entire population that might theoretically be interested. If only a share $s$ of the market is reached, replace $N$ with $sN$.

The discounted sum matters because a dollar of profit received later is worth less to the firm than a dollar received now. The expression does not apply unchanged if margins, customer numbers, or the discount rate vary by year; in that case, discount each year’s expected incremental profit separately.

Worked Example: Testing a Project Against Its Research Cost

A firm can spend $C=\$170{,}000$ on a one-period research project. If it succeeds, the innovation will generate an incremental profit of $V=\$500{,}000$, measured at the decision date. The probability of success is $p=0.40$. Assume failure produces no incremental profit and the cost is paid upfront in either outcome. Should the firm invest?

Method. Calculate expected incremental profit and subtract the cost paid regardless of the outcome.

The expected incremental profit is $pV=0.40(\$500{,}000)=\$200{,}000$. Therefore, $\operatorname{ENPV}=\$200{,}000-\$170{,}000=\$30{,}000$. The project has positive expected net present value, so a risk-neutral firm should invest under the stated assumptions.

The break-even success probability follows from $pV-C=0$: $p=C/V=\$170{,}000/\$500{,}000=0.34$. The assumed probability of $0.40$ is above this threshold.

Check. At the break-even probability, expected profit is $0.34(\$500{,}000)=\$170{,}000$, exactly equal to the research cost. The actual probability exceeds that threshold by $0.06$, producing $0.06(\$500{,}000)=\$30{,}000$ of expected net value.

Common mistake avoided. The firm pays the research cost even if the project fails. Subtracting the cost only in the success case would incorrectly calculate $p(V-C)=0.40(\$330{,}000)=\$132{,}000$. The correct calculation is $pV-C=\$30{,}000$.

Why the Expected-Value Rule Holds

The investment rule follows from calculating the payoff in each possible outcome and weighting each payoff by its probability. In the basic two-outcome model, success has probability $p$, while failure has probability $1-p$. Because the firm pays $C$ in either outcome, its net payoff is $V-C$ upon success and $-C$ upon failure.

Write the outcome payoffs.
Success gives net payoff $V-C$; failure gives net payoff $-C$, because the upfront research cost is not recovered.

Weight each payoff by its probability.
Expected net payoff is $p(V-C)+(1-p)(-C)$.

Simplify.
$p(V-C)+(1-p)(-C)=pV-pC-C+pC=pV-C$.

Thus the cost is subtracted once, not multiplied by the probability of success. The same result can be obtained by taking expected incremental profit, $pV$, and subtracting the certain cost $C$. If a firm is not risk-neutral, expected net present value alone may not describe its choice: the firm’s willingness to bear risk can also matter.

Market Size, Discounting, and the Investment Threshold

The following example shows how market size can change a decision even when the success probability, margin, research cost, and discount rate stay fixed. It uses a two-year horizon with year-end profits, so the present value of successful profits depends on both years.

Worked Example: Does a Larger Market Make Research Worthwhile?

A hypothetical firm is considering a product that, if successful, earns a margin of $\$50$ per customer per year. It expects to reach 20% of the potential market in each of the next two years. The annual discount rate is $r=10\%$, and research costs $\$50{,}000$ upfront. The probability of success is $0.25$. Compare a potential market of 9,680 customers with one of 12,100 customers. Assume customer counts and margins remain constant, and failure yields no incremental profit.

Method. For each market, calculate the annual profit conditional on success, discount both years’ profits, multiply that present value by the success probability, and subtract the upfront research cost.

The discount factors for the two year-end payments are $1/1.10=10/11$ and $1/(1.10)^2=100/121$. Their sum is $110/121+100/121=210/121$.

For the 9,680-customer market, the firm expects to serve $0.20(9{,}680)=1{,}936$ customers. Conditional annual profit is $1{,}936(\$50)=\$96{,}800$. The successful-project present value is $\$96{,}800(210/121)=\$168{,}000$. Expected profit before research cost is $0.25(\$168{,}000)=\$42{,}000$. Thus, $\operatorname{ENPV}=\$42{,}000-\$50{,}000=-\$8{,}000$; the firm should reject this project under the assumptions.

For the 12,100-customer market, the firm expects to serve $0.20(12{,}100)=2{,}420$ customers. Conditional annual profit is $2{,}420(\$50)=\$121{,}000$. The successful-project present value is $\$121{,}000(210/121)=\$210{,}000$. Expected profit before research cost is $0.25(\$210{,}000)=\$52{,}500$. Therefore, $\operatorname{ENPV}=\$52{,}500-\$50{,}000=\$2{,}500$, so the larger market makes the project just worthwhile.

Check. In both cases, expected profit before the research cost equals one-quarter of the successful-project present value. The increase from $\$42{,}000$ to $\$52{,}500$ is $\$10{,}500$, which equals one-quarter of the $\$42{,}000$ increase in successful-project value.

Common mistake avoided. Discounting only one year’s profit would omit the second payment; using undiscounted two-year profit would overstate the project’s present value. The two year-end profits must be discounted separately and then added.

When $V>0$, the basic model gives a useful break-even threshold for success probability:

p^*=\frac{C}{V}.

A project has positive expected net present value when $p>p^*$, provided $p^*$ is a feasible probability. If $C>V$, then $p^*>1$, so even certain success would not cover the research cost in this model. If the firm holds the success probability and cost fixed, the minimum successful-project value is $V^*=C/p$. These thresholds make clear why larger markets can help: when a larger market raises the firm’s incremental profits, it raises $V$ and lowers the probability of success needed to justify research.

Choosing Research Effort

A firm may be able to spend more on research effort to improve the probability of success. In that case, it must compare the additional expected payoff from effort with its additional cost. Let $e$ denote the firm’s research effort, $p(e)$ the probability of success at that effort, and $C(e)$ the cost. If successful-project value $V$ is fixed, the firm chooses effort to maximize

\operatorname{ENPV}(e)=p(e)V-C(e).

For an interior optimum, where the best effort is neither the minimum nor maximum available, differentiable functions imply the first-order condition $Vp'(e)=C'(e)$. The left-hand side is the marginal expected benefit of effort: an increase in success probability of $p'(e)$ is multiplied by the value $V$ received upon success. The right-hand side is the marginal cost. A maximum also requires the objective to be concave locally; when $Vp''(e)-C''(e)<0$, the second derivative is negative. If the first-order solution lies outside the feasible range of effort, the firm must compare feasible choices, including boundary values.

Worked Example: Finding the Profit-Maximizing Research Effort

A hypothetical project has a successful innovation value of $\$100{,}000$. Research effort $e$ is between 0 and 1, the success probability is $p(e)=0.10+0.40e$, and the research cost is $C(e)=\$40{,}000e^2$. Assume the probability formula applies throughout the feasible range. Find the effort that maximizes expected net present value and calculate that value.

Method. Substitute the probability and cost functions into the expected-value objective, differentiate, check the second derivative, and verify the solution is feasible.

The objective is $\operatorname{ENPV}(e)=(0.10+0.40e)(\$100{,}000)-\$40{,}000e^2 =\$10{,}000+\$40{,}000e-\$40{,}000e^2$. Its first derivative is $\operatorname{ENPV}'(e)=\$40{,}000-\$80{,}000e$. Setting this equal to zero gives $e=40{,}000/80{,}000=0.50$, which is within the permitted interval $[0,1]$.

The second derivative is $\operatorname{ENPV}''(e)=-\$80{,}000<0$, so the objective is strictly concave and the interior solution is a maximum. At $e=0.50$, the probability of success is $0.10+0.40(0.50)=0.30$. The research cost is $\$40{,}000(0.50)^2=\$10{,}000$. Expected profit before research cost is $0.30(\$100{,}000)=\$30{,}000$, giving $\operatorname{ENPV}=\$30{,}000-\$10{,}000=\$20{,}000$.

Check. At zero effort, net value is $\$10{,}000$. At maximum effort $e=1$, success probability is $0.50$, cost is $\$40{,}000$, and net value is $\$10{,}000$. Both boundary values are below the $\$20{,}000$ value at $e=0.50$.

Common mistake avoided. Choosing the effort that maximizes success probability alone would select $e=1$. But effort also has a cost, so the firm maximizes $p(e)V-C(e)$, not $p(e)$ by itself.

Private Returns and Broader Economic Effects

The private investment rule gives a disciplined way to organize a firm’s decision, but its conclusion is limited to the firm’s own payoff and the assumptions in the model. A firm may fail to invest even when research would benefit other producers or consumers, because those gains do not necessarily enter its $V$. Conversely, a positive private expected return does not by itself establish that the project’s total social benefits exceed its social costs.

In applied problems, identify what the success payoff includes before calculating. It should be the incremental private profit attributable to the innovation, not gross sales if those sales require costs to produce. Keep the time horizon and discounting consistent, account for costs paid in both success and failure states, and distinguish a larger market from a higher probability of success. Each can raise expected value, but through a different part of the calculation.

Key takeaway: In the basic risk-neutral model, research is privately worthwhile when the probability-weighted present value of incremental profit exceeds the research cost. Market size can raise the payoff from success, while research effort is worthwhile only when its marginal expected benefit justifies its marginal cost.

Check Your Understanding

For each problem, assume the stated research cost is paid whether the project succeeds or fails, unless the problem says otherwise.

A research project costs $\$65{,}000$. Success yields incremental profit worth $\$240{,}000$ at the decision date, and failure yields zero incremental profit. Find the break-even success probability and the expected net present value if the success probability is $0.30$. (Answer: break-even probability $0.2708$; expected net present value $\$7{,}000$.)
A product would earn $\$60$ per customer per year for two years if successful. The firm expects to serve 20% of a market of 5,000 customers, with year-end profits and a 10% annual discount rate. If success probability is $0.40$ and research costs $\$45{,}000$, should the firm invest? (Answer: successful-project present value $\$104{,}132.23$; expected net present value $-\$3{,}347.11$; reject.)
A project has $V=\$150{,}000$, $p(e)=0.05+0.50e$, and $C(e)=\$30{,}000e^2$, for $0\leq e\leq1$. Find the effort that maximizes expected net present value and that maximum value. (Answer: $e=1$; maximum expected net present value $\$52{,}500$.)
A firm’s research cost is $\$96{,}000$, and successful innovation would yield an incremental present value of $\$320{,}000$. If the estimated success probability is $0.28$, calculate the expected net present value and state whether the project passes the private-investment test. (Answer: $-\$6{,}400$; reject under the risk-neutral model.)

What You'll Learn