Jorge Quitério Search
Technology

Valuing R&D projects as real options

A discounted cash flow based on the median market can reject a project that is worth pursuing once its uncertainty and the right to stop are valued. For staged R&D, greater uncertainty increases value. A simulation decomposes the difference between the valuations.

Consider a research proposal submitted to an investment committee. The finance team builds a discounted cash flow on the expected market, the net present value is negative, and the project is rejected. A year later a competitor launches the product.

The analysis was not careless, but the model answered a different question from the one the committee faced. It valued a commitment: full investment and launch regardless of outcome. An R&D project works differently. It is a sequence of decisions, each taken with more information than the last, and the project can be stopped at every stage. This article examines the value of that right to stop and why uncertainty increases it.

Drug development is the standard example: a compound passes through laboratory studies and three phases of clinical trials, and at each stage the company can stop. Everyday decisions have the same structure. A buyer who pays for a structural survey before committing to a house is buying an option: a small certain cost for the right to walk away if the survey finds serious damage. The survey is worth more the less is known about the house, which is the central result of this article in its simplest form.

The term real options comes from Stewart Myers (1977), who observed that many corporate assets, growth opportunities above all, can be viewed as call options whose value depends on future discretionary investment. The financial theory underneath is Black and Scholes (1973). Dixit and Pindyck's Investment under Uncertainty (1994) and Trigeorgis' Real Options (1996) developed the idea for investment decisions in general. De Weck's roadmapping method puts a financial model, the technology's value, in every roadmap (section 8 in technology roadmaps as quantitative models), which is where this valuation belongs.

Three valuations of one project

The project in the figure has three stages: research (€2M, succeeds with probability p1p_1), development (€10M in year 1, succeeds with p2p_2), and a launch decision in year 5, when the market value VV is known and the launch investment KK is due. VV is uncertain: lognormal with median €60M and spread σ\sigma. With discount factor dtd_t:

  1. Planned on the median market: assume VV equals its median, NPVmed=−C1−p1C2d1+p1p2(Vmed−K) d5\text{NPV}_{\text{med}} = -C_1 - p_1 C_2 d_1 + p_1 p_2 (V_{\text{med}} - K)\, d_5.
  2. Committed to launch: use the mean of VV, which for a lognormal is Vmed eσ2/2V_{\text{med}}\,e^{\sigma^2/2}, larger than the median: NPVcommit=−C1−p1C2d1+p1p2(EV−K) d5\text{NPV}_{\text{commit}} = -C_1 - p_1 C_2 d_1 + p_1 p_2 (\mathbb{E}V - K)\, d_5.
  3. With the right to stop: launch only if V>KV > K, so the last stage pays E[max⁡(V−K,0)]\mathbb{E}[\max(V - K, 0)] instead of EV−K\mathbb{E}V - K.

The difference between 1 and 2 is the cost of planning with a single number, as discussed in the value of the stochastic solution. The difference between 2 and 3 is the option value.

The figure below illustrates the result: the right to stop is always worth something, and it is worth more the more uncertain the market.

  1. Default settings. With the defaults, the plan on the median market is negative (about −€1.8M). The committed value, using the mean, is positive; with the right to stop it rises further, to about €2.3M. The same project receives three different valuations.
  2. Low uncertainty. Set σ\sigma to 0.05. All three values converge: with no uncertainty there is nothing to learn, and an option to wait for information is worthless.
  3. High uncertainty. Set σ\sigma to 1. The option value grows, because the upside of a large market is retained while the right to stop truncates the downside. This is the central and counter-intuitive result of real options theory: greater uncertainty increases value, provided the decision can still be taken later.
  4. Distribution of outcomes. Even when the expected value is positive, most individual projects lose money (the bar at about −€2M is the research that failed). R&D portfolios depend on a few large successes; judging each project by its most likely outcome would reject all of them.

The closed form. If ln⁡V∼N(ln⁡Vmed,σ2)\ln V \sim N(\ln V_{\text{med}}, \sigma^2), then

E[max⁡(V−K,0)]=EV  Φ(d1)−K Φ(d2),d2=ln⁡(Vmed/K)σ,d1=d2+σ. \mathbb{E}[\max(V - K, 0)] = \mathbb{E}V\;\Phi(d_1) - K\,\Phi(d_2), \qquad d_2 = \frac{\ln(V_{\text{med}}/K)}{\sigma},\quad d_1 = d_2 + \sigma .

Proof: P(V>K)=P(Z>ln⁡(K/Vmed)/σ)=Φ(d2)P(V > K) = P\big(Z > \ln(K/V_{\text{med}})/\sigma\big) = \Phi(d_2), and E[V 1{V>K}]=Vmed eσ2/2 Φ(d2+σ)\mathbb{E}[V\,\mathbf 1\{V>K\}] = V_{\text{med}}\,e^{\sigma^2/2}\,\Phi(d_2 + \sigma) by completing the square in the normal integral. Because max⁡(V−K,0)≥V−K\max(V-K,0) \ge V - K for every VV, the option value can never be below the committed value; and because max⁡(V−K,0)\max(V-K,0) is convex in VV, spreading VV out (raising σ\sigma with the median fixed) raises its expectation.

None of this means every negative-NPV project should be funded. It means the valuation has to describe the decision actually being taken. For staged R&D that is a sequence of options, and the stage gates are where their value is collected, provided the organisation is willing to stop when the information says so. If no project is ever stopped, the option value in the model is not realised.

References

  1. F. Black and M. Scholes, "The pricing of options and corporate liabilities", Journal of Political Economy, 81(3), 637-654, 1973.
  2. S. C. Myers, "Determinants of corporate borrowing", Journal of Financial Economics, 5(2), 147-175, 1977.
  3. A. K. Dixit and R. S. Pindyck, Investment under Uncertainty, Princeton University Press, 1994.
  4. L. Trigeorgis, Real Options: Managerial Flexibility and Strategy in Resource Allocation, MIT Press, 1996.
  5. O. L. de Weck, Technology Roadmapping and Development: A Quantitative Approach to the Management of Technology, Springer, 2022.