Valuing technical projects is key to accurate portfolio management, stage-gate decisions, and resource allocation issues. The methods vary according to whether a project falls into Category 1 Incremental, Category 2 Next-Generation, or Category 3 Radical and Breakthrough Innovation.

For Horizon 1 Incremental projects discounted cash flow (DCF) is a quick and easy way to calculate the value of a project. This is because for incremental projects the market is well understood and projected sales and costs to develop a product can be estimated with high accuracy. Timing of the costs and timing of the sales volumes can also reliably be estimated. For such projects there is little need to add more complexity to the evaluation process.

For Horizon 2 Next-Generation projects discounted cash flow does not work well as a method for evaluating projects. This is because of the increased uncertainty of both the costs that will be involved and the resulting revenue streams that are anticipated. Additionally the timing of both the cost and revenue streams is uncertain. For the complexity is that more than one technical approach is likely to be an option for next-generation projects so the probabilities of using one approach or the other has to be taken into consideration. Early work done by the strategic decisions group took into account these factors by utilizing decision trees to mimic the technical choices and options that a project might utilize as well as the sales revenues and pricing structures that might occur after launch. Using decision trees with conditional probabilities allowed uncertain events and a project to be summed to give an overall estimate of a project’s ROI or total value to the Corporation. Because alternative strategies are possible for each project choosing the best one requires finding the “efficient frontier”. That is a strategy that yields the best value for a given investment level. The methods SDG uses are outlined in the book Decision Quality by Carl Spetzler et al.

This form of evaluating R&D projects now takes the form of risk analysis. Risk analysis applies analytical tools to identify, describe, quantify, and explain uncertainty and its consequences for R&D projects. When choosing between competing alternatives decision trees are used. When just quantifying the risk or the uncertainty the tool of choice is Monte Carlo simulations.

An Example of a Simple Decision Tree

A decision tree is a visual model consisting of nodes and branches such as shown in the “An Example of a Simple Decision Tree” figure. Usually grows from left to right beginning with the root decision to start a project and the branches represent two or more competing options available to the project leaders. At the end of the branches there is either a value mode or in uncertainty mode representing the possible outcomes along with their respective probabilities. For next-generation R&D projects these models can become quite complex. The advantage however is that software is now been developed to make this process easy to implement.

The caution to this approach is one of “garbage in, then garbage out”. It is difficult for project teams and project leaders to expend an appropriate level of energy assigning probabilities and values to the decision trees. Sometimes the estimates are thoughtless and error-prone. Other times project teams spend more time developing the models than they could’ve spent in doing the actual experiments themselves. One approach to resolving this dilemma is to specify during the stage gate process for any stage, the amount of time the project team should spend on the model. The project team should then share at the gate review which of the nodes are the most uncertain. Using a sensitivity analysis the project team can then decide which of the uncertainties they have are worthy of further work in assigning more robust values.

For companies with a large number of projects going down the stage-gate process David Walwyn and his colleagues argued in How To Manage Risk Better that a Jarrow and Turnbull model for credit risky bond pricing can be of applied to R&D projects. Their method works out well because stage and gate processes involve evaluation that takes place set points. There typically a fixed number of hurdles and for large organizations, they average out to be equally spaced in time. The projects’ progress can then be considered as a discrete-time process. The project reviews are considered as “paid” or “default” in a bond model. For companies with good historic data, this offers a comparative approach to projects useful for R&D management. Its weakness is that it lacks the rigor that some CFOs, general managers, and executive teams desire.

Moving now to the evaluation of Horizon 3 Radical and Breakthrough projects, the methods become more qualitative and less quantitative. Attempts to keep the valuation methods quantitative involve the use of Real Options. The reason Real Options are used is because they have the ability to quantify: (1) a right but not an obligation to use the technology, (2) account for flexibility over the way in which the product is developed and marketed, (3) allow for choosing when and how often during the development process the project might continue, (4) the uncertain cost of expenditures is a project continues, and the (5) uncertain value of the project as it is commercialized.

Real Options can capture the value of managerial operating flexibility. Real Options can occur naturally, may be built-in or acquired a cost. Examples, where a firm may acquire Real Options, comes from intellectual property rights (such as patents licenses leases etc.), ownership of natural resources (such as real estate and oil fields). technological know-how, reputation, brand name, strategic alliances, market position, organizational capabilities, infrastructure, employees, etc. Some common behaviors that are accounted for by Real Options are the ability to defer, stage or abandon, expansion or growth, shutdown or restart, switching, multiple options, interacting options. With such inputs, Real Options can answer such questions as: (1) how valuable is a right to invest at a future time if the circumstances are right, (2) how valuable is it to have resources deployed elsewhere until circumstances to develop the current project are appropriate, (3) to defer development until prices sufficiently high, or (4) move to a new business model. Another advantage of Real Options is that it captures value not only from direct cash flows from the product developed but also from additional cross-selling opportunities that may open up after the initial product is launched. This includes the ability to expand into new products, reach new customers, and find new partners. Real Options also takes into account the asymmetry in probability distributions for any of the events or choices that a company might make. As such, the evolution of real option valuation models has increased in sophistication from static or mechanistic models to controllable cash flow models, to dynamic game-theoretic or industry equilibrium models.
Because of the complexity of the Real Options process, the cost of using it are extremely high and it suffers just as all valuation methodologies do from “garbage in, garbage out”. As such, although in theory Real Options provides R&D management with a very solid value of R&D programs, in practice this method is rarely used. More common is to use a sensitivity analysis first on the important variables affecting a project’s cost and value, then utilize a decision tree to assess likely outcomes.