Supply chain · AI · Digital twins · Decisions under uncertainty

Is your R&D worth funding? Value it as an option

R&D does not commit you to a launch. It buys the right to decide later. Value it that way and better funding decisions follow.

· Dr. Jonathan D. Linton · 4 min read

Why does NPV undervalue R&D?

R&D should be valued as an option, not as a fixed investment: it buys the right, but not the obligation, to launch later, and net present value (NPV) leaves the value of that choice out. NPV works well for a new machine or a building, where the money buys an asset you will use. R&D is different. If the results disappoint or the market moves, you stop. If they are good, you go ahead. That ability to decide later, once more is known, is worth something, and NPV assumes it away.

The practical result is that many R&D cases look marginal or negative on NPV, especially long-term ones. Teams respond by stretching forecasts until the numbers work, and the debate turns into an argument about one forecast.

What is a better question to ask?

Turn the question around. Instead of forecasting the profits and checking whether they cover the spend, ask: what would the benefits have to be worth for this spend to make sense? That is a threshold, and leaders are usually better at judging whether a threshold is realistic than at producing a precise forecast.

Option pricing gives that threshold directly. The R&D spend is the price paid for the option. The cost of launching is what you would pay to exercise it. The time until the launch decision, the risk-free rate (the yield on a government bond of the same term) and the uncertainty in the value of the benefits, known as sigma, complete the picture. Solving the model in reverse gives the minimum value of benefits that justifies the spend.

Did it work on a real decision?

In 2006, Dr. David Goldenberg and Dr. Jonathan D. Linton applied this method to one of the largest R&D bets there is: nuclear fusion [1]. Planned research spending was about US$70 billion (2002 dollars), with commercial power expected around 2050. Building the plants was estimated at US$43.8 trillion in 2050 dollars.

The threshold came out at US$1.075 trillion in operating-cost savings, valued in 2002. Spread over projected world energy use, fusion would need to generate power about 2.4¢ per kWh cheaper than the alternatives. Across 14 scenarios for plant uptime, adoption and construction costs, the figure ranged from 0.6¢ to 2.5¢ per kWh, against an estimated fusion operating cost of about 1¢ per kWh. The research was financially justified across a wide range of reasonable assumptions, a clearer answer than any single 48-year forecast could give.

You can reproduce the case, and test your own numbers, with our free R&D investment threshold calculator.

Why does a higher sigma make R&D more valuable?

This is the result that surprises people most. In options theory, uncertainty is measured by sigma: the higher the sigma, the greater the risk and the greater the potential profit. A higher sigma lowers the threshold. Because you launch only if things turn out well, a wider range of outcomes raises the chance of a very good result, while the downside stays limited to what you spent on the R&D.

This is also why the choice of sigma matters. We favour using the sigma of stocks listed in the same industry. For nuclear fusion that meant electrical utilities, whose stocks have a low sigma (18% a year), which makes the estimate more conservative: the bar the research had to clear was set high, and it still cleared it.

That does not mean uncertainty is good in itself. It means that when you have the right to wait and decide, uncertainty is a reason to keep options open, through staged funding, pilots and stage gates, rather than a reason to reject the project outright.

What should decision-makers do?

  1. Separate the decision to explore from the decision to commit. Fund R&D as the price of an option, and make the launch a separate decision.
  2. Set the threshold first. Agree what the benefits would need to be worth before debating the forecast.
  3. Test a range, not a point. Check how the threshold moves with sigma, timing and launch cost.
  4. Revisit at each stage. As results come in, the uncertainty narrows and the threshold changes.

The same idea extends to staged commitments, such as whether a patent is worth filing [2], and to choosing between competing technologies [3]. These are the kinds of decisions we model in decision support under uncertainty.

References

  1. Goldenberg, D., & Linton, J. (2006). Nuclear fusion R&D. Energy Risk, January 2006, 2–5.
  2. Goldenberg, D. H., & Linton, J. D. (2012). The patent paradox – New insights through decision support using compound options. Technological Forecasting and Social Change, 79(1), 180–185. doi:10.1016/j.techfore.2011.08.017
  3. Goldenberg, D. H. (2010). Managing VC investments in competing technologies using exotic options and secondary markets. The Journal of Private Equity, 13(4), 86–103. doi:10.3905/jpe.2010.13.4.086

Questions people ask

How should R&D projects be valued?

As options. R&D buys the right, but not the obligation, to launch later if the results and the market are good. Option-pricing methods capture the value of that flexibility; a standard net present value assumes you will go ahead whatever happens and so tends to undervalue R&D.

What is a real option?

A real option is the right, but not the obligation, to take a business action later, such as launching a product, expanding a plant or abandoning a project. Its value comes from being able to wait until more is known and act only if the outcome is good.

What is an R&D investment threshold?

It is the minimum value of benefits, valued today, that an R&D programme must be expected to unlock for its cost to be justified. Comparing a realistic benefit estimate with the threshold is often easier and more honest than relying on a single profit forecast.