Quantifying the Financial Value of R&D
Executive Summary
Art and serendipity play important roles in R&D but shouldn't dominate investment decisions. Unfortunately, they often do.
Art and serendipity play important roles in R&D but shouldn't dominate investment decisions. Unfortunately, they often do.
by Nikolas Vrettos and Michael Steiner
- With the increasing costs of R&D and of productivity requirements, it is ever more important to quantify the risks and rewards of individual R&D projects so that managers can make effective choices among them.
- The most common method for project valuation—discounted cash flow (DCF)—fails to allow R&D management to make project-changing decisions throughout the life of the program and therefore can greatly overstate a project's risk.
- The decision-tree-analysis method, although coming closer than DCF to the real value of a project, does not take into account the potential value of deferring action.
- An option-pricing model, although harder for an organization to accept because of its newness, yields more accurate results, particularly at the earliest stages of a project's life.
"Trying to assess basic research by its practicality is like trying to judge Mozart by how much money the Salzburg Festival brings in each year."
This observation by Austrian naturalist Konrad Lorenz de-scribes the guiding philosophy of most R&D organizations. Their results have depended more on genius, intuition, serendipity, and insight than on aggressive management of resources and focused decision-making.
The problem is that such a philosophy is increasingly unaffordable. Not only has R&D spending increased by 15% annually over the last 15 years, but the average R&D-to-sales ratio has also increased, from 11% in the ‘70s and mid-‘80s to 15% in the ‘90s. And it is still rising. The total cost of researching and developing new drugs has recently been estimated at about $590 million—much higher than the approximately $300 million estimated for the ‘70s and mid-'80s. At the same time, applications for NCEs (new chemical entities) and NMEs (new medical entities)have decreased in proportion to the increase in R&D expenses, reducing productivity.
There are, of course, specific reasons for this cost explosion. Research is becoming more complex, technologies are evolving much faster, and new drugs attack not the symptoms but the causes of diseases. At the same time, managers are demanding more information before they make decisions, requiring more clinical trials and tests, and setting higher standards for innovation and hence approval. When the cost consciousness of the various health authorities around the world is added, it becomes clear that this industry, like many others, must make its R&D investment far more productive than it has historically been. Recent statistics have indicated that, with current development practices, sales for many products will not be sufficient to both cover expenses and yield an acceptable return on investment.
The Evaluation Dilemma
All these elements together suggest the need for comprehensive management of the R&D portfolio. By focusing its portfolio strategy, a company hopes to tip the risk-reward balance in its favor. Doing so is not an easy task, however, because there is widespread aversion to evaluating R&D, especially research. This aversion stems from R&D's inherent characteristics—its intangibility, discontinuity of output, and uncertainty—and the ingrained belief that research is a serendipitous activity.
Only a few companies use state-of-the-art methods to evaluate and quantify project values. Recent surveys and our experience suggest that the sophistication of evaluation methods varies considerably between companies and across the pipeline.
We've found that about 75% of companies make research decisions subjectively, and only 25% use processes with modeling features. In development, however, modeling becomes the key decision support, putting more emphasis on financial considerations. That is understandable because the closer a project is to completion, the more tangible its output becomes.
Nonetheless, even for evaluating development, where modeling does support decisions, companies rarely use sophisticated financial analysis tools with risk-adjusted parameters and sensitivity variations; as a result, go-no-go decisions are often based on deterministic, single-outcome models. Finally, most companies don't use a comprehensive method for quantifying a project's value throughout the pipeline.
In this article, we discuss various tools for assessing project values—in particular, three methods: risk-adjusted discounted cash flow (DCF), decision-tree analysis (DTA), and an option-pricing model (OPM). We elaborate on the methods, describe their relative strengths and weaknesses, make comparisons through a sample calculation, and assess their suitability for quantifying projects at different stages of development. Our conclusion is that OPM, while little used in the industry, is the most accurate valuation tool available. Both DCF—the most commonly used method—and DTA significantly overstate the risk of early-stage projects in part because of their inability to account for progress in an R&D project.
Indeed, any valuation method needs to take into account three interrelated characteristics of R&D projects. The first is that the R&D process increases knowledge incrementally. R&D is an information-gathering process in which uncertainty is greatest at the beginning. The further a project proceeds, the better the understanding of the molecule's structure, safety, efficacy, quality, dosage, risk/benefit ratio, and cost/performance ratio. Accordingly, a project can be separated into distinct stages, allowing for periodic assessment and facilitating go-no-go decisions.
Second, the R&D process gives managers the ability to defer actions. Splitting R&D into various stages allows multiple decision points, giving management more time and data to re-examine the course of events. Periodic re-evaluation increases the opportunity to avoid costly mistakes and takes into account any external events in the project's favor (changes in competitors or regulations, for example) that have occurred since the project began.
Finally, R&D allows for gradual investment. Only those R&D investments that have already been made in a project become sunk costs. All subsequent investments are open to future decisions.
The Trouble with Two Common Methods
The most common valuation technique is DCF, which is used on projects in all stages of development. The system is based on discounting the cash-flow projections (revenues minus investments) of a project over a period of time to account for the project's risks and uncertainties, as well as the cost of capital. The net present value (NPV) of the calculation is the project value.
The most crucial and most subjective element of the equation is the derivation of the project uncertainties, and hence the discounting element. These commercial and technical uncertainties link directly with the cash-flow projections of, respectively, revenues and investments. Commercial uncertainties affect revenues: what is the probability that the eventual product will meet the market requirements and the estimated demand? Technical uncertainties affect investments: how likely is it that the project will lead to a product that has the required attributes?
Analysts quantify these uncertainties into a DCF calculation using one of two basic approaches. In the first, they estimate the project's inherent probabilities for commercial and technological uncertainty. Then they factor in revenues and investments accordingly (through Monte Carlo simulations, expected values, or both), discounting them by the given cost of capital (as defined by the financial markets). In the second, an analyst adjusts the discount rate by adding a premium to the cost of capital. Typically, the resulting hurdle rates are four to five times the cost of capital. With either approach, if the calculated NPV of the project is positive, management will decide to invest. If the value is negative, it won't.
But DCF has a number of problems. When analysts use the second approach—adding a premium to the discount rate—decision-makers are presented with only a rough estimate. Indeed, in both approaches the inputs are often inaccurate, creating inflated hurdle rates. Most DCF calculations end up favoring projects with lower risks and late expenses.
Moreover, DCF fails to take into account the three key characteristics of R&D project valuation. Not recognizing that knowledge is accumulated gradually, DCF assesses a project with a winner-take-all philosophy, as if the decision to proceed with a project is made only once. The cumulative total investment in the project is therefore seen as a one-time, irreversible event. By the same token, DCF doesn't account for the flexibility to defer actions in response to events.
Decision-tree analysis more accurately recognizes the influence of real-life events in the course of an R&D project. It allows a given project to be separated into distinct decision points, each with possible options for action. Key decisions, critical hurdles, and major uncertainties are explicitly identified and logically connected in a comprehensive project flow.
There are many similarities between the DCF and DTA methods. Both need the same input parameters, such as investments, potential revenues, and the related risks (the latter, however, at a much more detailed level in DTA), and both are based on the principle of discounting to find the NPV.
DTA differs, however, in two important ways: the calculation equation changes, since it must incorporate all possible outcomes; and investments are factored in with the respective probability of success (see Exhibit 1).
Although DTA is more systematic and accurate than DCF, and comes closer to the real value of a project, it cannot account for one of the major characteristics of an R&D program: the money value of time, and hence the value of deferring action. This element, which plays a big role in the total value of a project, has a cash value, can be calculated, and is defined as the option value.
An Option-Pricing Model
The ability to defer action is the analogue of the call option for common stocks in the financial derivatives markets. A financial option gives an investor the right (but not the obligation) to buy an asset (in this case the current stock of a company) at a certain strike price at a future time. This option can be exercised within a specific time frame. For this right, the investor pays a premium—the option price.
The value of the option depends on the value of the stock itself and the strike price. At a stock price today of $30 and a strike price of $25, a call option will be worth at least $5. In reality, this value is much higher if the markets believe that the stock price will increase considerably in the future. An increase—for example, from $30 to $45—will have a payoff of $20 ($45 minus $25). Subtracting from this the price of the option, in this case $10.70, the result will be the net profit. If, on the other hand, the stock price falls below today's price, the option will be worthless. No one would invest $25 to receive less.
Accordingly, the option value is calculated to include the chances for upward and downward swings. It is therefore sensitive to the uncertainties surrounding the stock (like interest rates, taxes, and market evolution). It is these uncertainties, however, that make options interesting. Because of the fixed strike price, the upside potential if the stock develops favorably can be great; otherwise, only the initial investment in the option is lost.
R&D investments can be viewed in a similar light. Because future events can change the project path considerably, resulting in big benefits, decision-makers can treat R&D projects as nothing more than options (in the financial market sense) on future benefits (see Exhibit 2). The quantified R&D project value is the analogue of the option price; the present value (PV) of the expected profits from the project is the current stock price (asset). The investment needed for the project corresponds to the strike price. The volatility of the profits (risk) is analogous to the stock volatility. The investment period (project time frame) represents the expiration time of the option, and the riskless rate of return (for discounting) is the same for both.
A comparative analysis of DCF (the traditional investment analysis) and OPM on three of the parameters–volatility, expiration time of the option, and riskless rate of return–shows that OPM is best at addressing R&D's project characteristics.
With DCF, as volatility and risk increase, the NPV of the investment is discounted at much higher rates, resulting in a value decline. The downside risk looks substantial, since the method assumes that the whole investment will be lost. As volatility and risk increase with OPM, however, the value of the project increases too, because only the upside potential is influenced; there is no impact on the downside losses. These losses are limited, in any event, to the investments incurred through the last decision point. For R&D, this means that a wide array of speculative or partially defined applications and new paths (as are found in most innovative work, such as research) feed only into the upside potential.
OPM also more adequately accounts for the impact of time-to-expiration on project value. As the length of a project increases, the payback period becomes longer, hence increasing the discount under DCF and reducing the value. With OPM, a prolonged period offers the possibility of re-examining the course of events, avoiding costly errors, and waiting for a positive turn.
Finally, DCF undervalues the impact of riskless rate of return on project value, translating to higher discount rates and consequently lower project present values. If things change for the better, higher discount rates make the money for future optional investments cheaper, pushing up the value of an R&D project calculated by OPM. This increased value is especially worthwhile for projects that create opportunities for new ideas and endeavors—such as technology platforms at the front end of research—usually found in the discovery part of the pipeline.
The Valuation Methods Side by Side
To show the differences among DCF, DTA, and OPM, we've used a hypothetical follow-up drug project in the pharmaceutical industry (see Exhibit 3).
The project lasts six years, with specific probabilities of success and additional investments each year. At the end of the project, one can expect certain profits or a failure—that is, non-approval. The profits at the end of the period are the values of the expected cash flow from 2004 on. Investments outlined here are non-discounted values; the discount rate is 9% and the riskless rate is 4%. For simplicity, the values used here are the expected average values for the various parameters. Doing the calculations with each method for each stage of the project leads to three conclusions (see Exhibit 4).
1. The OPM approach approximates the project value better than DCF and DTA, especially in the early stages.
The OPM logic indicates that the time-to-expiration and the associated risk affect the value positively by offering the possibility of deferring actions accordingly. In this example, the managerial options are valuable enough to reverse the DCF recommendation to abandon the project (because of negative value in the first two years).
The option value increases every year for two reasons. First, the total incremental investment required to take advantage of the opportunity decreases every year as the project progresses; because the previous years' investments are sunk costs, they aren't considered in the evaluation. Second, as 2003 approaches, the option value is discounted less and less.
The higher value of the option each year does not suggest that we should now, in 1998, also pay the value of the option for each subsequent year. This option value is the amount we would pay in any particular year for the right to continue making the annual investments required to achieve the projected positive outcome of $1,119. Then we can exercise our option at the end of the last stage, if conditions are favorable.
2. DTA gives a better quantification than the classic DCF.
By splitting and reassembling the project at many points, DTA is better than DCF at describing the possibilities of the various branches. Unlike OPM, however, it does not include the money value of time. It undervalues the overall project, in comparison with OPM, and in the first year it even leads to a wrong decision: to abandon the project.
3. Toward the end, all three methods converge, and the differences become less apparent.
As the project approaches completion, its inherent characteristics become more certain; the outcome is more predictable. Because there are now fewer chances to defer actions, the money value of time is less important in computing the total project value. Even DCF yields satisfactory results, since the discounting period is shorter and the investments to be made are only a small proportion of the total needed.
The applicability of the various valuation methods varies stage by stage (see Exhibit 5). But while OPM allows for a more solid evaluation in all stages of a project's life—particularly the early stages, when DCF and DTA can produce inaccurate results—none of the three methods yield an absolute and precise value for a project over the whole pipeline. Obtaining accurate figures is a common problem with existing methods. Quantification, which is pivotal for the relative ranking of projects requiring budget decisions, thereby becomes more difficult.
These three methods capture the inherent characteristics of a project differently, but all are potential solutions to the R&D organization's typical aversion to accepting the importance of economic considerations in research. Debates on which approach to use generally focus on the tradeoff between the complexity of the method and the benefit it yields. It is clear that each method differs from the others in data input and collection, calculation complexity, and implementation (see Exhibit 6). OPM brings with it the additional challenge of being new and difficult to explain. It is more complex intellectually, making it harder under any circumstances to convert those who hesitate to change.
DCF methods, when first introduced, clashed with the intuitive methods of analysis that then prevailed. Similarly, OPM, even though its data collection and calculation requirements are not complex, could present a bureaucratic challenge to R&D organizations comfortable with DCF.
Each organization needs to consider these factors in relation to its internal culture. Questions about its culture include: How willing is the organization to change? When the staff are shown the benefits of a more complex method, will they be willing to learn? Can the needed communications and training programs be implemented to make OPM work?
No matter which method is chosen, creating awareness within R&D that quantification is necessary for business success can have a huge impact. Controlling and managing the quantification process efficiently and effectively across various projects, departments, and regions is surely a major challenge, but doing so will make the decisions about each project transparent, objective, and widely accepted. After all, what gets measured gets done.
Incidentally, whether an option on Mozart's Jupiter Symphony was worth an investment in 1788 is still being debated by the authors.
Nikolas Vrettos is a manager in the Düsseldorf office of The Boston Consulting Group. Michael Steiner is a vice president in the firm's Munich office.