Reexamining the Role of NPV in IT Investment Decisions: A Framework for Analysis

Of the many management tools that higher education has adapted from the corporate world, calculation of “net present value” (NPV) is one of the simplest and most pervasive. NPV calculation discounts future costs and benefits based on an assumption of interest rates in order to effectively recast all costs and benefits into the present for comparison.¹ While often used in making decisions about major IT projects, NPV alone does not provide a sufficient basis for selecting among alternatives for IT projects.

Definition

NPV is based on the concept that a dollar spent or received next year is equivalent to a lesser amount this year because money held today can earn interest and increase by next year. Economically, 94 cents now, earning 6.3%,² is the same as a dollar a year from now; similarly, a dollar 10 years from now is equivalent to 55 cents now. This is an important concept because a system’s benefits typically accrue years after the costs are incurred to procure or develop that system. To say that a dollar of benefits 10 years from now balances a dollar of costs now simply isn’t correct. Calculating the discounting for net costs or benefits in each year and summing creates the “net present value” of all cash flows over the lifetime of an IT system.³

Costs and Benefits; Tangible and Intangible; “True NPV”

A positive NPV means that the value of benefits in present dollars is greater than the value of costs in present dollars — and vice versa for negative NPV. The decision-making assumption is that positive NPV is good (provides positive value to the college/university) and that negative NPV is bad (costs more than it is worth). So why would an institution engage in a project that has a negative NPV (i.e., that costs more than it provides in benefits)? If your answer is, “well, there are intangible benefits” (benefits whose value is difficult to quantify in dollars), then what you are really saying is that the NPV would have been positive if you could have figured out how to quantify those difficult-to-quantify benefits. So one problem with using NPV as a decision tool is not in the calculation per se but, rather, in the range of benefits and costs that are included in that calculation and in how they are quantified.

When it comes to IT systems, many of the benefits we expect are intangible — or at least not very easy to quantify with a high degree of certainty. Therefore, NPV calculated for tangible benefits alone should not be presented as the single measure for making a decision. And attempting to quantify intangibles just clouds the issues (how reliable is the quantification of inherently difficult-to-quantify items?), so some other type of presentation is necessary. Institutions need to create a “true NPV,” one that includes a qualitative presentation of intangible benefits as well as a quantitative presentation of tangibles.

There are various ways to present intangible benefits. Determining whether a new system will have a barely perceptible reduction in administrative effort versus a reduction in effort by approximately half is more believable than attempting to put a specific dollar value on its impact on thousands of staff. So a gross determination of impact (“reducing total administrative effort for xyz task by 50%” or “some administrators will save a few minutes”) is one approach that is often useful. This approach might be called “coarsely quantified” and can be taken to whatever level of detail as can be reasonably validated.

Another approach is to work backward, asking: “What is this intangible benefit worth?” For example, you can’t really put a dollar value on the impact that a new research computing center will have on faculty recruitment and retention, but based on discussions with many faculty and deans, you know it will have a positive impact. If the NPV of a new research computing center (taking into account all the quantifiable costs and benefits ) is negative $20M, then you can ask whether the institution wants to spend $20M over the expected lifetime of the center (say, 10 years) for the set of intangible benefits this might buy, including benefit to faculty recruitment and retention. This reframes the question and provides further input to making the decision.⁴

Decision makers need as much information as possible about intangible benefits and costs, including how those intangibles impact different constituencies (faculty, researchers, staff, students) and to what extent. Do as much analysis and presentation of intangible benefits as you can defend, but don’t fall into the trap of attempting to boil everything down to a single number. IT leadership should be presenting all the information in the best available form — fully quantified, coarsely quantified, or just narrative — to appropriately frame alternatives for senior decision makers. That is showing the “true NPV.”

Developing Alternatives

Another, more difficult issue with NPV (or even with “true NPV”) as a decision-making tool concerns the number of alternatives in the comparison. Say, for example, that you are analyzing alternatives for dealing with your aging student system. Replacing the system is a decision you are likely to make once every 20 years or so (or maybe every 30 years, based on the age of some systems I’ve seen!). You determine the alternatives (e.g., develop a new system, license from a vendor, move to the cloud) and estimate costs and benefits for each one. The problem is that reality is likely to be much messier than any alternative you have analyzed: 20 years is a really long time in information technology, and your alternatives represent only a few “idealized” choices. For example, a more realistic scenario might be that you keep your current system for several more years, investing a significant amount in a “technology refresh” at some point and eventually moving to a cloud-based system after a period of operating the existing system in parallel for several years to allow faculty time to convert. Another several years further on, your cloud vendor will falter, or some new approach will be demonstrated, and you will repeat this cycle. We know that system life-cycles aren’t nearly as simple as our typical models make them out to be, but coming up with all the alternatives — with every possible twist and turn — is just too difficult, and uncertain, an approach to take. In the real world, rather than making one decision that lasts 20 or 30 years, you are likely to have multiple decision points of varying magnitude (fix, update, refresh, replace) along the way to that 20- or 30-year system life.

Even if we could analyze more complex life-cycles at the outset, we are likely to miss those important sea changes that will inevitably occur, both in IT and in higher education (new approaches to pedagogy, governmental regulation, varying financial resources, new technologies, even corporate takeovers). Who would have been so prescient as to put “move to the cloud” into a life-cycle analysis of a student system being discussed even 15 years ago? Taking into account all of the possibilities over the realistic system life isn’t just unmanageable (spreadsheets don’t have that many columns!); doing so is actually impossible, at least without a really good crystal ball.

Analyzing to the Next Decision Point

Another way of attacking this problem (“too many life-cycles, too much uncertainty”) is to realize that we don’t really need to look at entire system life-cycles but, rather, at just enough of the life-cycle to get us to the next decision point. For example, instead of looking at all of the alternatives for replacing that aging student system, you could look at replacing now versus performing a technology refresh to get you through a few more years. You need to ask yourself these questions:

1. Are you in a period of pedagogy or technology change that would argue for waiting a few more years before analyzing possible replacements? The last five years are a good example of just such a period as MOOCs, cloud-based technology, and several new vendors have changed the landscape for student systems.
2. Can the current system last several more years, and if it needs a technology refresh, what would that cost?
3. Is the cost and upheaval of converting to a new system something that can be borne by the institution now and over the next several years?

“True NPV” analysis will help, but there are many other factors that will enter into answering these questions.⁵ And of course, here is where your experience in the messy real-life life-cycle of systems comes in. Idealized life-cycles that we might wish we had just don’t happen very often in real life.

Conclusion

No matter how much we might try to simplify it, the real world of IT investment is not just about quantifying the quantifiable and making a single decision at the start of a project. It is about analyzing all of the relevant factors as well as the potential decision points throughout the system life-cycle. NPV is a useful decision metric, but successful IT decision making requires a more sophisticated approach using all of the tools and the experience we have at our disposal to get to the “true NPV.”

Notes

1. Future value of an amount = amount * (1 + interest rate) ^ number of years in the future. Present value = amount / (1 + interest rate) ^ number of years back to the present.
2. This figure is approximately the 10-year average university endowment return as recently reported by NACUBO (see NACUBO, “Educational Endowments’ Investment Returns,” Commonfund Study of Endowments (NCSE), January 27, 2016.  There is significant variance year to year, and among institutions, so the interest rate appropriate to your situation should be verified with your financial office.
3. We focus on systems, but the same type of analysis can be applied to starting up a new service and looking 5, 10, or more years into the future for calculating costs and benefits.
4. But it doesn’t provide a ready approach to comparing multiple $20M expenditures with different types of intangible benefits — something that deans and provosts and chancellors have to do all the time!
5. You might also want to look at Jerrold M. Grochow, “IT Infrastructure Projects: A Framework for Analysis,” EDUCAUSE Review, January 26, 2015, and “IT Infrastructure Projects: A Framework for Analysis,” ECAR Research Bulletin, July 17, 2014.

---

Jerrold M. Grochow ([email protected]) retired as MIT’s Vice President for Information Services & Technology and served as Internet2’s Vice President for NET+ Services. Now a consultant to higher education institutions and organizations, he has been recognized as a Distinguished Engineer by the ACM and is a Senior Member of the IEEE.