Purpose
Net Present Value (NPV) is a valuable tool for instrumenting incomplete development efforts, including IT projects. It
can be used to assess the value of continuing a project, understanding if it has added value to date, weighing its
value relative to other projects, and trading between various different means by which the project might be finished.
NPV provides a meaningful alternative to the conventional wisdom that an effort has no value until it ships, by
objectively acknowledging work already done and providing a basis for ongoing value management.
Definition
NPV = the distribution empirically derived from the sum of all random variables characterizing monetary benefits
expected over time, minus the sum of all random variables characterizing expected costs over time. This is not a
trivial calculation, but it is straightforward. Note that it results in a distribution, not a single number, though one
commonly uses the mean and standard deviation - two numbers - to characterize the result. A benefit of its use is that
the standard deviation is a highly appropriate measure of uncertainty or risk.
The calculation uses streams of distributions for benefits and costs, an appropriate discount rate representing that
future money is worth less, and Monte Carlo simulation to permit arithmetic on these streams of distributions. While it
is not necessary to divide the streams any further than benefits and costs, we have found it appropriate to do so due
to the various classes of stakeholders typically found in larger IT organizations, who know about some types of
benefits or costs, but not others. Thus, a common formulation accounts for mission effectiveness benefits and
operational effectiveness (efficiency) benefits separately. Similarly, costs are divided into direct development costs,
integration costs, and costs of ownership such as maintenance activities. In this formulation, the NPV is the
distribution of the following formula:
where:
-
IV = Investment value
-
MEi = Revenue or mission effectiveness benefits stream
-
OEj = Operational effectiveness benefits stream
-
DCk = Direct development costs stream
-
DIl = Integration costs stream •
-
Ms = total costs of ownership stream
-
rME, rOE, rDC, rDI and rM all are discount rates accounting for the time value of money. They could all be the
same, or they could all be different.
Again, note that IV, MEi, OEj, DCk, DIl, and Ms all are random variables. Note also that the summation indices are
different, due to the fact that, in general, benefits and costs might accrue at different intervals. In practice, we
usually make these intervals all the same.
A valuable property of NPV as an investment value metric is that it is continuous. It answers two key questions
supporting value-based decision making:
-
Are we creating value? (Is the program healthy? Should we intervene for some reason? Cut our losses?)
-
Is this program worth continuing? (Is the program still needed and worth investing in? Should its content be
adjusted?)
Analysis
The best way to think about performing this calculation is in the context of a spreadsheet or matrix. One arranges the
kinds of benefits and costs in the rows. One might place all considered mission effectiveness or revenue benefits in
the first few rows, then operational effectiveness benefits in the next few rows, then direct costs, and so on. Next, a
set of columns is prepared, a set of three columns for each period from the current date out to the projected time
horizon. For example, one might have a time horizon of five years, and the periods might be quarterly. Thus, the number
of columns necessary would be the cost/benefit title, then 3 columns for each of 20 periods. Why 3 columns? This is the
simplest method of satisfying the requirement that each element of each stream be a distribution, rather than a single
number. Three monetary values is a convenient method of describing a triangular distribution (lowest possible, highest
possible, expected value), and triangular distributions have been found particularly applicable to this kind of value
estimation analysis. A triangular distribution isn't the only way to characterize these future values, but it is what
we recommend until the reader has experience with the method as a whole. It is also interesting to see the emotional
difference associated with requesting triangular distributions for estimates, instead of single numbers, from
stakeholders. The stakeholder asked for a single number (how much is it going to cost, for example) will inevitably
game the questioner. A discussion surrounding distributions is generally far more honest, because the stakeholder
quickly realizes that s/he will have an opportunity to discuss uncertainties and risks with the questioner.
The next part of the analysis requires automation for reasonable performance. A Monte Carlo simulation must be run on
every distribution (random variable) in the spreadsheet. (You cannot do arithmetic directly on distributions. Monte
Carlo simulation is the easiest and most well-known method for doing arithmetic on random variables.) 1000 trials is
generally sufficient. 10,000, or 100,000 trials is better. There are a few Microsoft Excel spreadsheet plug-in
applications that will do this, such as Oracle's Crystal Ball. Excel itself has a rand() function, so it is possible to
build the Monte Carlo simulation entirely within Excel, though that isn't recommended. For many of these plug-in
applications, a fourth column, that is 4 instead of 3 columns per period, will be useful so that there is workspace for
the result of each Monte Carlo trial.
More difficult than developing the Monte Carlo simulation is the analysis necessary to monetize benefits that
traditionally may not be quantified in this way. In fact, right about now you may be asking yourself if this is worth
it! Yet it most assuredly is. The analysis work to monetize mission effectiveness benefits, and/or to make estimates
about future revenues in a rigorous, disciplined manner, is extremely valuable to the portfolio management process, and
necessary if one is to deal effectively and objectively with portfolio management decisions. Monetizing mission
effectiveness benefits is beyond the scope of this guideline, but there are numerous references on the topic. One very
useful suggested reference is How to Measure Anything, Finding the Value of Intangibles in Business, by
Douglas W. Hubbard, ISBN 978-0-470-11012-6.
|