Capital Budgeting: IT Project Portfolio Optimization Redux
This is the second in a series of articles that examines decision-making methods for the selection or rejection of individual projects throughout the project portfolio management process. These methods determine whether or not a given project (either proposed or in process) should be included in your next capital budget.
Capital budgeting is a multi-faceted activity that includes: the formulation and articulation of long-term goals; searching for new and profitable uses for investment funds; the proactive preparing of technical resources, marketing (of both the external or internal kind), and financial forecasts and estimates; the preparation of appropriation and control budgets and integration of these budgets in the firm's information system; the economic evaluation of alternative projects; and the post-audit of performance of past projects.
Although all of these capital budgeting activities are important, this series narrowly examines only the following, albeit very-powerful, methodologies for selecting or rejecting projects:
- Optimization using deterministic input variables (Previous article)
- Optimization using stochastic input variables (This article)
- Decision-tree analysis, with and without real options (Next article)
- Other (See References 12, 13, and 14)
These articles will focus on the sorting of projects based on their risk-adjusted returns calculated as their net present value (NPV), which is based on cash. However, other bases such as economic value added (EVA), which is based on profit, are commonly employed. For a discussion of the rudiments of NPV, see Appendix A of the previous article—Reference 15 below.
My rationale for preferring NPV over EVA: The refrain "Cash is a fact, but profit is an opinion." reflects the thought that the profit and loss statement (also known as the income statement) is subject to much interpretation using the GAAP (Generally Accepted Accounting Principles—a widely accepted set of rules, conventions, standards, and procedures for reporting financial information, as established by the Financial Accounting Standards Board), whereas cash is tangible and well understood without ambiguity.
Furthermore, these articles put financial performance at the top of the value chain. (Although your company may insist on profitability (financial viability), a regulator may insist on its social or economic viability.) Insofar as projects return more in financial resources than they absorb, they will be selected for inclusion in the next budget. But, as discussed in the previous article, when the benefits of diversification are also considered, this financial-performance rule for selection may be waived.
Why Organizations Undertake [Different Kinds of] Projects
First, take a look at the fundamental question of why organizations undertake projects and, then, the different kinds of projects that they might want to take on.
When investors buy securities, they expect to earn, in the long run, rates of return that will compensate them for time and risk: not more and not less. This is not the case, however, in investments in real assets such as new computer hardware or software, where each investment is unique in terms of such factors as timing, the way it fits into the overall operation of the firm, the firm's operating expertise, and the marketing network available. But, above all, unlike investments in the security market (where a known set of securities is available to all), investments in real assets are the result of conditions or goals that are unique to a particular firm. Thus, a firm can develop investment proposals whose costs are below—even substantially below—the present value of their expected future cash flows. Such investments give investors expectations of compensation that exceeds the market's required rate of return. Consequently, they increase demand or the firm's stock and bid up the price. Clearly, the search for, selection, and adoption of such investments is an important objective of the firm because they help to increase the value of the stock.
Classification by type of cash flow
Projects' forecasted cash flows are examined and classified using criteria such as whether they are short- or long-lived and whether or not they are expected to change sign more than once. Cash flow estimation is the most critical, and most difficult, part of the capital budgeting process. Cash flow components sometimes must be forecasted many years into the future, and estimation errors are bound to occur. However, large firms evaluate and accept many projects every year, and as long as the cash flow estimates are unbiased and the errors are random, the estimation errors will tend to cancel each other out. That is, some projects will have NPV estimates that are too high and some will have estimates that are too low, but the average realized NPV on all projects accepted should be relatively close to the aggregate NPV estimate. Later, I'll comment on the fact that these forecasts are often biased and offer some suggestions on how to deal with this problem.
Classification by size
A company might classify projects by size, as follows:
- Routine proposals involving small investments
- Capital projects that require no more than a certain amount of cash outlay
- Major proposals
The size of the proposed project usually determines not only the depth of analysis to which the project is subjected but also the level within the company at which the project is approved or disapproved. For example, the company might establish the following approval authority:
- Plant manager: Investments up to $10,000
- Division manager: Investments up to $50,000
- Vice President: Investments up to $150,000
- Executive Vice President: Investments up to $500,000
- Executive Committee: Investments over $500,000
It is unlikely that the plant manager would use sophisticated capital budgeting techniques when making investments of under $10,000; but such techniques are warranted when large investment proposals are analyzed.
Classification by purpose
When an investment proposal is advanced, the individual or group advancing the proposal usually submits an appropriation request that provides relevant information about the proposal. In addition to the dollar amount, the proposal usually specifies the function or purpose of the investment. These might be categorized in the following way:
- Upgrading an existing software application
- Developing a new software application
- Expanding existing products in existing markets
- Expanding existing products into new markets
Whereas the benefits and costs of routine upgrading of existing software applications are only roughly estimated, very careful analysis is required when it comes to such projects as expansion to new markets, developing a new software application, and the like. In short, the larger the proposed investment and the less routine its nature, the more formal and rigorous the analysis should be.
Relationship between investment projects
The classification of proposals by size and purpose helps determine who in the firm will make the decision to accept or reject the proposal and what resources will be allocated to the analysis. But, there is one more classification important to the evaluation of the project's profitability. (Profitability can come from increased revenues from external product sales or decreased operations costs from internal use ... say, of newly developed software.) You can distinguish among
- Independent projects
- Dependent projects
- Mutually exclusive projects
This classification is important when several investment proposals are being evaluated simultaneously.
Independent, dependent, and mutually exclusive projects
Two projects are said to be economically independent if the acceptance or rejection of one has no effect on the cash flows of the other. If the cash flows from an investment are affected by the decision to accept or reject another investment, the first investment is said to be economically dependent on the second. And, two investments are said to be mutually exclusive if the potential benefits derived from one investment will completely disappear if the other investment is accepted. The acceptance of one of two (or more) mutually exclusive projects results in the rejection of all of the others.
Two projects could be mutually exclusive because of limited capital or manpower or bothSiri Krishnathe two resources simulated in the optimization study discussed below.
Sometimes, the amount of cash available for investment by the firm in any one period of time is limited by management; in other words, there's capital rationing. In those cases, alternative investments compete for funds.
Clearly, many other classifications are possible. Some firms assign a priority rating to alternative proposals; classifying projects as urgent, required, desirable, and so on. Others classify investment alternatives by the location of the projects within the firm, division, or even department. The various schemes are not mutually exclusive; a firm can, and many do, use all of the above-mentioned classifications at one stage or another of their budgeting process.
I've taken the time to discuss these rather mundane matters before advancing to a state-of-the-art, computer-assisted capital budgeting study to emphasize that the high-tech solution that follows does not take place in a vacuum!
Simulation and Optimization
Computer simulation is a tool that can be used to incorporate uncertainty (risk) explicitly into spreadsheet models. A simulation model is the same as a regular spreadsheet model except that some cells include random quantities. Each time the spreadsheet recalculates, new values of the random quantities occur, and these typically lead to different bottom-line results. By forcing the spreadsheet to recalculate many times, a business manager is able to discover the results most likely to occur, those that are least likely to occur, and best-case and worst-case results. This is just as if he or she ran hundreds or thousands of what-if analyses on the spreadsheet, all in one sitting.
Computer optimization uses simulation, sometimes called Monte Carlo simulation, to do a risk analysis on each possible solution generated during an optimization.
In the previous article, an optimization was conducted on an economic model that didn't incorporate uncertainty (risk) in the definition of the input variables (mean and standard deviation in that case). In this article, uncertainty (risk) will be included in the definition of the input variables (manpower and capital usage).
It's important to note that simulation and optimization are sometimes applied only when large projects are being evaluated and after considerable effort has been made to estimate future cash-flows.
Capital budget (portfolio) optimization enables the decision maker to set goals and constraints for the project portfolio at the same time. Then, optimization can use a variety of what-if scenarios to focus on what the portfolio delivers rather than on whether individual projects are good or bad. By asking managers to describe the outputs they want from a portfolio, the debate is around the trade-off of strategic priorities rather than individual projects. This avoids many of the conflicts with sponsors and project managers inherent in other approaches and results in decisions that build confidence among company stakeholders.
Figure 1 illustrates a project selection scenario that's subject to resource limitation and uncertain usage. This screen shot shows the Palisade RISKOptimizer optimization/simulation tool (a Microsoft Excel spreadsheet add-in) maximizing the NPV contributed by the selected projects. Ten projects (A,B,C,D,E,F,G,H,I, and J) are under consideration. Each project uses an uncertain amount of manpower (in thousands of hours) and capital (in millions of dollars). The simulation/optimization maximizes the mean NPV subject to the constraint that there is at most a 5% chance of using more capital or labor than is available. 70,000 man-hours and $85 million are available. Adjustable cells (C3 to C12) will be "1" for those projects selected or "0" for those not selected by the optimization process.
The optimization selected Projects B, C, D, and I, which earn an NPV of $107 million. There is no chance of a manpower shortage, but a 4% chance of a capital shortage. An authoritative discussion of this and other optimizations under uncertainty can be found in Reference 1. In addition, an overview of how the optimization engine is implemented is available in Appendix B below.
Figure 1: Capital budgeting with resource limitation and uncertain usage.
One shortcoming in the model shown in Figure 1 is the tacit assumption that each unit of manpower is fungible: That is, in one hour, day, or week, Nate, your most-junior programmer, will be as productive as Jane, your most-senior programmer. But, that's generally not the case. So, to introduce these differences into your model, you could create a standard person and assign a standard-person coefficient to each programmer. Thus, Nate's might be 0.9 and Jane's might be 2.1. Then, the values for manpower in the spreadsheet model could be entered in standard-person units.
Alternatively, pay could be used to differentiate one programmer from another. Where this unit of measure reflects a constant, corporate-wide cost per unit of productivity, pay needs no adjustment before it is entered into the spreadsheet. Here, $1000 would presumably buy the project more than twice as many hours of Nate's time as Sally's. There are, of course, other refinements that could be made to this model. But, it's here simply to introduce the process of optimization of an uncertain system. By the time you've read this entire series of articles on Capital Budgeting, you should be able to identify some of the other assumptions made above.