How Financial Firms Decide on Technology

(Part Five)

2.3.2¡¡¡EEvaluating Oppoutunities

¡¡¡¡Once a project is at least initially defined, there is a process by which the initial idea is converted into a proposal that can be evaluated by management for approval or rejection of funding.
¡¡¡¡In the last ten years, it has become more or less standard practices to develop a business case or business plan for any substantial IT investment (some small maintenance projects are simply done on request), although the content, sophistication and formality of this process varied substantially. The most typical of these project proposals (assuming a mid-size to large project) take the form of a business plan which includes a qualitiative description of the objectives, competitive environment, a description of the opportunity and, in some cases, an implementation plan. While the form of these plans varies widely, there are some general points of comparison.
¡¡¡¡For the qualitative portion, the major issue is whether the plan explicitly addresses changes in the business environment, or is primarily inward focused. For minor systems enhancement projects with no strategic objective (or even major investments that are not strategic such as year 2000 repairs), it makes sense for the plan to focus entirely on internal issues. However, to the extent that the investment is made for competitive reasons or is likely to spur a reaction from competitors, it is important to qualitatively evaluate whether the business environment will remain static and , if not, examine possible scenarios that are likely to occur. The assumption of a static business environment is a common decision bias that can particularly plague strategic IT investments; Clemens (1991) terms this the "trap of the vanishing status quo".
¡¡¡¡For the quantative financial evaluation, most IT evaluation methods have their roots in traditional capital budgeting procedures such as discounted cash flow analysis(DCF). However, while these techniques can work well for projects where costs and benefits ar well defined (e.g. purchasing off-the -shelf software in pursuit of operational cost savings), it is increasingly recognized that simple application of DCF approaches is not sufficient for IT investments. This is because much of the value of modern IT investments is likely to be difficult to quantify --such as revenue enhancements or cost savings through improved customer service, product variety, or timeliness. One commonly used stategy is to value non-quantifiable benefits at zero, although this strategy will systematically bias project evaluations to unnecessarily reject projects.
¡¡¡¡Recognizing the limitation of the DCF approach, several alternative approaches have been proposed. One method is to base tghe case entirely on qualitative analysis; unfortunately, this approach often leads to hightly subjective judgements and is likely to err on the side of accepting bad projects. Kaplan, recognizing this problem in the context of evaluating computer integrated manufacturing (CIM), proposed using a variant of DCF; a firm calculates the present value of the investment using all the components that can be quantified and then compares this prliminary value to the qualitative list of other benefits and costs. In other cases, where the evaluation is make difficult because of future uncertainties (e.g. market growth and acceptance; response of competitors ) decision trees or other types of probability based assessment tools may simplify investment decisios. Finally, for some types of investments or decisions (for example, the decision whether to invest immediately or defer), advanced techniques such as real options can be applied.
¡¡¡¡Although this discussion has focused primarily on evaluating the benefit part of the quantitative evaluation, there are other difficulties in estimating the cost of IT projects, particualarly those involving software development. Existing models such as COCOMO of function points estimation are known to improve the ability to predict project length, staffing requirements and total costs, although they are k own to be systmeatically off by as much as 400%. However, the accuracy of these estimates can also be improved later in the project when specifications are well defined or by customizing the models to the experience of a particular organization. However, despite the fact that these tools and approaches are readily available many firms still utilize "seat of the pants" estimants or lock in schedules and cost estimants before the projects are fully defined.