Incorporating asymmetric dependency patterns in the evaluation of IS/IT projects using real option analysis

The objective of my dissertation is to create a general approach to evaluating IS/IT projects using Real Option Analysis (ROA). This is an important problem because an IT Project Portfolio (ITPP) can represent hundreds of projects, millions of dollars of investment and hundreds of thousands of employee hours. Therefore, any advance in the techniques used to manage an ITPP will save a significant amount of limited resources. A primary obstacle in using traditional methods to evaluate IS/IT projects is that they are notoriously risky. Cost overruns of 100%, or even outright failure are commonplace (Standish Group, 2009). When project volatility is high metrics such as NPV and ROI are of limited value. The weakness of these measures is that they are point estimates of static business environments and do not account for managerial flexibility. ROA has been a primary stream of MIS research for the last decade and is a suggested approach for evaluating projects in IS management frameworks such as COBIT and Val IT. There are several known issues that need to be addressed before project portfolios can be evaluated using ROA. Unlike investments in financial portfolios, projects in MIS are clearly not independent of each other. That is, companies invest in particular technologies that actively interact and often depend on each other, e.g. Java, Oracle, SAP etc. A second issue is that IS/IT projects and their driving variables can be non-normally distributed, non-linearly related, and asymmetrically correlated. These characteristics, especially inter and intra-project interactions between non-normally distributed variables, violate the assumptions of the basic Black-Scholes, Binomial, and Margrabe ROA models used in the current MIS research. This work suggests using a Monte Carlo approach to calculating option value using mathematical copulas. Copulas allow each variable that make up a multivariate distribution to be considered separately from the dependence structure between the variables. This work can be seen as a logical next step in the MIS literature, which has thus far focused on evaluating individual projects without interactions, and will enable practitioners to manage their ITPP as a whole, instead of as individual projects