A portfolio approach to risk in capital budgeting using Monte Carlo simulation

This thesis provides a methodology to evaluate groups of interrelated projects that contain high risk. Two types of interdependencies are evaluated. First, cross-temporal interedependencies (time dependent) are where, for example, a decision to implement a followup project is dependent on how well the original project succeeded. Second is crosssectional where the fate of one project is immediately im pacted by what happens to the other project. Semivariance is used to measure risk and may fit managers' views of risk better than variance because it measures only deviations that are less than a desired level of return.

Two major findings from this thesis are that, first, by using semivariance, an ideal group of projects (largest return and least risk of all groups of projects) was more easily found than with variance. Second, when projects were evaluated with dependencies, the best group was one where projects had a negative interrelationship.

Companies where most capital projects have low risk, small capital expenditures, and only moderate profits may find this methodology time consuming and expensive. But companies with few capital projects which are large in size, may find that this methodology will help them to economically evaluate long-term, risky capital projects with less prejudice than traditional evaluation methods.