Probabilistic Budget Estimating
Menton, N.
MCR, LLC

While developing a fixed-year cost estimate is an important process, ultimately decision-makers are likely to be more interested in estimates that will directly help in the creation of a specific program's budget. A probabilistic estimate of a program's budget can show the expected budget given all the modeled uncertainties in cost and schedule. For programs with already existing budgets, this type of estimate can also show the likelihood of overruns for both the program's lifetime and each fiscal year. Unfortunately, little research has been devoted to implementing comprehensive probabilistic budget estimates. Unrealistic assumptions are often made when converting fixed-year to then-year estimates (and subsequently budget estimates), most notably when schedule is assumed to be constant. Intuitively, it is likely that a significant portion of uncertainty in fixed year cost estimates could be the result of deviations from fixed schedules. Holding the schedule constant may force the budget estimate into a shorter or longer time period than is appropriate.

This paper will give a step-by-step description of what can be done with the historical data and models we have now to create probabilistic budget estimates. Uncertainties associated with schedule, phasing, inflation and budget outlays will be addressed. An example of how a budget probability density function can be conditioned on schedule will also be shown using Crystal Ball's™ filter function. Areas for future research will be noted throughout and summarized in the paper's conclusion.