Description
Vary each input of a value model one at a time between pessimistic and optimistic estimates to see which assumptions drive the result and deserve further research.
A project's value, such as net present value, depends on assumptions about market size, share, price and costs. One-way sensitivity analysis builds a spreadsheet model, gives each input a pessimistic, expected and optimistic value, and sets each input to its extremes in turn while holding the others at expected. Tabulated ranges show which assumptions dominate. In the textbook scooter example, market share swung value from about minus 104 to plus 173 million while market size barely mattered. Its practical use is deciding where additional information is worth buying: research effort should go to the inputs with the widest swing. A tornado chart, sorting inputs by range, is the usual display. Interactions between variables are not captured.
Attributes
- Discipline
- Game Theory & Decision Science
- Loss type
- Unmigrated
- Stage
- Strategy
- Data
- Analytical / expert
- Effort
- Hours
Steps
- Build a spreadsheet model that computes the outcome (NPV, revenue, cost to serve) from explicit assumptions.
- Give each input a pessimistic, expected and optimistic value, ideally from evidence or calibrated estimates.
- Holding all other inputs at expected, set one input to its pessimistic then optimistic value and record the outcome.
- Repeat for every input and tabulate the outcome ranges.
- Sort inputs by range, optionally as a tornado chart.
- Direct research or data collection to the inputs with the widest ranges, and revisit the analysis as estimates improve.
Sources
- Four Ways to do Project Analysis (Crawford, n.d.)
- Uncertainty Analysis with High Dimensional Dependence Modelling (Kurowicka & Cooke, 2006)
Also known as
- Sensitivity analysis
- What-if analysis
- One-at-a-time sensitivity analysis
- Pessimistic-expected-optimistic analysis
Categories
- Strategy & Collaboration
- Game Theory & Decision Science