Economic analysis usually assumes future cash flows are known. In reality they aren’t, and engineering decisions have to account for that.
The two related concepts get distinguished:
- Risk — outcomes have known probabilities. We can say “there’s a 30% chance the project takes a year longer than planned.” Decisions under risk use probability-weighted analyses (decision trees, expected values, Monte Carlo).
- Uncertainty — outcomes are possible but probabilities aren’t known. We know cost might be higher than the estimate but can’t quantify how likely. Decisions under uncertainty use techniques that don’t require probabilities: Sensitivity analysis, Break-even analysis, scenario analysis.
In engineering practice the line is blurry — most projects mix the two — but the methods break down cleanly:
Methods for uncertainty (no probabilities). Sensitivity analysis varies one input at a time and observes the effect on the outcome. Break-even analysis finds the value of an input that brings the outcome to a threshold (PW = 0, or some other target).
Methods for risk (probabilities known or estimable). Decision trees model sequential decisions and chance events; each leaf has a probability and a payoff, and the tree is folded back to compute expected values. Monte Carlo simulation samples thousands of possible futures from probability distributions on inputs and characterises the resulting distribution of outcomes.
The reason this matters: a project’s expected PW may look good while its worst-case PW is catastrophic, or its outcomes may be very sensitive to one assumption (the exchange rate, say) that’s hard to estimate. Risk and uncertainty analysis reveals these structures that point estimates of PW or IRR hide.
Common pitfalls:
- Ignoring interdependencies. Sensitivity and break-even analyses vary one parameter at a time, which underestimates risk if parameters are correlated (e.g., demand drops and prices drop together in a recession).
- Overconfidence in probabilities. Decision trees with made-up probabilities give a precise-looking answer that’s only as good as the inputs. The structure of the tree is often more informative than its numerical output.
- Confusing risk with uncertainty. Treating uncertainty as if it were risk (assigning probabilities you can’t actually defend) produces false precision; treating risk as uncertainty wastes the probability information you have.
The four core techniques covered in Engineering Economics: Sensitivity analysis, Break-even analysis, decision tree, and expected-value analysis. For the corresponding management framework see Risk management.