Simulation Analysis in Capital Budgeting

A single NPV figure looks precise, but it hides a lot of guesswork. Sales forecasts, raw material costs, and discount rates are all assumptions, not certainties, and a project that looks profitable on paper can turn unprofitable the moment one of those assumptions shifts. Simulation analysis, most commonly done through the Monte Carlo method, deals with this problem directly by replacing fixed assumptions with ranges of possible values and showing decision-makers the full spread of outcomes a project could realistically produce.

What simulation analysis actually means 🔗

In capital budgeting, simulation analysis is a technique where a project is modelled mathematically, and the model is then run repeatedly, each time with different combinations of values for the uncertain variables. The Institute of Chartered Accountants of India describes it as a technique involving a very large number of calculations to arrive at the possible outcomes and probabilities attached to a given decision, with the Monte Carlo variant tying together sensitivity and probability distribution analysis. The name comes from the city of Monte Carlo, known for its casinos, because the method leans on random number generation the same way a roulette wheel does.

Instead of saying “sales will be 10,000 units,” the model says “sales could range from 7,000 to 13,000 units, and values closer to 10,000 are more likely than the extremes.” The computer then draws a random value from that range, does the same for every other uncertain input, calculates the resulting NPV or IRR, and repeats the entire process thousands of times.

Why point estimates and sensitivity analysis are not enough 🔗

Most students first learn sensitivity analysis, where you change one variable at a time, holding everything else constant, to see how much the NPV moves. It is useful for spotting which variable a project is most exposed to, but it has a clear weakness: real projects rarely have their variables moving one at a time. Sales volume, price, and cost typically shift together, and sensitivity analysis cannot capture that combined effect.

Where scenario analysis fits in 🔗

Scenario analysis tries to fix this by building a few discrete pictures, such as best case, base case, and worst case, and assigning probabilities to each. This is an improvement, but it still limits the analysis to a handful of fixed combinations. Analysts researching CFA-level risk methods note that sensitivity analysis, scenario analysis, and Monte Carlo simulation are all used together to assess the stand-alone risk of a capital project, with each technique building on the limitations of the one before it. Simulation analysis is the natural extension of this chain because it lets every uncertain variable move simultaneously, across a continuous range of values, rather than being locked into three or four discrete scenarios.

How a Monte Carlo simulation is actually run 🔗

Running a simulation is a systematic process rather than a one-step calculation. It generally follows these stages.

1. Build the base financial model 🔗

Start with a standard capital budgeting model that links inputs like sales volume, selling price, variable cost, fixed cost, and the discount rate to outputs like NPV and IRR. This model has to work correctly for a single set of fixed values before it is worth simulating.

2. Identify the uncertain variables 🔗

Not every input needs a probability distribution. Focus on the two or three variables that genuinely swing the outcome, usually revenue drivers and major cost items, since adding distributions to every minor input just slows the model down without adding useful insight.

3. Assign a probability distribution to each variable 🔗

This is the step that separates simulation from a simple spreadsheet. Each uncertain input is described using a statistical distribution instead of one number. A demand forecast might follow a normal distribution centred on the expected value, while a cost that has a firm floor but an uncertain ceiling might follow a triangular distribution. A simplified example for a new product launch might look like this:

4. Run the simulation 🔗

Specialised software, or spreadsheet add-ins built for this purpose, randomly draws one value from each distribution, plugs the combination into the financial model, and records the resulting NPV. This single pass is repeated a large number of times, often several thousand iterations, so that the full range of possible combinations gets represented.

5. Study the output distribution 🔗

Once all the iterations are complete, the results form a probability distribution of NPV rather than a single number. A widely cited case published by the Institute of Management Accountants describes a capital budgeting model where Monte Carlo software generated a distribution showing roughly a 72 percent probability of a positive NPV, with the average of the simulated outcomes landing close to the NPV calculated using the most likely single-point values. That gap between the two numbers, however small, is exactly the kind of insight a static NPV calculation cannot offer.

Reading the results: more than just the average 🔗

The mean NPV from a simulation is rarely the most useful number by itself. What managers actually rely on includes:

• Probability of a positive NPV: the share of iterations where the project clears the acceptance threshold.
• Standard deviation of NPV: a measure of how spread out the outcomes are, indicating how risky the project is in absolute terms.
• Coefficient of variation: standard deviation divided by mean NPV, useful for comparing risk across projects of different sizes.
• Probability of loss: the chance that the project destroys value rather than creates it.

A risk management study on capital budgeting notes that this kind of probabilistic method is not meant to replace the standard deterministic NPV calculation but to enhance it, giving decision-makers a measure of the variance attached to a return estimate. Two projects with an identical average NPV can carry very different levels of risk once you look at the spread of their simulated outcomes, and that spread is often the deciding factor between two competing proposals.

Simulation compared with sensitivity and scenario analysis 🔗

Where simulation analysis is genuinely useful 🔗

Simulation analysis earns its place in capital budgeting for a few concrete reasons.

It captures interdependence between variables. Real-world sales and cost figures move together, and simulation lets you build those relationships into the model instead of ignoring them.

It quantifies risk instead of describing it. “This project is risky” becomes “this project has a 25 percent chance of a negative NPV,” which is far more useful for comparing projects or setting contingency reserves.

It supports better communication with non-finance stakeholders. A probability curve is often easier for a board member or investor to grasp at a glance than a table of ratios, which matters when the outcome of a capital budgeting decision needs buy-in from people outside the finance function.

Where it falls short 🔗

Simulation analysis is not a substitute for sound judgement. The quality of the output depends entirely on how accurately the probability distributions represent reality, and poorly chosen distributions produce a convincing-looking result that is still wrong. Building and running a proper simulation also demands more time, data, and often specialised software compared to a simple sensitivity table, which is why smaller organisations sometimes stick to scenario analysis instead. As one detailed guide on capital budgeting techniques points out, managers are generally advised to use simulation analysis alongside scenario and decision tree analysis rather than as a stand-alone method, so that the limitations of one technique are offset by the strengths of another.

It’s also worth remembering that simulation does not eliminate risk, it only measures it more precisely. A project can still show a comfortable 80 percent probability of a positive NPV and end up in the unlucky 20 percent. Academic work on Monte Carlo methods for financial risk assessment applies the same logic well beyond capital budgeting, extending it to stand-alone project risk, portfolio value-at-risk calculations, and comparisons between risky and risk-free investment strategies, which shows how central this idea of a distributed outcome, rather than a single guess, has become across finance.

A practical way to think about it 🔗

Consider a company evaluating a new manufacturing line. A traditional NPV calculation gives one number, and the project either clears the hurdle rate or it doesn’t. Running the same model through a simulation instead produces a range: perhaps the project is profitable in seven out of ten outcomes, with a small but real chance of a loss if raw material costs spike and demand disappoints at the same time. That single piece of information, the shape and spread of possible outcomes, often matters more to a finance team than the average NPV itself, because it tells them exactly how much downside they are exposed to before they commit capital.

What do you think? If two projects show the same average NPV but very different probability distributions, which one would you choose, and would your answer change if your organisation had limited cash reserves to absorb a bad outcome?

References

1. https://resource.cdn.icai.org/74830bos60509-cp3.pdf
2. https://analystprep.com/study-notes/cfa-level-2/sensitivity-analysis-scenario-analysis-and-simulation-analysis/
3. https://www.imanet.org/-/media/125d2e3fa1434a2ca33a95520d373e39.ashx
4. https://crmsindonesia.org/publications/risk-analysis-in-capital-budgeting/
5. https://fastercapital.com/content/Sensitivity-Analysis–How-to-Perform-Sensitivity-Analysis-for-Capital-Budgeting-Decisions.html
6. https://www.researchgate.net/publication/221527265_Introduction_to_Financial_Risk_Assessment_using_Monte_Carlo_Simulation