Sensitivity Analysis: Testing Project Viability Under Uncertainty

Every agricultural project – whether it’s a new irrigation scheme, a crop diversification program, or a rural agro-processing unit – is built on a set of assumptions. Prices may rise or fall. Yields can disappoint. Construction costs can overrun. Sensitivity analysis is the tool that stress-tests these assumptions before money is committed. By systematically changing one variable at a time and observing what happens to project worth, decision-makers can see exactly where a project is vulnerable – and what would have to go wrong before it stops being worth doing.

What is sensitivity analysis? 🔗

Sensitivity analysis is a technique for investigating the impact of changes in project variables on the base-case scenario – that is, the most probable outcome used in the original project appraisal. It focuses on the two most widely used measures of project worth: Net Present Value (NPV) and the Internal Rate of Return (IRR). When either of these shifts significantly in response to a small change in an input, the project is said to be “sensitive” to that variable.

The core logic is straightforward. Sensitivity analysis involves assessing the effect of changes in one input variable at a time on NPV, while other inputs remain unchanged. This isolation of variables is what makes the technique so useful – it pinpoints exactly which assumptions carry the most risk.

In agricultural project analysis, those variables typically include output prices, input costs, crop yields, project investment costs, and the timing of benefits. Because farming is exposed to weather, market volatility, and policy shifts, the list of uncertain inputs is often long – making sensitivity analysis especially critical in this sector.

Why it matters: projects are built on forecasts, not certainties 🔗

Sensitivity analysis recognises that there is no such thing as an accurate forecast. A project analysis prepared today might rely on commodity price projections spanning ten or twenty years. The further out those projections reach, the less reliable they become. Sensitivity analysis doesn’t claim to predict what will happen – it shows how much would have to change before the project’s viability is called into question.

This is particularly relevant in agriculture, where the gap between projected and actual outcomes can be substantial. A World Bank-supported agricultural project in Nepal, for example, found that returns were highly sensitive to changes in benefits – with the Economic Rate of Return dropping sharply if output fell by 10% or production costs rose by the same margin. Such findings directly shape how a project is designed and what risks need to be managed before implementation begins.

The purpose of sensitivity analysis in project appraisal 🔗

The purpose of sensitivity analysis is fourfold: to identify the key variables that influence cost and benefit streams; to investigate the consequences of adverse changes in those variables; to assess whether project decisions are likely to be affected by such changes; and to identify actions that could mitigate possible adverse effects.

Put simply, it answers the question every project sponsor needs to ask: what has to go wrong, and by how much, before this project is no longer worth doing?

How sensitivity analysis is carried out 🔗

The process follows a structured sequence. Analysts first identify key variables, then calculate the effect of likely changes in those variables on the base-case IRR or NPV, and finally consider possible combinations of variables that may change simultaneously in an adverse direction.

Step 1: Establish the base case 🔗

The starting point is the base case – a full project model built on the most probable values for all inputs. This produces a baseline NPV and IRR. If a project’s NPV is positive in the base case, it is considered viable; if negative, it should be rejected. All sensitivity tests are measured against this baseline.

Step 2: Identify the key variables 🔗

Not every input needs to be tested. A preliminary set of key variables is typically chosen on the basis of variables that are numerically large (such as investment cost), essential variables whose value is critical to project design, variables occurring early in project life (such as initial fixed operating costs), and variables affected by broader economic changes.

In an agricultural project, the most commonly tested variables are output prices (the price received for crops or livestock products), input costs (fertilisers, seeds, labour, irrigation), crop yields or production volumes, and total investment costs.

Step 3: Vary each variable and recalculate 🔗

NPV is generally most sensitive to changes in unit sales and unit prices, rather than changes in cost per unit, tax rates, or salvage values. The analyst changes one variable – say, the output price – by a fixed percentage (typically 10% or 20%) while holding all other inputs constant, then recalculates the NPV or IRR. This is repeated for each key variable in turn.

The result is a clear picture of which inputs move the project worth the most. Sensitivity analysis shows that the NPV and IRR of a project are most vulnerable to changes in sales price and variable costs per unit, and less vulnerable to changes in fixed costs. In farming projects, this often means that commodity prices and yields sit at the top of the sensitivity ranking.

Step 4: Calculate sensitivity indicators and switching values 🔗

Two specific outputs make sensitivity analysis especially actionable. The first is the sensitivity indicator (SI), which compares the percentage change in NPV with the percentage change in the variable being tested. A high SI means that small changes in that variable produce large swings in project worth – flagging it as a high-risk input.

The second is the switching value. This is the value at which the project’s NPV falls to zero (or IRR drops to the minimum acceptable rate) – in other words, the point at which the project becomes unviable. The IRR can be interpreted as a break-even required return: if the firm’s actual cost of capital stays below the IRR, the project has a positive NPV; if costs of capital rise above it, the NPV turns negative.

Switching values are especially useful because they translate abstract sensitivity indicators into practical thresholds. If an output price only needs to fall by 5% to wipe out the NPV, that is a very different risk profile from a project where prices could fall 40% and still remain viable.

Variables most commonly tested in agricultural projects 🔗

Agricultural projects have their own characteristic set of sensitive variables. While every project is different, the following inputs are routinely subjected to sensitivity testing:

Output prices: Commodity prices are among the most volatile inputs in any farm-level or agribusiness project. A fall in the market price of a crop directly reduces revenue and compresses the project’s net benefit stream. Because price forecasts over a ten-year horizon are inherently uncertain, output prices typically generate the highest sensitivity indicators.

Yields and production volumes: Crop yields depend on weather, pest and disease pressure, soil conditions, and farmer management practices. In new projects – especially those introducing unfamiliar technologies – the gap between projected and actual yields can be significant. A 10% shortfall in yields reduces both revenue and the project’s internal rate of return.

Investment and capital costs: Construction delays, procurement problems, and inflation can push investment costs above initial estimates. Because investment costs occur early in a project’s life and are relatively unaffected by discounting, cost overruns can have a disproportionate impact on NPV.

Operating and input costs: Rising prices for fertilisers, fuel, labour, or water can squeeze net farm income. If a project’s economics depend on a narrow margin between output revenue and input costs, even a moderate increase in operating expenses may shift the NPV significantly.

Project timeline and disbursement delays: Benefits in most agricultural projects take time to materialise – fields must be cleared, crops established, markets developed. Delays in the onset of benefits, or longer-than-expected construction periods, reduce the present value of the benefit stream and can erode viability.

Sensitivity analysis vs. scenario analysis 🔗

Sensitivity analysis varies one input at a time. This is its strength – and its limitation. In the real world, several inputs are subject to change simultaneously, and inputs are often related to each other. For example, raising the price of a product will often result in a decrease in demand. Analysing each variable in isolation can therefore understate the risk when multiple adverse conditions occur together.

Scenario analysis addresses this by creating pessimistic, most-likely, and optimistic scenarios that change several variables at once. In a pessimistic agricultural scenario, low crop prices, below-average yields, and higher-than-expected input costs might all be combined. This gives a worst-case NPV that is more informative than any single-variable test. Both techniques are used together in thorough project appraisals – sensitivity analysis to identify the critical variables, and scenario analysis to test coherent combinations of those variables.

Limitations to keep in mind 🔗

Sensitivity analysis is a powerful planning tool, but it has boundaries. It shows what would happen if a variable changes, but it says nothing about the probability of that change occurring. A key limitation is that sensitivity analysis varies one input at a time and does not indicate how likely any particular change is. A project may appear highly sensitive to a commodity price decline, but if that price is underpinned by long-term contracts or government support, the risk may be manageable in practice.

Similarly, sensitivity analysis does not model the interdependencies between variables. In agriculture, drought that reduces yields will typically also depress local commodity prices – a double adverse effect that a single-variable sensitivity test would not capture. For complex projects in volatile environments, Monte Carlo simulation – which assigns probability distributions to multiple variables simultaneously – provides a more complete risk picture, though it requires considerably more data and analytical capacity.

Despite these limitations, sensitivity analysis holds a central position in model-based decision support, as it effectively identifies the interdependencies among variables within a project model. It is a standard requirement in the economic and financial appraisal of development projects by institutions such as the World Bank, the Asian Development Bank, and major bilateral donors.

What sensitivity analysis means for project decisions 🔗

The real value of sensitivity analysis lies not in the numbers themselves, but in what they prompt decision-makers to do. When a project is found to be highly sensitive to a particular variable, that finding should trigger concrete responses.

If output prices are the dominant risk, the project design might be adjusted to include market linkages, forward contracts, or price support mechanisms. If yields are uncertain, the project might phase in benefits more conservatively, or invest more heavily in extension services and farmer training. If investment costs are a concern, contingency provisions can be built into the budget.

Sensitivity analysis is more than a calculation – it requires professional judgement to identify which assumptions are most uncertain and whether expected returns fairly reflect project risk. The base-case NPV tells you what the project is worth under the most likely conditions. Sensitivity analysis tells you how confident you can be in that number – and what it would take to be wrong.

What do you think? If two agricultural projects have the same base-case NPV but different sensitivity profiles – one highly sensitive to output prices, the other to investment costs – which would you consider the higher-risk investment, and why? And given that most agricultural projects involve multiple uncertain variables, should sensitivity analysis always be accompanied by scenario analysis before a final funding decision is made?

References

1. https://www.upet.ro/annals/economics/pdf/2009/20090205.pdf
2. https://analystprep.com/study-notes/cfa-level-2/sensitivity-analysis-scenario-analysis-and-simulation-analysis/
3. https://www.tutor2u.net/business/reference/investment-appraisal-sensitivity-analysis
4. https://www.gafspfund.org/sites/default/files/inline-files/project%20appraisal%20document%20(pad)%20-%20P164319_May%202018_pre_Negs%20(Recovered).pdf
5. https://uq.pressbooks.pub/introduction-financial-management/chapter/module-9-npv-modelling-and-analysis/
6. https://financialmanagementpro.com/sensitivity-analysis-in-capital-budgeting/
7. https://www.pastpaperhero.com/resources/cfa-level1-capital-budgeting-and-cost-of-capital-risk-analysis-sensitivity-and-scenarios
8. https://www.researchgate.net/publication/387780045_Net_Present_Value_NPV_Sensitivity_Analysis_Understanding_Risk_in_Investment_Projects
9. https://www.vickeryholman.com/news/residual-valuation-and-sensitivity-analysis-in-development-appraisals/