Sensitivity testing in a viability appraisal

Viability appraisal sensitivity testing exists because the residual land value is a small difference between two very large numbers. On a scheme with a gross development value of £12,000,000 and total costs of £11,000,000, a 5 per cent movement in sales values changes the residual by £600,000, which can be most of the surplus the section 106 negotiation is about. A single base case tells the reader what the appraiser assumed. Sensitivity tells them how much that assumption matters.

Planning practice guidance on viability expects assessments to be transparent and to be publicly available, and it expects them to be based on evidenced inputs. Sensitivity analysis is how an appraisal shows its working rather than asserting a conclusion.

Why a single residual figure is not enough

Three reasons, and the third is the one that persuades officers.

The arithmetic amplifies. Because costs and values are of similar magnitude, percentage changes in either produce far larger percentage changes in the residual. That is a property of the method, not a weakness in the inputs.

The inputs are ranges, not points. Sales values come from a spread of comparables. Build costs come from an indexed data set with a range around the median. Profit is a judgement about risk. Presenting any of them as a single figure misstates the evidence available.

And the decision the council is taking is about risk allocation. A scheme that is viable in the base case and marginal on a modest downside needs a different answer from one that fails across the range. Sensitivity is what allows a reviewer to distinguish them, and it is what makes a review mechanism a reasoned response rather than a default.

The variables worth testing

Not everything needs testing. Test what moves the answer and what remains uncertain.

• Sales values or capital values, expressed per square metre and cross-referenced to the comparables schedule.
• Build costs, with the source and index date stated, plus any abnormal costs separately because they carry different evidence.
• Developer's profit, tested as a range and on a stated basis, whether on gross development value or on cost.
• Finance rate and the drawdown assumption.
• Sales rate and the absorption period, which on a larger scheme affects finance cost more than the interest rate does.
• Programme length, including a pre-commencement period for discharge of conditions.
• Policy costs: affordable housing quantum and tenure, transfer values, the community infrastructure levy where a charging schedule applies, and biodiversity net gain delivery cost.
• The benchmark land value itself.

Two items commonly missed. Affordable transfer values, which vary considerably by tenure and by registered provider appetite and are often inserted as a single percentage of open market value without evidence. And net gain, which on a constrained site may mean buying off-site units at a price that has to be evidenced rather than estimated.

Ranges that are defensible

The test a reviewer applies is whether the range was chosen from evidence or chosen to produce an answer.

Symmetry is the first thing they look at. An appraisal that tests values at minus 10 per cent and costs at plus 10 per cent, and never the reverse, has presented a downside case and called it sensitivity. Test both directions.

Justify each range from its own source. For build costs, the spread in the published tender price data and the movement in the index since the estimate date. For values, the actual spread of the comparables rather than a round number. For profit, the range seen on comparable schemes and the risk profile of this one. Where a range cannot be evidenced, say what it is: a judgement, with the reasoning.

And state the base case unambiguously, including the date the inputs were compiled, because appraisals get recycled and an eighteen-month-old cost base is a reason on its own for a reviewer to rebuild the model.

Presenting the matrix

Two presentations do most of the work.

A two-way matrix, with sales value movement on one axis and build cost movement on the other, showing the residual land value in each cell. Mark the benchmark land value on it, so a reader can see immediately which combinations produce a surplus and which do not.

Illustrative example, with the figures shown purely to demonstrate the format:

With a benchmark land value of, say, £900,000 in that illustration, the scheme supports full policy compliance only in the upper left of the matrix, and the negotiation is really about which cell the parties think is most likely.

The second presentation is a one-at-a-time table ranking the variables by how much a stated movement in each changes the residual. It answers the question a case officer actually has, which is what this appraisal turns on. Usually it turns on sales values and build costs, and occasionally on the benchmark land value, in which case the argument is a valuation argument rather than a development appraisal one.

The benchmark land value needs testing too

Most sensitivity sections test the development and leave the land value as a fixed number. That is where the largest single uncertainty often sits.

Under the guidance, the benchmark is established using existing use value plus a premium, with alternative use value relevant only where it has permission or a realistic prospect of it. Both components are contestable. The existing use value is a valuation with its own range, and the premium is a judgement about what would reasonably persuade a landowner to release the site.

So run the appraisal against a range of benchmark values, and set out the basis of each. Where the conclusion flips between a low and a high benchmark, the assessment should say so plainly rather than presenting the favourable end as the answer. A Viability Assessment that concedes this openly is treated more seriously than one that does not, and viability work is an area where councils expect a suitably qualified author and independent review.

Linking sensitivity to a review mechanism

This is the practical payoff. Where the sensitivity work shows a scheme that is viable in the base case and marginal on a downside, the proportionate answer is usually a review mechanism rather than a permanent reduction in obligations.

The output should inform three things. The trigger point, whether that is a date, a commencement threshold or a sales milestone. The formula for sharing any surplus above the base case. And the cap, where one is agreed. A matrix that shows the residual under different sales value outcomes is exactly the evidence needed to set a sensible sharing arrangement.

Where the reduction being sought is in affordable housing, the position also has to be reconciled with the Affordable Housing Statement, because the two documents are read together and a mismatch in tenure or unit numbers between them is the first thing a reviewer notices.

How a reviewer reads it

Assume the appraisal will be rebuilt. The district valuer or an independent consultant acting for the authority will substitute their own inputs, and the question becomes whether your model is transparent enough for them to do that.

What helps:

1. A clear input schedule with sources and dates for every figure.
2. Abnormal costs itemised with supporting quotations or estimates rather than a single allowance.
3. Cash flow shown, not just a static residual, where finance is material.
4. Sensitivity presented in both directions, with the reasoning for each range.
5. An honest statement of which variable the conclusion depends on.

What harms the case: an undisclosed contingency layered on top of an already conservative cost, a profit level asserted without comparables, a benchmark land value derived from the price actually paid for the site, and a sensitivity section that tests only the variables that do not matter.

One further point of practice. Because these documents are published, write the sensitivity section knowing that objectors, the parish council and a local surveyor will read it. An appraisal that survives that reading is usually the one that has tested its own assumptions hardest.

Browse the full list of report types to see how the viability and obligations documents fit together, or start a draft from a postcode on PlanScope.

This is general guidance, not legal advice; check the position with your local planning authority or a qualified professional before relying on it. PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval. For specialist technical reports, qualified professional input may be required.