Sensitivity Analysis for Real-Estate Feasibility: Which Variables Actually Move the Answer

Direct answer: In development appraisal, the output you care about — profit, or residual land value — is far more volatile than any single input. A 5% move in achievable price can move residual land value by a third or more. Sensitivity analysis is therefore not an optional refinement; it is the only honest way to state a feasibility result. Test six variables individually, then test realistic combinations, and report a band with the assumptions that produced each end of it.

Why the output is leveraged

Residual land value is a difference between two large numbers. Revenue and cost are each of the order of GDV; land is often a small fraction of GDV. Small percentage errors in the large numbers become large percentage errors in the small residual. The same applies to profit when land price is already fixed.

If GDV = ₹29.25 cr and residual land = ₹4.27 cr,
a 5% GDV fall (₹1.46 cr) removes ~34% of the residual.
A simultaneous 5% cost overrun (₹0.93 cr) removes another ~22%.

(Figures continue the assumption set in https://devpartner.in/insights/residual-land-value-explained.)

The six variables that matter

1. Achievable price (₹ per sq ft, by product and floor). Highest leverage. Test −10%, −5%, base, +5%.
2. Construction cost (₹ per sq ft of gross built-up). Test +5%, +10%, +15%. Include the effect of a deeper basement.
3. Time — approval period, construction period, and the sales tail. Time acts through finance cost and through price drift. Test +6 and +12 months.
4. Efficiency — carpet ÷ gross built-up, and loading. Test ±3 percentage points (https://devpartner.in/insights/development-efficiency-fsi-to-saleable).
5. Finance — interest rate and average exposure. Test ±150 bps and a slower collection profile. Where a policy benchmark is referenced, use the rate published by the Reserve Bank of India on a stated date rather than a remembered figure, and record the lender's term-sheet spread over that benchmark separately.
6. Approval risk — probability and consequence of not obtaining an assumed entitlement (premium FSI, additional height, land-use change). Model as a scenario, not as a discount rate tweak.

Building the grid

Step 1 — one-way sensitivity. Vary each variable alone; record the change in residual land value or profit. Rank the variables by impact. This tells you where diligence money should go.

Step 2 — two-way table. Cross the top two variables, usually price and construction cost:

| RLV (₹ cr) | Cost −5% | Cost base | Cost +5% | Cost +10% | |---|---|---|---|---| | Price +5% | 6.68 | 5.64 | 4.60 | 3.56 | | Price base | 5.32 | 4.28 | 3.24 | 2.20 | | Price −5% | 3.96 | 2.91 | 1.87 | 0.83 | | Price −10% | 2.59 | 1.55 | 0.51 | (0.53) |

(Arithmetic recomputed on the assumption set of the residual draft: GDV ₹29.25 cr, development cost ₹18.54 cr of which marketing at 4% of GDV moves with price, finance ₹2.14 cr held flat, risk at 2% of GDV, developer margin at 20% of development cost. Replace with your own model output — the shape of the surface matters more than the cells.)

Step 3 — coherent scenarios. Real downside is correlated: weak prices usually come with slower absorption and a longer sales tail, which raises finance cost. Build three or four named scenarios rather than 64 mechanical permutations:

• Base — central assumptions, stated sources.
• Slow market — price −7%, absorption 30% slower, +9 months, finance +50 bps.
• Cost shock — construction +12%, programme +3 months.
• Entitlement failure — premium FSI not obtained; scheme falls to base FSI, parking eased, revenue reduced.

Step 4 — breakeven. Report the price at which the deal returns zero margin, and the cost overrun the deal can absorb. Breakevens are more persuasive to a credit committee than probabilities.

Step 5 — decision rule. Fix, in advance, what result would cause you to walk away. Sensitivity analysis that never changes a decision is decoration.

How to report it

• State the base case, then the band: "residual land value ₹1.9–5.7 crore; central ₹4.3 crore".
• Attach the assumption table with sources and dates.
• Show the breakeven price and breakeven cost.
• Say which input is doing the most work — usually price.
• Do not average scenarios into a single number. A weighted average of incompatible futures is not a forecast.

Mistakes and risks

• Varying only the comfortable inputs. Testing cost while holding price flat is not analysis.
• Independent shocks only. Downside variables move together.
• Optimistic base disguised by a wide range. A range around a wrong centre is still wrong.
• Using sensitivity to justify a price. The grid should constrain the bid, not decorate it.
• No time dimension. Delay is the most under-modelled variable in Indian development appraisal, and it compounds through finance.
• Changing several inputs without a version record. Keep an auditable assumption set per run.

Checklist

• [ ] Six variables each tested one-way and ranked by impact
• [ ] Two-way price/cost table produced
• [ ] Three to four coherent named scenarios modelled
• [ ] Breakeven price and breakeven cost overrun stated
• [ ] Approval risk modelled as a scenario, not a fudge factor
• [ ] Finance modelled on the cash-flow profile with a dated rate source
• [ ] Assumption set versioned and attached to the output
• [ ] Walk-away rule agreed before negotiation

FAQs

Is sensitivity analysis the same as risk analysis? No. Sensitivity shows how the answer responds to inputs; risk analysis attaches likelihood and mitigation. Do both.

Should I use Monte Carlo simulation? Only if the input distributions are evidence-based. Simulated precision on invented distributions is worse than a clear four-scenario table.

How wide should the band be? Wide enough to reflect genuine uncertainty in your market and narrow enough to support a decision. If it spans viable to unviable, the answer is "more diligence", not "proceed".

What if the client wants one number? Give the central case with the breakeven. One number without a breakeven is not decision support.

Next step

Model the base case in /development-potential and /construction-cost-calculator, then run the grid in /residual-land-value-calculator. Record the assumption set against the opportunity at /opportunities/new.

Sources

• Policy rates and monetary policy announcements — Reserve Bank of India (accessed 6 August 2026): https://www.rbi.org.in/Scripts/BS_PressReleaseDisplay.aspx

This article states no jurisdiction-specific rate, fee or market figure as fact. Every number in it is either arithmetic on stated assumptions or an explicitly labelled assumption.

Disclaimer. Decision support only, not investment advice. All figures above are illustrative assumptions, not market data or forecasts. Verify prices, costs, absorption, programme, interest rates and entitlement assumptions with qualified valuers, quantity surveyors, lenders, architects and legal counsel, and with the competent statutory authorities, before relying on any appraisal.

Methodology: `/methodology`.