Why pro forma errors are expensive
Development underwriting concentrates enormous consequence into a small number of cells. A misplaced assumption about interest capitalisation or promote timing does not shave a few basis points off a return; it changes whether the deal is financeable and who earns what. These seven errors recur in sponsor models across asset classes, and each has cost somebody a deal or a promote.
1. Modelling annually instead of monthly
An annual model cannot show the month peak equity occurs, cannot capitalise interest against a real draw schedule, and cannot test a covenant at the moment it breaks. Development spend and revenue are profoundly uneven, and averaging them across a year hides the entire financing question.
Everything downstream inherits the error. Peak equity is understated, so the equity raise is undersized. Interest is understated, so cost is understated, so yield on cost flatters. Sponsors discover it during construction, which is the most expensive moment to discover anything.
2. Interest as a flat percentage of the facility
Construction interest accrues on the drawn balance, which grows along an S-curve, and it capitalises into cost during the build. Applying the rate to the full commitment from day one overstates cost; applying it to an average balance understates it in the late, expensive months.
Do it properly and the interest line is typically 15% to 30% different from the flat approximation. On a scheme where the development spread is 120 basis points, that difference is the deal.
3. Contingency that is never drawn
Many models carry contingency as a line in the budget that sits untouched to completion. Real contingency gets consumed, usually early, usually in groundworks. A model where contingency never draws is telling you nothing about what happens when it does.
Model it as a drawable line with a consumption profile, and run a case where it is fully consumed by 60% completion. That case is what a credit committee wants to see, because it is the one that happens.
4. Straight-line absorption
Units do not sell or lease evenly. There is a slow pre-launch period, a release spike, a long middle, and a tail of hard-to-shift units — often the ones with the worst aspect, which are also the ones carrying the assumed premium.
Straight-line absorption pulls revenue forward, which shortens the hold, which flatters IRR disproportionately because IRR is so sensitive to timing. A realistic curve with a slow tail frequently moves levered IRR by 200 to 400 basis points.
5. Exit cap rate equal to entry
Assuming you exit at the cap rate you underwrote at is an assumption that the market in three years is identical to today. Committees expect at least 25 to 50 basis points of expansion as a base case, and a downside testing 100.
Cap rate sensitivity usually dominates every operational variable in the model. If a 50 basis point move destroys the promote, that is the single most important fact about the deal, and it should be on the summary tab rather than buried in a sensitivity appendix.
6. A waterfall that pays promote too early
This is the error that costs sponsors relationships. Promote is calculated on cumulative distributions without testing, in each period, whether the preferred return has actually been caught up. The result overstates sponsor economics, sometimes substantially.
Investor counsel finds it. Not always before closing, which is worse: a promote paid on a mis-modelled waterfall becomes a clawback conversation, and clawback conversations end partnerships. Test the catch-up every period, model clawback where the documents provide for it, and have the waterfall reviewed independently before it goes out.
7. No delay scenario
Programme delay is the most probable adverse event in development and the least frequently modelled. A six-month slip extends interest, defers revenue, pushes the exit into a different market and can breach a facility's longstop date.
It should be a switch, not a separate workbook: set delay to six months and watch IRR, peak equity and covenant headroom respond. If the deal only works on programme, that is worth knowing before land commitment rather than after.
What to do about it
Six of these seven are structural rather than analytical — they are fixed by building the model differently, not by having better market judgement. That is the encouraging part: they are all avoidable, and an independent review catches them in days rather than in construction.