Underwriting construction risk an investor cannot see from the model

Underwriting construction risk an investor never manages directly is a documents exercise, not a modeling one. The spreadsheet can show you what happens if cost moves. Only the contract tells you whether it can.

The gap between the model and the contract

Your model has a construction cost line, a contingency line and a sensitivity that flexes both. You can run a 5 percent overrun and see what it does to the return in about ten seconds.

What the model cannot tell you is the probability attached to that 5 percent, and that probability does not live in a spreadsheet. It lives in a set of documents describing what the price covers, how much of the work is bought, and who pays when something moves.

This is the structural problem with underwriting construction from a capital seat. The risk is real, it is material to the return, and every instrument you normally use to measure risk is the wrong instrument for it.

The temptation is to handle it with a governance answer: more reporting, a bigger contingency, a tighter approval threshold. Those help after the fact. None of them changes the exposure you took on the day the price was fixed.

The exposure was set by four documents, and all four existed before you funded.

Underwriting construction risk an investor should start with the delivery method

Before any document, establish how the job is being bought, because it decides which risks exist at all.

Lump sum, or design bid build. Design finished, price competed, contractor holds the cost risk for the defined scope. The risk sits in design completeness and in change orders, and the owner has the least visibility into how the price was built.

Construction manager at risk. The builder advises during design, then converts to a guaranteed maximum price. Open book, more visibility, and a structural tension: the party that estimated the job is the party that guarantees the estimate, and it wrote the qualifications defining the guarantee. That mechanism is described in CM at risk explained for owners.

Design build. Single point of responsibility, fewer interfaces, less owner control over specification. Risk concentrates in the performance requirements rather than in the drawings.

None of the three is safer than the others. They move risk to different places, and an underwriting process that does not know which one it is looking at is measuring the wrong things.

The four documents that set your exposure

The qualifications and assumptions page. Every exclusion is a cost the owner carries. An exclusion with a quantity cap is a bounded risk. An exclusion without one is unbounded, and the number of uncapped exclusions is the fastest read available on how the price was assembled.

The buyout log. The share of the price under executed subcontract. Work still priced from an estimate can move; work under contract mostly cannot. This is the difference between a price and a forecast.

The allowance schedule. Each allowance is a deferred decision with a placeholder figure. The total, and how many are carried against superseded design, sizes the part of the number that has not been decided yet.

The change order clause. Markups, unit rates and the pricing basis for work that does not exist yet. These terms govern the price of every change for the life of the job, and they are competitive for about a week before signature and never again.

Four documents, usually under forty pages combined, and they describe more of your actual exposure than the entire financial model does.

What none of the four does is refer to the others, which is where the findings sit. The qualifications page does not say which of its exclusions has an allowance behind it. The buyout log does not say which unbought package carries the specification most likely to be challenged. Reading them one at a time produces four clean summaries and misses the exposure.

That is also why a sponsor reading its own documents rarely produces the finding. The sponsor read each one when it arrived, months apart, in a different frame of mind, and nobody has since put them on the same desk on the same afternoon.

Where the delivery method changes the question

The same four documents mean different things depending on how the job is bought.

On a construction manager at risk job the qualifications page matters most, because the contractor wrote both the estimate and the list of things the estimate excludes. On a lump sum job the drawings matter most, because ambiguity in them is the primary source of change orders and the contractor has every reason to find it.

On a design build job the performance specification matters most. A requirement written as an outcome rather than a component transfers design risk to the builder, which sounds attractive until you discover the builder satisfies it with the cheapest compliant option. The specification is the whole negotiation.

Schedule risk moves too. A construction manager at risk job usually has a schedule built by the party that will execute it, which makes it realistic and also makes its float allocation worth reading carefully.

The underwriting question is not which method is safest. It is whether the documents that carry the risk for this method have been read by somebody whose fee does not depend on the answer.

A worked example

A family office underwrites a $66 million industrial development as a limited partner. The model runs sensitivities at 3, 5 and 8 percent construction overrun and the return holds at 5 percent.

The documents tell a different story. The qualifications page carries three uncapped exclusions: unsuitable soils, rock removal and utility relocations. Buyout is 34 percent. Allowances total $3.7 million, including a $900,000 site work allowance carried against a geotechnical report that was never commissioned.

The model treats overrun as a smooth percentage. The documents describe a job where site work is essentially unpriced and three of the largest risks have no ceiling on them.

A 5 percent movement is not the relevant scenario. The relevant scenario is site work landing 40 percent above the allowance with rock in it, which is a $2.1 million movement concentrated in one category, and the model never tested a shape like that because nobody told it the shape existed.

What to require before you commit capital

1. Confirm the delivery method in writing and state which risks it moves to the owner.
2. Require the qualifications page and count the exclusions that carry no quantity cap.
3. Require the buyout log as a percentage and a dollar figure, and require it updated monthly afterwards.
4. Require the allowance schedule with the design basis for each item.
5. Require the change order markups and rates to be fixed before signature.
6. Run your sensitivity on the three largest named exposures rather than on a smooth percentage.
7. Require that somebody independent of the sponsor and the contractor has read all of the above.

Item six changes more underwriting conclusions than the other six combined. Real overruns are lumpy and concentrated, not smooth, and a model that only tests smooth movement is testing the scenario least likely to occur.

What we do

We read the four documents and give you the named exposures with dollar ranges, so your sensitivity runs on the actual shape of the risk rather than on a percentage. We act only for owners and capital, never for contractors, and we do not produce a competing estimate. The register is written to be handed to an investment committee without translation. The underlying reading is the readiness review, and the lender and investor memo is the short form of it.

Questions people ask