Construction Loan Monitoring: Catching Hidden Cost Overruns

Construction loan monitoring fails for a structural reason: the general contractor reports to the owner, the owner reports to the lender, and the only party with eyes on the actual field conditions is the one with the strongest incentive to look current. By the time a budget problem reaches the lender’s desk, it has usually been absorbed into contingency, shuffled between line items, or quietly sat on for several draw cycles.

Rich Lamphere has spent over 40 years in construction — union carpenter, framing contractor, general contractor, developer, owner’s rep, construction manager — and now runs TL Pacific in Los Angeles, where he does third-party project reviews and works for court-appointed receivers on distressed assets. He built a program called Project Risk Intel specifically to answer two questions for lenders and owners: will this finish on the schedule being reported, and will it finish within the budget being reported.

Below: how overruns stay concealed, what an unbiased review actually inspects, where the real unknowns sit on ground-up work, and how to use AI on a thousand-page file dump without trusting the output blindly.

From the Real Estate Pros Show

This article draws on an interview with Rich Lamphere of TL Pacific LLC on the Real Estate Pros Show, hosted by Meghan Escobar.

Why the Numbers a Lender Sees Are Not the Numbers in the Field

The reporting chain has three links and degrades at every one. The superintendent and project team know what is actually happening on site. The general contractor packages that into an owner report. The owner packages that into a draw request and lender update. Each handoff is an opportunity to round in the direction that keeps money flowing.

Lamphere’s observation after four decades of sitting on both sides of that chain is blunt: it is amazing how little time and money lenders spend to monitor, from a third-party unbiased perspective, what is actually being reported from the field. On a large ground-up multifamily project in a market like Los Angeles, that is tens of millions of dollars of exposure verified largely by the party with the most reason to look current.

The gap is not usually fraud. It is a contractor who believes they will make the number back on the next phase, an owner who does not want to trigger a loan covenant conversation, and a lender whose inspection is a photo set and a percentage-complete figure that nobody independently tested. Everyone in the chain is managing a problem they expect to solve before it has to be disclosed.

That is why the question worth asking is not “what does the report say” but “what controls produced this report, and do they work.” A percentage-complete number with no underlying verification is an opinion. A draw backed by tested quantities, subcontract commitments, and change order logs is a fact. Most lenders are funding on the first and assuming the second.

How an Overrun Stays Hidden: Contingency, Line-Item Shuffling, and Schedule Drift

There are three mechanisms that keep a construction budget overrun invisible, and they work in combination. Lamphere describes the pattern he sees repeatedly: a contractor reporting one thing on budget and schedule when in reality they are upside down — hiding the fact that they are $2 million over, because it has not come to fruition yet.

• Drawing contingency early. Contingency is supposed to cover unknowns in later phases. When it gets consumed in the first third of the job to cover known shortfalls, the budget still reports at 100% of the contract amount. Nothing looks wrong on the summary page. The exposure only surfaces when a real unknown appears and there is nothing left to absorb it.
• Moving money between line items. Overruns in one trade get covered by buyout savings or optimistic remaining-to-complete figures in another. Each individual line stays plausible. The total stays intact. You cannot see it from a top-line budget report — only from comparing committed costs against the original estimate line by line.
• Sitting on it. Some contractors simply know it will eventually catch up to them and say nothing, hoping for a productivity gain, a change order, or a scope reduction that closes the gap before disclosure becomes unavoidable.

The same applies to schedule. Float gets consumed quietly, activities get reported at percentages that assume a recovery that has not been planned, and the critical path drifts without a single missed milestone on paper. Both the cost gap and the schedule gap can stay covered for months — which is exactly how long it takes for the cure to become expensive.

It’s amazing to me how little time and money lenders spend to monitor, from a third-party, unbiased perspective, the reporting from the field. You might have a contractor that’s reporting one thing, budget-wise and schedule-wise, and in reality they’re upside down.

— Rich Lamphere, TL Pacific

What a Third-Party Review Actually Inspects

An unbiased review is not another site walk. It is an audit of the control system that produces the numbers. Lamphere’s Project Risk Intel approach works through four steps on any project a lender has funded or an owner-developer has in progress.

1. Examine the controls in place. What cost tracking, change management, subcontract administration, and schedule updating processes actually exist on this job — and which ones exist only on the org chart.
2. Test whether the reporting is accurate. Compare what is being reported against the underlying documentation rather than against the previous report.
3. Identify where the breakdowns are. Pinpoint which link in the chain is losing information and whether it is a control that is missing or a control that is being ignored.
4. Answer the two questions. Is it really going to finish in the time they say, and is it really going to finish within the budget they are reporting.

The leverage point is pattern. As Lamphere puts it, if there is a two-year schedule and they are a year into it, there is already a pattern of reporting and information being exchanged. That history is the test set. A single month’s draw tells you almost nothing; twelve months of draws, change orders, RFIs, and schedule updates tell you whether the reporting has been tracking reality or drifting from it.

This is also why the work is cheap relative to the risk. Reviewing an existing document trail costs a fraction of what the lender has outstanding — and it is the difference between finding a problem at month 12 and finding it at month 20.

Ground-Up vs Remodel: Where the Unknowns Actually Live

On ground-up construction, the unknowns are front-loaded into the excavation phase. Roughly 90% of Lamphere’s work is ground-up new construction, and his read on risk distribution is specific: the unknowns are what you do not see when you are doing all the excavation.

In Los Angeles, where projects routinely go 30 or 40 feet down for subterranean parking, that phase is where the budget gets decided. Soil conditions, groundwater, undocumented fill, utilities, shoring behavior — you never know what you are going to find. Once you are out of the ground, the surprises thin out considerably. Vertical construction on a new structure is largely a known quantity priced against plans and specs.

The practical consequence for anyone monitoring a ground-up loan: weight your scrutiny toward the first phase. If contingency is being consumed during excavation and shoring, that may be legitimate — those are the real unknowns. If contingency is being consumed after the podium slab, the question is why, because the risk profile of the remaining work does not justify it.

Remodels and tenant improvements carry a different exposure. You are working on something somebody else built, and you discover things you were not anticipating throughout the job, not just at the start. The unknowns are distributed across the whole schedule rather than concentrated at the front, which means a remodel budget never earns the same confidence a ground-up budget does once it is out of the ground.

The Distressed Project Deep Dive for Receivers and Buyers

When a court-appointed receiver or interim trustee takes over a stalled project, the first deliverable is not a plan — it is an accurate statement of where the asset actually stands. Lamphere describes the work as turning over all the rocks: reviewing all the documentation of the current status of the asset, determining what needs to be done to cure the distress, then summarizing it for the receiver or trustee.

That sequence matters. A receiver cannot evaluate whether to fund completion, sell as-is, or restructure without knowing what has been paid for versus what has been installed, which subcontracts are still live, where lien exposure sits, and what the realistic cost and schedule to completion are from today’s condition. Those answers live in files, not in the previous owner’s narrative.

This niche is expanding for a structural reason Lamphere points to: a large volume of commercial loans in this country is coming due, and not all of them will refinance at current terms. Projects that pencilled at 2021 debt costs do not pencil at 2025 debt costs, and partially built assets are the hardest category to resolve because they generate no income and keep accruing carry.

For investors, that is the opportunity side of the same trade. Lamphere is aligning with capital partners to acquire distressed assets, add value, and take them back to market. The diligence requirement is the same as the receiver’s — you need an honest accounting of cost-to-complete before you price the acquisition, because the seller’s number is the one produced by the chain that already failed.

Using AI on Plans, Budgets, and File Dumps Without Trusting It Blindly

The time compression on estimating is real and it is large. Lamphere sent a multimillion-dollar bid package — hundreds of pages of plans and specifications — to his son at 11pm. By 6am he had a full bid spread back in Excel, five or six tabs of structured information. That is work that would normally have taken him at least a couple of weeks to reach the same point.

The distressed-asset application is similar in kind. On one current project there are thousands of pages across dozens of files and hundreds of files within files. Before these tools, reaching a defensible conclusion about where that asset actually stands would have taken an extraordinary amount of manual review.

Two constraints keep it honest. First, output quality is a function of the prompts and the inputs — you have to put your own knowledge into it and ask the questions that produce the result you actually need. Second, the first pass is a draft, not an answer. Lamphere reads through everything: every number, every line of verbiage, every piece of formatting, because the output goes out under his name and has to be right.

The human element is the validation layer. Forty years of pricing and building is what tells you that a line item is low, that a schedule duration is unrealistic, or that a file set is missing something material. AI gets you to a working draft in hours instead of weeks. It does not know when a number is wrong.

Frequently asked questions

What are the earliest signs a construction project is over budget before the contractor admits it?

The earliest signal is contingency being drawn down faster than the work is progressing, particularly after the excavation phase is complete. On a ground-up project, early contingency use during excavation and shoring can be legitimate because that is where the genuine unknowns sit. Contingency consumption after you are out of the ground deserves an explanation.

The second signal is movement between line items. If buyout savings in one trade are repeatedly offsetting overruns in another while the total stays flat, the budget is being managed rather than reported. A third is a schedule that hits every milestone on paper while float quietly disappears from the critical path.

Who pays for third-party construction monitoring — the lender, the owner, or the developer?

It varies by deal structure, and all three commission this work. Lamphere’s Project Risk Intel is geared toward lenders, owners, and developers who want to understand whether their projects are going as well as they are being reported, and he has also done this work for court-appointed receivers, interim trustees, and private REITs.

In practice, whoever carries the capital risk has the strongest reason to pay for independent verification. For a construction loan, that is usually the lender, often with the cost passed through as a loan expense. An equity partner who is not operating the project has the same exposure and the same reason to fund a review.

At what point in a project schedule is an independent review most useful?

Once there is enough reporting history to test against — Lamphere’s example is a year into a two-year schedule. At that point there is an established pattern of reporting and a trail of exchanged information, so you can compare what was said against what happened rather than taking a single snapshot.

Earlier reviews are still useful for evaluating controls and the quality of the original estimate, but they cannot detect drift. The worst time is after a draw has been refused or a milestone has been publicly missed, because by then the overrun has already surfaced and the cure options are narrower and more expensive.

What does a receiver need to know first when taking over a distressed construction project?

The current physical and financial status of the asset, documented rather than described. That means reviewing everything in the file — contracts, draws, change orders, subcontract status, permits, inspection records — to establish what has actually been paid for versus what has actually been installed.

Only after that can anyone determine what must be done to cure the distress and what it will cost. Lamphere’s process for receivers and interim trustees is to turn over all the rocks, do the deep dive, then deliver a summary that supports the decision about whether to complete, sell, or restructure.

Can AI replace a construction manager for estimating and budget review?

No. It compresses the production work dramatically — a bid spread that would take a couple of weeks can come back overnight — but the output is a draft that requires someone with field experience to validate. Lamphere reads every line of what comes back before it goes anywhere.

The judgment that tells you a unit price is unrealistic, a duration will not hold, or a file set is missing something material comes from having built the work. AI organizes information at a speed no person can match. It does not know when the information is wrong.

The bottom line

If you are carrying capital on a ground-up project and your only view into it comes through the contractor’s report to the owner, commission an independent review of the documentation you already have — the draws, change orders, schedule updates, and contingency ledger — and have someone who has priced the work test the pattern against reality. It costs a fraction of what you have at risk, and the difference between finding the gap at month 12 and month 20 is usually the difference between a cure and a workout.