Executive Summary: Construction lending fails in three distinct ways, and no single data source catches all three. A contractor can hold a valid license and still be twelve creditors deep in UCC filings; a contractor with a clean collateral position can still be defending four active suits that will drain working capital before your advance is repaid. This post walks the three checks that actually matter for construction credit, states plainly where each one is available, and identifies the one state where all three can be pulled from a single provider.
Why Does Construction Lending Need Three Independent Checks Instead of One?
Most underwriting stacks were built for general commercial credit and then pointed at construction borrowers without modification. That works until the file goes bad, at which point the post-mortem almost always identifies a risk that lived outside the data the underwriter actually pulled.
What Does a Contractor License Tell You That a Credit File Does Not?
A license is a statement about legal standing to perform the work, not about willingness or ability to pay. It answers a narrower question than credit data, and that narrowness is the point. In New York City, any individual or business performing home improvement work costing more than $200 on a residential building must hold a Home Improvement Contractor license from the Department of Consumer and Worker Protection.[1] The requirement sits in the New York City Administrative Code, and the city publishes a public list of unlicensed operators it has taken action against.[2]
The credit consequence is direct rather than reputational. New York courts have held that an unlicensed home improvement contractor cannot enforce the contract or perfect a mechanics lien to recover on the work performed.[3] If your borrower's receivables depend on contracts they lack standing to enforce, the receivable you are lending against is worth materially less than its face value. That is not a compliance detail. It is a collateral valuation problem.
Enforcement is active, not theoretical. California's Contractors State License Board closed a statewide enforcement round in November 2025 that produced 119 legal actions, including 46 sweep operations across 17 counties, 454 site visits, 27 stop-work orders, and 26 individuals referred to district attorneys.[4] Effective July 1, 2026, California raises the minimum civil penalty for unlicensed contracting to $1,500, with the minimums adjustable for inflation every five years against the California Consumer Price Index.[5] A stop-work order on an active job is a cash flow event for your borrower whether or not you were watching for it.
What Do UCC Filings Reveal About Collateral That Is Already Spoken For?
UCC Article 9 financing statements are the public record of who claims a security interest in a borrower's personal property. For construction borrowers, that typically means equipment, receivables, and inventory. New York processes Article 9 filings through the Department of State and exposes a public inquiry system covering financing statements, amendments, and federal tax lien notices filed under Article 10-A of the Lien Law.[6]
The trap in construction specifically is fixtures. A security interest perfected by a central-index UCC filing on goods that become fixtures is subordinate to conflicting interests of any owner or encumbrancer of the real property other than the debtor, regardless of filing time, unless the secured party filed in the real property records as a fixture filing.[7] That means a UCC search alone can leave you confident about a priority position you do not actually hold. If you are lending against equipment that will be installed into a building, your search and your perfection strategy both need to account for the real property side.
What Does Litigation History Add That Neither of the First Two Catches?
Court records capture the disputes that have already escalated past negotiation. In construction that usually means mechanics lien foreclosures, payment suits, bond claims, and breach actions that pull the contractor's principals into depositions instead of onto jobsites. Highwire, a contractor pre-qualification platform, framed the requirement in an evaluation call as a two-part question:
"Are there any open liens against them? Any litigations associated to this specific contractor?"
That is the whole check stated in one sentence. Neither half is answerable from a license lookup or a UCC search. Lien claims that reached court and litigation that names the contractor as a defendant live in court dockets, and nowhere else.
How Bad Is the 2026 Construction Credit Picture?
The case for adding checks depends on whether the underlying risk is rising. On current data it is.
Are Contractor Insolvencies Actually Rising?
Total US bankruptcy filings reached 574,314 in 2025, an 11 percent increase over the 517,308 filed in 2024 and 26.8 percent above the 452,990 filed in 2023.[8] Business filings tracked by the federal judiciary rose 11.4 percent to 25,960 in the twelve months ending March 31, 2026.[9] The commercial reorganization picture is sharper still: commercial Chapter 11 filings in January 2026 ran 76 percent above January 2025, and Subchapter V small business filings rose 67 percent year over year in the first quarter of 2026.[10]
Construction sits directly in the path of that trend because of its working capital structure. Contractors carry payroll and materials cost ahead of payment, which makes them acutely sensitive to any slowdown in draws.[11]
What Are Lien and Surety Data Saying About Payment Stress?
• Lien activity has turned. The NCS Credit Lien Index registered 48 in Q4 2025, falling below the neutral 50 mark for the first time since Q1 2023, which the index reads as contraction in overall construction momentum rather than improvement in payment behavior.[12]
• Liens work, which is why they get filed. More than 64 percent of mechanics liens are paid within three months of filing without any foreclosure action, meaning a filed lien is a strong signal that informal collection already failed.[13]
• Surety loss ratios are running high. Direct loss ratios for US surety writers hit 24.9 percent through the first nine months of 2024, the highest in five years, and underwriters have responded by tightening bonding terms amid rising subcontractor default.[14]
• Bank construction exposure was already deteriorating. US banks' delinquent construction loans reached $4.86 billion in Q1 2024, the highest level in at least three years, with the residential construction delinquency ratio at 1.03 percent.[15]
• Regulators are tracking it. The FDIC's 2026 Risk Review, covering data through December 2025, treats construction and land development within its commercial real estate credit quality analysis.[16]
• Contractor sentiment has cooled. In the Associated General Contractors 2026 Hiring and Business Outlook, economic slowdown or recession was the top-cited concern at 62 percent of firms, up from fifth place the prior year, while 82 percent still reported difficulty hiring hourly craft workers.[17]
Rising insolvency, tightening surety capacity, and a labor market that still constrains delivery produce exactly the kind of borrower who looks fine on a bank statement and is already being sued.
What Does Each Layer Miss on Its Own?
Understanding the gaps is what makes the sequencing decision obvious.
Why Does a Clean License Check Still Leave You Exposed?
License status is binary and current. It tells you nothing about accumulated obligations. A contractor can renew on time every cycle while quietly stacking receivables financing behind three other secured parties. License checks also have jurisdictional shape that surprises people: New York has no statewide general contractor license, with licensing handled at the city and county level, while Texas licenses specific trades through the Department of Licensing and Regulation rather than issuing a general contractor license at all.
Why Is a UCC Search Not a Lien Search?
This is the most common conflation in construction underwriting. A UCC search returns Article 9 personal property security interests. It does not return mechanics liens, which are creatures of state lien law recorded against real property, usually at the county level. It does not return judgment liens. It does not return the pending suit that will become a judgment lien in seven months. A borrower with zero UCC filings and four open mechanics lien foreclosures reads as clean on a UCC-only workflow.
Why Do Two Layers Still Leave a Gap?
Run license and UCC together and you have legal standing plus collateral position. What you still lack is the trajectory. Litigation is the forward-looking layer, because a suit filed today is a judgment lien or a settlement obligation later. That is the sequence that turns a performing file into a charge-off, and it is visible in dockets months before it appears anywhere in your borrower's financials.
Why Is New York the Sharpest Case for the Triple Check?
Coverage reality determines workflow design more than any theory about ideal underwriting. Here is the honest map.
Which Checks Actually Overlap in New York?
Cobalt Intelligence's coverage differs by product, and the differences matter:
• Secretary of State entity data. All 50 states, pulled live from the state source at request time.
• TIN and EIN verification. All states. OFAC screening is global.
• UCC filing data. Roughly 10 to 11 states, returned in the same API call as SOS data and billed as a separate credit.
• Contractor licensing. A few states only, including New York, Texas, and California.
• Court records. New York State and Miami-Dade County, Florida. That is the entire footprint. Not nationwide, not federal, and not PACER.
Lay those footprints over each other and one state is left standing. New York is the only state where SOS status, UCC filings, contractor licensing, and court records are all available from Cobalt. If you are underwriting New York construction, you can assemble the full picture from one integration. If you are underwriting anywhere else, you cannot, and you should plan accordingly rather than discover it mid-integration.
What Does the New York Overlap Let You Automate?
It lets you make the whole triple check a single pre-funding decision step rather than a mix of API responses and analyst tabs. Entity status and UCC filings return together. The license check confirms standing to contract. The court records call returns judgment details, case numbers, case types, divisions, filing dates, and parties, with amounts included where the underlying record contains them. Not every court record carries a dollar figure, and any workflow that assumes one will break on real data.
New York also happens to be where the concentration is. Roughly 80 percent of the judgments filed by Cobalt's funder customers land in New York or Miami-Dade, which is why those two jurisdictions were built first. New York is the center of alternative lending, and Miami-Dade became the second hub after a large migration of funders to South Florida during the pandemic. The coverage was chosen by looking at where customers actually file.
What Are Your Options for Running These Three Checks Today?
Before getting to any specific integration, it is worth being clear about the alternatives, because several of them are perfectly reasonable answers depending on your volume and jurisdiction mix.
What Do the Established Options Each Solve?
PACER covers federal district, bankruptcy, and appellate dockets nationally. If your primary concern is contractor bankruptcy filings, PACER is the authoritative source and there is no substitute for it. It does not cover state civil courts, which is where the mechanics lien foreclosures and most construction payment disputes actually sit.
Unicourt aggregates state and federal court data across a broad footprint with an API. LexisNexis offers deep public records coverage with the research tooling and the pricing that come with an incumbent legal data provider. CSC and Wolters Kluwer are the established names in UCC search and filing, with nationwide coverage and long-standing lender relationships. Manual courthouse search remains the fallback for any jurisdiction no API reaches, which in construction is often the county recorder where mechanics liens live.
Several funders run multiple sources at once because no single one covers the whole need. One MCA operator processing 500 files a day described the current state directly: "we run New York court separately and then we run Unicourt, and then we run UCC searches." That is three systems for one decision.
Where Does Cobalt Fit?
The differentiator is not breadth. It is integration friction inside a jurisdiction Cobalt covers. Court records, SOS, UCC, and contractor license all sit behind one API key, one billing relationship, and one credit model, at one credit per lookup, the same unit price as an SOS pull. For a funder whose New York volume justifies automating the check on every application rather than on the deep-dive subset, collapsing three vendor integrations into one call path is the value. For a funder whose book is spread across thirty states, it is not, and pretending otherwise wastes everyone's evaluation time.
Cucumber Capital put the economic condition on it clearly: "If courts are cheap enough, then it's worth it to run on every [application] automatically." That is the right test. Court data at deep-dive pricing gets run on the 10 percent of files someone flags. Court data at SOS pricing gets run on everything, which is where it actually changes outcomes.
Underwriting New York or Miami-Dade construction files and want to see what the triple check returns on your own borrowers? Book a walkthrough with the Cobalt team and run test-mode lookups before spending a credit.
How Do You Actually Call the Court Records Endpoint?
The mechanics are simple, with one design constraint that trips up first integrations.
What Does a Real Request Look Like?
curl -G "https://apigateway.cobaltintelligence.com/courtCases" \
-H "x-api-key: YOUR_API_KEY" \
--data-urlencode "businessName=Hudson Valley Structural LLC" \
--data-urlencode "jurisdiction=newYork" \
--data-urlencode "callbackUrl=https://underwriting.yourdomain.com/webhooks/court-cases"Valid `jurisdiction` values are `newYork`, `miamiDade`, `testNewYork`, and `testMiamiDade`. The two test values run the full request path without consuming credits, which means you can build and validate your entire callback handler before your first billed lookup. Use them.
How Should You Handle the Async Callback?
The endpoint is asynchronous only. There is no synchronous mode, and `callbackUrl` is required rather than optional. Typical completion runs 30 to 120 seconds because the data is pulled live from the court site at request time rather than served from a cache, so there is no staleness window to reason about. Build for it:
• Treat the initial response as an acknowledgment. Your request returns a handle, not case data. Anything that blocks an underwriter's screen on the response will time out.
• Make the callback handler idempotent. Retries and duplicate deliveries should not create duplicate case records or fire a second review task.
• Persist the raw payload before parsing. When a case type or division value shows up that your parser has not seen, you want the original response to reconcile against.
• Do not assume an amount field. Judgment amounts appear where the record contains them, and plenty of records do not.
• Set a timeout and a fallback path. Decide in advance what an underwriter does when a callback has not landed in five minutes, because that decision made in the moment is always worse.
Cobalt returns court data as a data source. It does not return a risk score, a decision, or a recommendation, and it does not monitor a portfolio for new filings on a schedule. Scoring, thresholds, and re-check cadence are yours to build, which is the correct division of labor between a data provider and a credit policy. Related patterns are covered in the construction lender court records workflow guide.
What Do You Do When Your Contractor Is Outside the Overlap?
This is the question that determines whether the check is usable for your book, and it deserves a straight answer rather than a roadmap deflection.
Which Layers Can You Still Get Everywhere?
Entity status through Secretary of State data works in all 50 states with a live pull. TIN and EIN verification works in all states. OFAC screening is global. Those three layers are available regardless of where your contractor operates, and they still catch a meaningful share of the failures: entities in bad standing, dissolved shells reactivated for a single draw, and mismatched tax identity.
UCC coverage across roughly 10 to 11 states means most funders will get partial collateral visibility and will need a second source for the rest. Contractor licensing narrows further. Court records narrow furthest, to New York State and Miami-Dade County.
How Should You Supplement the Gaps?
Objections to the coverage limit are legitimate, and two evaluators put them plainly. RoxWrite: "Most of our clients are not in New York." Elementix: "We'll stay away from the court stuff then. I wish the court stuff was rounded." Both are correct conclusions from their own book composition. If your construction volume is concentrated in Arizona and Colorado, the court records layer does not serve you today and no amount of positioning changes that.
The workable pattern for mixed books is to route by jurisdiction rather than to force one vendor onto every file. Run the checks Cobalt covers where it covers them, use PACER for federal bankruptcy nationally, and keep an aggregator or a manual search vendor for state courts outside the footprint. Broader jurisdictions are on the roadmap, including a planned human-assisted queue for unsupported jurisdictions that would take roughly an hour rather than seconds. That queue has not shipped, and you should build your current workflow as though it does not exist. Coverage details by state are in the state-by-state UCC filing coverage breakdown.
How Should You Sequence the Triple Check in an Underwriting Workflow?
Order matters, because each layer changes what the next one is worth checking.
What Order Should the Calls Run In?
Run entity status first. If the SOS record shows the entity dissolved, delinquent, or never formed in the state where the work is being performed, nothing downstream will save the file and you have spent one credit to find out. Run UCC and license next, in parallel; they answer independent questions and neither gates the other. Run court records last, and only on files that survive the first three, because it is the slowest call and the one where the async pattern adds workflow complexity.
That sequencing also matches how funders describe their own funnels. CorFinGroup, running 50 to 60 submissions a day with 10 to 15 deep dives, framed the goal as consolidation: "I'm looking for one cost effective solution for KYC and KYB." The cheap deterministic checks screen the volume. The richer checks run on the survivors.
What Should Trigger a Manual Review?
• License lapsed, suspended, or absent in a jurisdiction that requires one. In New York this directly affects enforceability of the contract and the contractor's mechanics lien rights.
• UCC filings from parties you do not recognize. Especially recent ones, and especially where the collateral description reaches receivables you assumed were unencumbered.
• Any open judgment, regardless of stated amount. A judgment with no amount in the record is not a small judgment; it is an unknown one.
• The same principal appearing as defendant across multiple entities. Repeat defendant patterns are a stronger signal than any single case.
• A fixture-heavy collateral package with only a central-index UCC filing. Your priority position may not be what your search implies.[7]
The wider case for pricing the check honestly against its cost was made by Gate Rock Capital in an evaluation call: "where I would pay $4 a pull is when you have the state index on court search." The willingness to pay tracks coverage, and coverage should be stated before pricing rather than after.
Public enforcement supplies the closing argument for why litigation exposure belongs in a pre-funding decision. In April 2026, the CEO of Prime Capital Ventures was sentenced to 97 months in federal prison over a wire fraud conspiracy tied to the collapse of his commercial lending business, with a forfeiture judgment exceeding $55 million.[18] Counterparty distress in commercial credit becomes court activity before it becomes public knowledge. The checks that catch it are the ones running before the wire goes out, which is also the argument in the construction lender license, bond, and UCC workflow guide.
Three checks, three different failure modes, one state where all three arrive from one integration. Build for the coverage you actually have.
.jpg){kind=small}
.png)