By Patrick Turcotte, COO, Docutrax. Last reviewed: 14 July 2026.

Contractual risk transfer is how an owner or general contractor arranges for a loss arising from someone else's work to be paid by someone else's insurer. It has three parts, and they are three separate promises: an indemnity clause in the contract, an insurance requirements schedule that says what coverage the other party must buy, and endorsements on that party's policy that make the coverage reach you.

Each part can be drafted correctly and still fail. This page covers what each one does, how the three fit together, and where the chain typically breaks.

What is contractual risk transfer?

Contractual risk transfer is an arrangement under which one party agrees to bear the financial consequences of a loss that would otherwise fall on another.

It works as a stack:

The indemnity clause creates the legal obligation. It is a promise from the downstream party, the indemnitor, to the upstream party, the indemnitee.

The insurance requirements schedule creates the funding. It tells the downstream party what coverage to carry, at what limits, and with what endorsements.

The endorsements on the downstream party's policy make the funding reachable. Additional insured status, waiver of subrogation, and primary and noncontributory each do a different job.

Missing any one of the three breaks the transfer. The clause without insurance is a promise from a party that may not be able to pay. Insurance without the endorsements is a policy you cannot claim against. And an unenforceable clause can take the insurance down with it, which is the mechanism most often missed.

Indemnify, defend, and hold harmless: what is the difference?

These three words usually appear together in a single sentence, and they are three different obligations. The distinction is not academic, and it is the one most often glossed over.

Indemnify means to pay or reimburse the other party for losses. The obligation typically does not arise until the end of a case, once the indemnitee has had judgment entered against it or has actually paid.

Defend is a separate and independent obligation to actively defend, or to fund the defence of, a claim brought against the indemnitee. It arises as soon as the claim is made, and it applies regardless of the merits. The allegations trigger it, not the outcome. A groundless claim can trigger a duty to defend. (Thomson Reuters; American Bar Association, Under Construction.)

Hold harmless is often treated as a synonym for indemnify, and in several states it is. Colorado, Delaware, Ohio and Louisiana read the two terms as a unit. Other states, California among them, treat them as distinct rights, and in those states the omission of hold harmless can mean the indemnifying party is not on the hook until the indemnitee has actually paid out. (Thomson Reuters.)

The practical weight sits on defend. Defence costs in construction routinely outrun the underlying claim, and the duty to defend is triggered earliest and most easily.

Two consequences follow, and both are frequently discovered late.

Under the common law of most states, an indemnitor has no duty to defend unless the contract says so. If the clause requires the other party to indemnify and hold harmless but does not require them to defend, many indemnitors will simply decline a tender of defence, and in most states they are entitled to. Some states go the other way: California, Montana, Oklahoma and South Dakota provide by statute that an indemnity agreement carries a duty to defend unless the contract negates it. (American Bar Association, Under Construction.)

The wording changes what the insurer owes. Where the indemnitor agreed to reimburse attorney fees rather than to defend, the indemnitor's insurer is not obligated to defend the indemnitee. (IRMI, In Defense of Insured Contracts.) The difference between "defend" and "reimburse the costs of defence" is one word in the contract and a very large number on the invoice.

What does an indemnification clause contain?

An indemnity clause has four moving parts, and each is negotiable.

The parties. Who indemnifies whom. In construction, typically the subcontractor indemnifies the general contractor and the owner. Watch the definition of indemnitee: it should reach the parties who actually need protection, including the owner, the construction manager, and their affiliates where the prime contract requires it.

The trigger. What has to happen for the clause to bite. "Arising out of the work" is broad. "Caused by the negligence of the subcontractor" is narrow. The gap between those two phrasings is often the entire dispute.

The scope of the obligation. Indemnify, defend, and hold harmless, individually or in combination. See above. The words are not interchangeable.

The form of indemnity. This is what state law regulates, and it is the layer most likely to be void.

Clauses fall into three functional forms:

● Broad form. The indemnitor covers the indemnitee's own negligence, even where the indemnitor did nothing wrong.

● Intermediate form. The indemnitor covers the indemnitee except where the indemnitee was solely at fault.

● Limited form. Also called comparative indemnity. The indemnitor covers only its own fault.

States treat these differently, and that is where the first failure appears.

Why is the insurance requirements schedule a separate clause?

Because an indemnity clause does not, by itself, oblige anyone to buy insurance.

A contractual duty to indemnify and hold harmless is not the legal equivalent of a duty to procure insurance covering that obligation. A subcontract can contain a perfectly enforceable indemnity clause and impose no insurance requirement at all, in which case the indemnitee has a promise and no funding behind it. Well-drafted construction contracts, including the standard AIA and ConsensusDocs forms, keep the two as separate clauses with separate headings for exactly this reason. (IRMI, In Defense of Insured Contracts.)

A working insurance requirements schedule specifies, at minimum:

● The coverages required: commercial general liability, automobile, workers compensation and employers liability, umbrella or excess, and any professional or pollution coverage the scope demands.

● The limits, per occurrence and in the aggregate, and whether the aggregate applies per project.

● Occurrence rather than claims-made coverage for general liability, so that a claim surfacing years after completion still has a policy to reach.

● The endorsements required, by ISO form numberor equivalent where possible: additional insured for ongoing operations and for completed operations, waiver of subrogation, and primary and noncontributory.

● How long completed operations coverage must be maintained after the work ends, which should track the statute of repose in the relevant state rather than the end of the project.

● The right to obtain the endorsements themselves, not only a certificate.

The last item is the one most often omitted and the cheapest to add.

Why do states limit indemnity in construction?

Legislatures have taken the view that broad and intermediate form indemnity produces bad incentives on high-hazard work: a party with no exposure to the consequences of its own negligence has less reason to prevent it.

Legislatures across the United States have grown increasingly hostile to indemnity clauses in construction agreements, and more and more prohibit broad form and intermediate form indemnity. (IRMI, Indemnity and Additional Insured Requirements.)

The specific scope varies by state and the boundaries move, so the governing state's law is a question for counsel on any given contract. The direction of travel is not in dispute.

Minnesota is a clear reported example, and it shows the whole chain failing at once. In Engineering & Construction Innovations v. L.H. Bolduc Co., decided by the Minnesota Supreme Court in 2013, a subcontractor drove sheet piling into a sewer pipeline while installing it exactly where the upstream contractor told it to, and a jury found the subcontractor not negligent. The court held the indemnity clause unenforceable, because Minnesota permits only limited form indemnity agreements by statute. (IRMI, same; Engineering & Construction Innovations v. L.H. Bolduc Co., 825 N.W.2d 695 (Minn. 2013).) And the additional insured coverage failed in the same decision: the endorsement covered the upstream contractor only for injury or damage caused by the subcontractor's acts or omissions, which the court read as requiring fault, and the jury had found none. The clause fell, the endorsement gave nothing, and the certificate had shown a clean additional insured grant throughout.

The Minnesota legislature then went further. Before 2013, the statute's insurance-procurement exception let parties reach the same broad indemnity result through an agreement to insure, and the court in Bolduc observed that the narrow exception had swallowed the rule. Effective August 1, 2013, amended section 337.05 voids any provision requiring a party to provide insurance coverage to other parties for those parties' own negligence or intentional acts, with a carve-out for insuring vicarious liability. The insurance workaround was closed by statute. (Minn. Stat. 337.02, 337.05.)

Texas closes it even more directly, and Texas is the market where this matters most. The Texas Anti-Indemnity Act, Insurance Code chapter 151, effective January 1, 2012, voids broad and intermediate form indemnity in construction contracts, and section 151.104 extends the prohibition to the insurance: a contract provision requiring additional insured coverage, and any endorsement or provision within the policy providing it, is void and unenforceable to the same extent as the indemnity it mirrors. The statute reaches into the policy itself. There is one deliberate carve-out, section 151.103, which preserves indemnity and additional insured coverage for claims of bodily injury or death of the indemnitor's own employees, the classic action-over scenario, precisely because those claims are why the protections exist. (Tex. Ins. Code 151.102 to 151.104.)

The consequence for a risk transfer program is structural. In most states, additional insured status is the independent path that survives when the indemnity clause falls. In a minority of states, Texas foremost among them, the statute limits the additional insured coverage too, and the fallback is only as good as the jurisdiction allows. Which state's law governs is not a drafting detail. It determines whether the chain has one anchor or two.

What is an "insured contract," and why does it decide the outcome?

This is where contract law and insurance meet, and it is the part of the mechanism that is invisible on a certificate.

A standard commercial general liability policy insures a contractual liability the insured has assumed, but only where the contract is an "insured contract" as the policy defines it. The definition sits inside the policy form, at Section V, item 9. Indemnity agreements have historically fallen within it.

Two separate things can take an indemnity clause outside that definition.

The statute. Where an indemnity clause is prohibited or void by statute, it no longer falls within the definition of insured contract. Insurance coverage is therefore no longer directly available for the indemnity agreement. (IRMI, Indemnity and Additional Insured Requirements.) The anti-indemnity statute does not merely void the clause. It removes the insurance behind it.

The endorsement. The CG 21 39 endorsement, often attached to contractors' CGL policies, removes coverage for indemnity agreements by deleting them from the policy's list of insured contracts. A policy endorsed with CG 21 39 simply does not insure the contractor's indemnity agreement, regardless of what the state does with the clause. The CG 24 26 endorsement does something narrower: it removes coverage for broad form indemnity, leaving the insured covered only for losses caused in whole or in part by the subcontractor. It was revised in ISO's 2013 CGL revisions to clarify that indemnity agreements are insured contracts only to the extent the agreement is permitted by law. (IRMI, same.)

Two things must therefore be true for an indemnity clause to be insured: the clause must be permitted where it applies, and the policy must not have deleted indemnity agreements from its insured contract definition.

A certificate of insurance confirms neither. It does not describe the subcontract, it does not describe the statute, and it does not list the endorsements an underwriter attached to take coverage away.

If the contract already indemnifies you, why also require additional insured status?

Because they pay for different things, and one of them covers a gap the other leaves open.

Defence costs. Additional insured status gives you direct access to the other party's policy as an insured in your own right, and a CGL policy provides uncapped defence costs for an additional insured that do not erode the policy limits. Where you rely only on contractual liability coverage, the coverage that backs the indemnity clause, defence costs are generally paid within the limits, unless the narrow supplementary payments conditions are met, same suit, same counsel, no conflict, which they frequently are not. Defence and indemnity together are capped, and available indemnity shrinks as the defence runs. (IRMI, Indemnity and Additional Insured Requirements.)

The action-over problem. For the indemnitor's insurer to owe a defence to the indemnitee under the indemnity clause, the indemnitor and the indemnitee generally have to be named in the same suit. In the most common construction scenario they are not. A subcontractor's employee is injured, and sues the general contractor. The employee cannot sue their own employer, because workers compensation is the exclusive remedy. The subcontractor is therefore never named in the suit. The general contractor can still enforce the clause by bringing the subcontractor in as a third-party defendant, which most states permit on a written indemnity notwithstanding workers compensation exclusivity. But that route means litigating against the subcontractor, proving the indemnity, and funding a defence in the meantime. (IRMI, In Defense of Insured Contracts.)

Additional insured status resolves that, because it does not depend on the subcontractor being sued or impleaded: the general contractor tenders directly to the subcontractor's carrier and the defence obligation is immediate. It is the reason the endorsement is required alongside the clause rather than instead of it. It is also why the Texas statute's employee-injury exception exists: the action-over claim is the one place the Texas legislature deliberately left both the indemnity and the additional insured coverage intact.What survives what. In some states the duty to defend can survive even where the indemnity clause is struck down by an anti-indemnity statute, on the reasoning that the duty to defend is independent of and broader than the duty to indemnify. Massachusetts courts have taken this position. (American Bar Association, Under Construction.) This varies, and it is a question for counsel, but it is worth knowing that the two obligations do not always fall together.

Additional insured status has its own failure modes: the endorsement's scope, its edition, and the entity named on it. Those are covered in the insurance endorsements guide.

What does verifying contractual risk transfer actually require?

Three documents, read against each other.

The subcontract. The indemnity clause, the form of indemnity, whether a duty to defend is expressly stated, and the insurance requirements schedule.

The governing state's law. Whether the form of indemnity used is enforceable where the work is performed, and whether the state is one of the minority whose statute also limits additional insured coverage. This is a legal question, and it belongs with counsel.

The policy. The insured contract definition, any endorsement that has narrowed or deleted it, the additional insured endorsement and its scope, and the exclusions that apply to the work.

A certificate check does not reach any of them. An endorsement check reaches the third partially: it examines the endorsements a broker or agent listed, not the endorsements an underwriter attached to remove coverage. Finding a CG 21 39, or a CG 24 26, or an insured contract definition that no longer reaches the clause, requires reading the policy.

That is a Comprehensive Policy Review, and Docutrax runs them on high-exposure work. Policies are reviewed by licensed P&C professionals, and construction accounts are handled by CRIS-certified Account Managers, on a base of more than 300,000 complete policy reviews since 2017. Where a policy carries an endorsement the contract did not anticipate, the finding is documented and the correction coordinated with the broker or agent. The facts go back to the client, and the client decides.

For lower-exposure relationships, verifying the certificate and the key endorsements is proportionate. The depth of review should track what the relationship can cost.