WYEA / Specialty insurance / Binding authority
Specialty insurance
Binding authority and program agreements across carriers
By WYEA · Published September 27, 2026 · Updated September 27, 2026
To know what each carrier has authorized, read its signed agreement together with every addendum that amends it, and quote the clause. The document engine we build does that reading across all of an MGA's carrier agreements, and returns each answer with the agreement and clause it came from.
What a binding authority or program agreement sets out
A binding authority agreement delegates underwriting authority from a carrier to an agent. At Lloyd's, the carrier side is a managing agent acting for a syndicate, and the agent is a coverholder. In US program business the same arrangement is usually called a program agreement, signed between an MGA or program administrator and an insurer. That insurer is often a fronting carrier that cedes most of the risk to reinsurers.
Delegation carries a large share of the business at Lloyd's. Lloyd's says delegated underwriting accounts for "approximately 45% of the market's premium income."
The agreement usually has a main body and several schedules. The schedules hold the terms an underwriter works to each day: the classes that may be written, maximum limits, territories, the rating basis, the premium the agent may write in a year, the risks that must be referred before binding, and the policy forms and endorsements the carrier has approved. Claims authority, bordereaux reporting and the carrier's audit rights sit in the main body or in schedules of their own.
In the United States, the NAIC's model Managing General Agents Act requires a written contract between the MGA and the insurer. Section 4 lists the underwriting guidelines it must contain, including the maximum annual premium volume, the basis of rates, the types of risks, maximum limits of liability, exclusions, territorial limitations, cancellation provisions and the maximum policy period. Section 5 requires the insurer, at least semi-annually, to "conduct an on-site review of the underwriting and claims processing operations" of the MGA.
Why several carriers make the terms harder to hold
Many MGAs write the same class through more than one carrier. A cyber program might be split between two carriers by size of risk, with a third added for capacity at the top of the book. Each carrier negotiates its own agreement, so the referral limit, the excluded territories and the approved wording can all differ for the same class.
The agreements also change during the year. An addendum raises the limit for one class, or replaces the approved cyber form with a new edition from a stated date. The signed agreement and its addenda are the terms. The summary grid on the team's shared drive is someone's reading of them, and it drifts as addenda arrive.
Errors from that drift tend to surface much later, in the carrier's audit or in a claim. A typical one is a policy issued on the wrong carrier's endorsement, or on a form edition that an addendum had already replaced.
Reading each agreement with the addenda that amend it
The engine reads the agreements, schedules, addenda and approved forms where your team already keeps them. Dropbox is the first connector, and your document system stays the system of record.
It works out which addendum amends which agreement, and which issued policy was written under which agreement, and it keeps that. An answer about a referral limit reflects the schedule as amended and quotes the clause. Every line opens the document it came from. A passage has to be found in one of your documents before it is written as an answer. An addendum that is a signed scan with no readable text goes to a person, unanswered.
any risk with a limit above USD 5,000,000 shall be referred to the CompanySchedule C, Referrals, as amended by Addendum 2 · effective 1 Jul 2026
limits exceeding USD 3,000,000 require the prior written approval of the CompanySchedule B, Referral Conditions · effective 1 Jan 2026
Addendum 1 changes the cyber referral limit, but it is a signed scan with no readable text. It is on your list, unanswered.
The carriers, clauses and dates in this example are invented. No client material appears anywhere on this site.
The engine reports what each agreement says. Whether a risk falls within authority, and whether to bind or refer it, stays with your underwriters. The engine does not issue, bind or send anything, and it makes no underwriting, pricing or claims decisions.
Drafting on each carrier's approved forms
The engine drafts from your own precedents, given the deal inputs and the precedent to start from. The same deal can be drafted on the forms of each carrier entity you write through, so a risk moved from one carrier to another starts on the second carrier's approved wording.
Where the carrier's approved forms and your house rules cover a point, the draft follows them. Language the engine suggests beyond them is labeled until your team adopts it. When a reviewer approves or corrects a clause, the next draft follows that choice. What the engine learns is kept as rules and records in your own system, and none of it trains a model. Drafts stay drafts until a reviewer approves them.
Carrier audits and the record behind an answer
Carriers audit the agents they delegate to, and the NAIC model act sets a minimum frequency for it. Many audit questions are about documents: what the agreement allowed on the date a policy was bound, and which form edition went out with it.
An answer from the engine points at the document as it read that day, and later edits do not change it. Copies of the pages you relied on are kept past your storage provider's file history. Every answer records who asked, who approved it, and when. We do not yet offer a view of that record for your team to open on its own. The security page lists what we have today and what we do not.
When a shared product is the better fit
If your program writes on one carrier's standard forms under a short agreement, a shared product will usually serve you better, and we say so on the first call. Bordereaux validation and premium accounting belong in delegated authority software built for that work. The engine reads and drafts documents.
See it on your own documents
Book a consultation or demo. After an NDA, we build a fixed-price, one-week prototype on your binding authority and program agreements, their addenda and the forms each carrier has approved. If a build follows, it is quoted at a fixed price, with deliverables in writing, before work starts.
Start a conversation
Tell us which documents take your team the longest.
Thirty minutes. Book it as a consultation on whether any of this is worth doing, or as a demo on your own documents. Say which when you book, or decide on the call.
Thanks. Your message is on its way. We'll reply within one business day.