The cost of saying yes

Delegated authority has become the growth engine of specialty insurance, yet the capacity behind it still pays to trust every agency from scratch. Change that fixed cost and the appeal and the economics of delegation change with it.

Two earlier papers in this series looked at capacity from the outside. Perspective 05 told coverholders that their line is priced on what it costs to trust them. Perspective 13 argued that the market should want founders, because the cost of trusting them is falling. This paper sits on the other side of the table and works through the capacity provider’s own economics: what a delegation costs to hold, why that cost has been rising, and what changes when the controls come built into the agency rather than inspected into it afterwards.

01  The engine capacity cannot do without

Delegated authority is no longer a distribution channel at the edge of the market. It is where the growth is.

Lloyd’s puts the share of its premium written through delegated authority at about 45 per cent, through 3,015 coverholders at the end of 2025, in a market that wrote £57.9 billion of gross premium that year. In the United States, Conning estimates managing general agents wrote US$128 billion in 2025, with statutory premium growing about 12 per cent against roughly 5 per cent for property and casualty as a whole; Gallagher Re puts MGAs at about one-eighth of the US market. In Australia, underwriting agencies write around $10.8 billion.

Capital has followed the flow. Premium ceded to reinsurers by the US program carriers Gallagher Re tracks reached US$21.2 billion in 2025, up 59 per cent in two years, and sidecars and private capital now build vehicles specifically to follow delegated portfolios. Fronted premium passed US$22 billion. The appetite for delegated business has rarely been stronger.

And yet, in the same period, 62 per cent of US MGAs told Conning that capacity for new programs was increasingly hard to find. AM Best describes capacity providers as “more selective”. Both things are true at once: capital wants delegated premium in aggregate, and hesitates over each individual agency. That hesitation is not caprice. It is arithmetic.

02  The economics of a delegation

From the capacity provider’s side, a binder is a portfolio of risks written by someone else, and its profit is what is left after four deductions: acquisition cost, claims, the cost of capital held against the portfolio, and the cost of overseeing the firm holding the pen. The first three scale with premium. The fourth mostly does not.

Oversight is a cost per relationship. Reviewing a new agency’s wording, rater, systems and controls costs much the same whatever the size of the agency’s portfolio. On that arithmetic, a small agency is rarely a bad risk; it is an expensive relationship. The capacity provider’s rational responses are to back fewer and larger agencies, to load the terms of the smaller ones, or to wait until an agency has a history. Each response is sensible for one provider and costly for the market, because specialist appetite usually starts small.

The costs of weak oversight are also real. AM Best found affiliated programs to be the third-leading cause of US insurer impairments between 2000 and 2022. Lloyd’s expense ratio rose to 35.6 per cent in 2025, which Lloyd’s attributed to commissions and acquisition costs among other things. And Moody’s warned this month that insurers leaning heavily on third-party underwriters risk becoming “commoditised capital providers”, with intermediaries taking a growing share of the profit. The capacity provider’s defence against all three is the same: knowing, from evidence rather than assurance, what is being written in its name.

03  The regulatory ratchet

The cost per relationship has been rising, and the direction is set.

In Australia, APRA’s Prudential Standard CPS 230 took effect on 1 July 2025. It requires insurers to treat providers of underwriting and claims services as material service providers unless they can justify otherwise, to keep a register of those providers, and to notify APRA within 20 business days of entering into or materially changing an arrangement that supports a critical operation. Existing contracts had to comply by their renewal or 1 July 2026, whichever came first. Every Australian binder now sits inside the insurer’s operational-risk framework, with the evidence that implies.

At Lloyd’s, the Coverholder Reporting Standards have set a mandatory data core for binding authorities since 2017, now at version 5.2. The Lloyd’s Market Association’s audit scope has widened what a coverholder must be able to evidence. And Lloyd’s own 2025 annual report records that Blueprint Two “has not yielded the benefits that were originally envisioned”, with parts of its original vision sunset in favour of common data standards and interoperability. The practical consequence is that delegated-authority data and controls are the responsibility of individual firms, not of shared market infrastructure.

More relationships, each more expensive to hold: that is the squeeze on the capacity side of delegated authority.

04  What good oversight asks for

Across the Lloyd’s code of practice for delegated underwriting, the reporting standards, the audit scope and CPS 230, the substance of good oversight comes down to a short list. Authority is enforced before a risk is bound, not reviewed after. Pricing can be reproduced from its inputs. Changes to rates and wordings are controlled and dated. Data arrives in a standard form, on time and complete. Licences and sanctions are checked. Every decision leaves a record. And the arrangement can be exited or taken over without losing the portfolio’s history.

None of that is new. What is expensive is how it is usually verified: by sampling, after the event, one agency at a time, through audits and bordereau queries that re-establish facts the agency’s systems should have held in the first place. The cost of oversight is largely the cost of reconstructing evidence.

05  Building the evidence in

Cuttleflow is designed so that the evidence is a by-product of the work. In that design, every agency on the platform trades on the same controls from its first risk, and none can be switched off by the agency.

Authority is encoded. Binder terms — products, territory, limits, rate authority and referral triggers — are set before the agency can trade. A risk outside authority is referred or declined; it cannot be bound.

Decisions are deterministic. Every price, eligibility decision, referral and bordereau line comes from a versioned rules engine or from a person with written authority. Artificial intelligence reads and drafts; it never sits on the bind path. Any premium can be reproduced from its inputs and the rule version in force.

Products are standard. Rating logic, coverage map, referral rules and bordereau mapping are authored and versioned by Cuttleflow. The agency configures values within closed parameter classes; it cannot write a rule, a rate or a clause.

Configuration is proven and signed off. An automated test rates every combination of factors and every extreme of every adjustment range before activation. The agency’s signatory attests that the configuration sits within its binding authority, Cuttleflow checks the binder evidence, and the configuration is frozen. The capacity provider receives the frozen configuration and the rating proof at activation.

Monitoring is continuous. Licence status is re-verified against the ASIC register, sanctions screening runs at pre-bind, and every state change and attestation is written to a hash-chained ledger before it takes effect.

Bordereaux are an output. Risk and premium bordereaux are produced in Lloyd’s Coverholder Reporting Standards v5.2 from the same data that rated and bound each risk, and export is blocked while any validation fails.

The effect on the cost table in section 02 is specific. Due diligence moves from each agency to each product template: a capacity provider that has reviewed the management liability template for one agency has reviewed it for every agency on it, and what remains to assess is the underwriter, the binder and a configuration snapshot. File review becomes a query against a record rather than a reconstruction. Bordereau corrections fall because the data that produced the bordereau is the data that bound the risk. The rows that scaled with the number of agencies start to scale with the number of templates instead.

06  Origination: seeing agencies before they choose

The second change is less discussed, and for a capacity provider building a delegated portfolio it is as valuable as the first. Capacity has the same problem every supplier to underwriting agencies has: finding good agencies before they commit to someone else. By the time a new agency is visible to the market, it has usually chosen its partner, and its terms were set in a conversation the next provider was not in.

The design lets an agency join Cuttleflow, configure its product and pass the rating proof before it holds a binder. At that point it has been verified against the Australian Business Register and the ASIC register, has chosen its product, has set its limits and appetite within the template, and has shown that its configuration works. With the agency’s consent, Cuttleflow would share that profile with capacity providers. The capacity provider sees a verified, configured agency at the moment it is looking for capacity. The placing broker negotiates and places the binder in the usual way.

07  The conclusion

The appeal of delegated authority to capacity has always been access: to specialist judgement, to distribution and to risks an insurer would not see on its own. Its drawback has been the cost of trusting each firm it delegates to, a cost that falls hardest on the smallest and newest agencies, where the specialist appetite usually is.

When controls are built into the agency from day one and the product is standard, that cost stops scaling with the number of agencies. A portfolio of many small specialists becomes diversification rather than overhead. Oversight becomes a matter of reading a record rather than rebuilding one. And the capacity provider can compete for an agency at formation, when the terms are set, rather than after it has signed elsewhere. The price of saying yes falls, and delegation becomes a better trade for the party whose capital stands behind it.

Sources & basis. Lloyd’s, Delegated authorities (delegated share of premium, about 45%); Lloyd’s 2025 Annual Report (3,015 coverholders; expense ratio 35.6%; Blueprint Two); Lloyd’s full-year results 2025 (GWP £57.9bn). Conning, US MGA market studies, July 2025 and July 2026 (US$128bn in 2025; spring 2025 survey: 62% find capacity for new programs increasingly difficult, down from 71%); Conning, fronting study, August 2026 (fronted premium above US$22bn), as reported by Reinsurance News and Artemis. Gallagher Re, 2026 MGA Market Report, July 2026 (about 12.5% of US P&C; program-carrier ceded premium US$21.2bn, +59% since 2023), as reported by Reinsurance News. AM Best, reports of May 2024 and June 2026 (affiliated programs as third-leading cause of US insurer impairments 2000–2022; “more selective” capacity), as reported by Reinsurance News. Moody’s, delegated underwriting report, September 2026, as reported by Reinsurance News. APRA, Prudential Standard CPS 230 Operational Risk Management and APRA implementation guidance. Lloyd’s Coverholder Reporting Standards v5.2; Lloyd’s Code of Practice — Delegated Underwriting; Lloyd’s Market Association coverholder audit scope. Australian underwriting agency premium: APRA-published statistics as cited in Perspectives 12 and 13 of this series. Market figures are as published by the sources above at the date of this paper. The description of Cuttleflow in sections 05 and 06 is a design claim, not a description of a live system or any named party’s capability. Companion papers: Perspective 05 (what capacity rewards), 08 (the end of small), 13 (the market that wants founders) and 16 (the authorised representative, turned toward underwriting). Nothing in this paper is financial product advice.

Cuttleflow Systems · Perspective 18 · 33°53′S · 151°16′E · Sydney