Insurtech Reprices Small-Business Risk Without Owning the Loss
Insurtechs selling small-business coverage rarely hold the risk they price. Most operate as managing general agents, collecting fees while reinsurers absorb the loss. That structure explains why a sleek quote flow can coexist with ugly loss ratios, and why the pricing inputs that matter have not changed as much as the marketing suggests.
The Risk Carrier That Never Pays
A managing general agent underwrites policies, issues them in a fronting carrier's name, and cedes most of the risk to reinsurers. The insurtech keeps a commission on premium and often a profit share if the book performs. It does not keep the float, and it does not absorb the tail. When a hailstorm hits a strip mall in Texas, the MGA's software does not pay that claim. A reinsurance treaty does.
This matters for small-business lines because the losses are lumpy and correlated. A business owner's policy bundles general liability and property coverage, so one weather event can trigger hundreds of claims across a single book. Workers compensation adds medical severity that develops over years. Commercial auto brings litigation. An MGA that prices these lines with a thin capital base is really a distribution and underwriting shop renting someone else's balance sheet.
Embedded coverage pushes the same model into point-of-sale channels. A payroll platform or a booking tool offers coverage at signup, and the MGA handles the underwriting behind it. The distribution is genuinely cheaper. The loss is not. A related piece on this site, Embedded Workers Compensation Pricing Rests on Payroll Data Feeds, makes the point that the pricing still depends on the quality of the payroll data feeding the model, not the elegance of the checkout screen.
How the MGA Stack Actually Works
Four parties sit in a typical stack. The insurtech originates and underwrites. A fronting carrier lends its licensed paper and takes a fee, often in the low single digits as a percentage of premium. One or more reinsurers provide capacity and hold the loss. The policyholder pays premium and hopes the claims process works.
The MGA's economics depend on the commission and profit share, not on the loss ratio directly. That creates a subtle incentive problem. A book with a 70% loss ratio might still pay the MGA well if the commission is 20% and the fronting fee is modest. A book at 95% wipes out the profit share and can trigger a treaty reset. The reinsurer feels that first, which is why capacity negotiations get tense after a bad cat year.
Fronting carriers have failed before when the risk they fronted turned out worse than described. State guaranty associations exist for exactly that scenario, and this site has covered the limits of that backstop in State Guaranty Limits on Long-Term Care Insurer Failures. Small-business books are smaller and more diversified than a long-term care block, but the principle holds: the entity that holds the loss is the one that matters when the model is wrong.
The fee structure can also mask adverse selection. If the MGA earns a percentage of premium, it has an incentive to grow the book even when the underlying risk is deteriorating. Reinsurers have learned to watch for this and often impose underwriting guidelines or premium caps. The tension between growth and discipline is not new, but it plays out faster in a digital distribution channel.
Documented Incidents and Their Lessons
Parametric insurance offers the clearest documented cases. The Caribbean Catastrophe Risk Insurance Facility pays governments after a storm when wind speed or a modelled loss estimate crosses a threshold. The African Risk Capacity does something similar for drought, using rainfall and satellite data to trigger sovereign payouts. Both have settled claims faster than traditional indemnity processes, which can take months of adjustment.
A coral reef policy in Quintana Roo, Mexico, is a smaller and more unusual example. It pays out when wind speeds and storm surge cross agreed thresholds, funding reef repair after a hurricane. The trigger is objective and the payout is quick. The trade-off is basis risk: if the storm damages the reef but the measured wind speed stays below the threshold, the policy pays nothing.
That trade-off is the whole story of parametric triggers. They remove the adjuster and the dispute, and they replace both with a definition. A business that needs cash within days after a named storm may find that acceptable. A business whose loss does not match the trigger definition will find the policy useless. The speed is real. So is the mismatch risk.
Where the Hype Outruns the Loss Data
AI underwriting gets marketed as a way to price small-business risk more accurately. The problem is data thickness. A regional insurer may have decades of claims on restaurants and contractors. A new entrant has a few years at best, often skewed by a benign catastrophe period. A model trained on thin data can look precise and be wrong in ways that only show up after a bad loss year.
Embedded coverage has a similar gap. Selling a policy at the point of sale is easy. Retaining the customer past the first renewal is not. Churn on embedded small-business books is not well documented publicly, and the acquisition cost only pays back if the policy stays on the books long enough. A book that turns over every 14 months behaves very differently from one that renews for five years.
Workers compensation still needs human adjusters for complex claims, and commercial auto telematics adoption remains uneven across fleets. Telematics can improve loss experience for long-haul operations with consistent drivers. A small delivery fleet with high turnover may see less benefit. The technology is real. The improvement is conditional.
What the Pricing Inputs Really Move
Loss ratios vary by line far more than by platform. A well-run business owner's policy book might run in the 50s to 60s. Workers compensation can sit lower or higher depending on mix and state. Commercial auto has been deteriorating in several markets. The platform does not change these fundamentals, though it can change how quickly a book reaches them.
Reinsurance cost is the input that moves MGA economics most. After a heavy catastrophe year, reinsurers raise attachment points and rates. The MGA's commission may stay the same, but the profit share becomes harder to earn. Some MGAs respond by tightening underwriting. Others shrink. The ones that survive usually have a differentiated data set or a niche where they understand the loss better than the reinsurer.
Cat modelling for parametric triggers is its own discipline. The model defines the trigger, and the trigger defines the payout. A poorly specified model can produce a policy that pays too often or almost never. Expense ratios for these operations often land in the 20% to 30% range, which leaves less room for error than the marketing implies.
Interest rates and investment income also matter, though less directly for MGAs that do not hold float. For carriers that do, a higher rate environment can offset a deteriorating loss ratio. For MGAs, the effect is indirect: reinsurers may demand higher returns, which tightens terms. The cost of capital flows through the entire stack.
Practical Steps for Brokers and Buyers
Ask who holds the loss before you ask about the app. The reinsurance treaty and the fronting carrier's rating matter more than the quote flow, because they determine whether the claim gets paid after a bad year.
Read the trigger definition in any parametric policy. Find out what measurement source is used, what threshold applies, and what happens if the event occurs but the measurement misses. A policy that pays on a modelled loss estimate behaves differently from one that pays on a physical wind gauge.
Request the reinsurance backing and the carrier's financial strength rating. An MGA with a single reinsurer and a thin fronting arrangement carries more counterparty risk than one with a diversified panel and a well-rated carrier.
Model churn before you count on embedded distribution. If the average policy life is short, the acquisition cost may never be recovered, regardless of how cheap the initial quote looks.
Demand loss ratio transparency, not feature lists. A carrier that will share its loss ratio by line and by vintage is telling you something. One that will not is telling you something too.
This article is informational and does not constitute personalised insurance, legal, or financial advice. Coverage terms and pricing vary by jurisdiction and circumstance.
