In the specialist insurance world, a managing general agent (MGA) can write hundreds of policies on a pricing model that fits on a single laptop. One London MGA did exactly that: it bound roughly 300 professional liability policies using an Excel spreadsheet built by a single actuary. The model was not peer-reviewed, its assumptions were opaque to brokers and buyers, and when two large claims hit in year three, the pricing buffer evaporated. This is not a story about fraud or malice—it is a story about how the MGA model, for all its speed and flexibility, can produce fragile pricing structures that look stable until they are not.

The Single-Spreadsheet Underwriter: How One Excel Model Became an MGA's Pricing Engine

The MGA in question, a London-based specialist writer, launched a professional liability line in 2019. To price the risk, the firm engaged a single consulting actuary—a respected practitioner with decades of experience but working alone. The actuary built a model in Excel that projected claim frequency, severity, and expense loads. The model was then used to rate each policy, with adjustments for industry class and revenue band.

By early 2023, the MGA had bound approximately 300 policies. Brokers and insureds were given pricing indications but no visibility into the underlying model. The actuary's work was not independently audited, and the MGA's small underwriting team relied on the output without a second actuarial opinion. As one market observer put it, the model was treated as a black box—trusted because of the actuary's reputation, not because its assumptions had been stress-tested.

The model assumed a 5% annual claim frequency, a severity curve based on three years of UK data, and fixed expense loads of 35% of premium. There was no tail factor for long-tail claims, and the reinsurance structure—a simple excess-of-loss treaty—was not modeled under different loss scenarios. The MGA's carrier partner, a mid-tier Lloyd's syndicate, did not require model pre-approval or independent validation. The pricing engine was, in effect, a single point of failure.

This example is not unique. A similar pattern appears in other lines: a London marine syndicate's collision model repriced one cargo fleet into two treaty layers, and a telematics rewrite changed how a state regulator priced personal auto risk. In each case, a single analytical tool drove significant premium decisions with limited external review.

Why Professional Liability Invites This Fragile Structure

Professional liability is a line of business that tempts MGAs toward thin actuarial modeling. Claims are rare—typically single-digit percentages of policies in force—but when they occur, they can be severe. A single misdiagnosis, missed deadline, or flawed audit can produce a loss that exceeds the premium from an entire book. This low-frequency, high-severity profile makes statistical modeling difficult because the data set is sparse, especially for niche professions.

Data scarcity means that actuaries must rely on assumptions about tail risk, correlation, and trend. In the London MGA's case, the severity curve was based on only three years of UK data, which may not reflect the full distribution of large claims. As one actuarial consultant noted, three years is barely enough to calibrate a frequency model, let alone a severity tail. The model also lacked a tail factor—an adjustment for claims that take many years to settle, such as those involving construction defects or financial advisory errors.

MGAs compete on speed and service. A broker placing a professional liability risk wants a quote in hours, not weeks. This pressure encourages MGAs to use models that are quick to run but may not be thoroughly validated. The carrier partner, often a larger insurer or Lloyd's syndicate, may have limited actuarial resources to audit every MGA's model. In practice, many carriers rely on the MGA's own underwriting judgment and historical loss ratios, not on independent model review.

Reinsurers, too, rarely vet the underlying pricing models of MGAs. They see aggregate data—premium, loss ratio, exposure—but not the actuarial assumptions. As long as the loss ratio stays below 40%, there is little incentive to ask questions. But when claims spike, the model's weaknesses become apparent, and by then, the policies have already been bound.

The Actuary Behind the Model: A Named Practitioner's Approach

The actuary who built the model, a fellow of the Institute and Faculty of Actuaries with over 20 years of experience, followed a standard approach for professional liability pricing. He collected data from publicly available loss databases and supplemented it with the MGA's own limited experience. He built a frequency-severity model using a Poisson distribution for claims counts and a lognormal distribution for severity. The model was implemented in Excel with Visual Basic macros to automate the rating engine.

Key assumptions included a 5% annual claim frequency and a severity distribution with a mean of roughly £150,000 and a standard deviation of £400,000. Expense loads were fixed at 35% of premium, covering commission, overhead, and profit. The model did not include a tail factor—an adjustment for the long tail of professional liability claims, which can take 5–10 years to close. The reinsurance structure was a simple excess-of-loss treaty with a £500,000 retention and a £5 million limit, but the model did not simulate how that structure would perform under different loss scenarios.

The actuary's work was not peer-reviewed. The MGA's underwriting team had no actuarial background and accepted the model as sound. When asked about validation, the actuary noted that his firm had used similar models for other clients without issues. But the model had never been stress-tested for correlation between claims, changes in legal environment, or economic cycles. As one industry analyst remarked, the model was built for a stable world, not for the one we live in.

This reliance on a single actuary is not uncommon in the MGA space. A German hospital group's surgeon fee schedule similarly rewrote national health fund reimbursement tables, highlighting how a single analytical framework can reshape pricing across a market. In both cases, the model's creator was respected, but the lack of independent review created hidden vulnerabilities.

What Happens When the Model Meets the Real World

For the first two years, the MGA's professional liability book performed well. Loss ratios stayed below 40%, and the carrier was pleased with the profitability. The MGA expanded its offering to new professions, including architects and management consultants. The model's assumptions appeared validated. But in year three, two large claims hit: a £1.2 million settlement for a construction design error and a £900,000 judgment for an investment advisor's negligence. Combined with a handful of smaller claims, the loss ratio jumped to roughly 75%.

The MGA had set aside a claims buffer—about 10% of premium—but it was exhausted by the two large claims. The reinsurance treaty was triggered, but the excess-of-loss structure meant the carrier retained the first £500,000 of each claim plus a share of the layer. The carrier's overall loss ratio for the book rose above 60%, triggering a model audit requirement in the reinsurance agreement.

The audit, conducted by a different actuarial firm, found that the original model had omitted correlation assumptions between claims. In professional liability, claims are not independent—a recession can trigger a wave of negligence suits, and a legal precedent can increase severity across the board. The model also lacked a tail factor, meaning it underestimated the ultimate cost of claims that take years to settle. The auditor estimated that the model underpriced the book by roughly 15–25% on a discounted basis.

The carrier demanded premium adjustments, which led to broker pushback. Some brokers threatened to move their business to other MGAs. The MGA revised its rating engine, adding a correlation factor and a tail factor, but the damage to its reputation was done. Several policies were non-renewed, and the book shrank. The actuary's model, once the engine of growth, became a liability.

Regulatory Gaps: No One Checks the Spreadsheet

The UK Financial Conduct Authority (FCA) does not require MGAs to submit their pricing models for pre-approval. Nor does it mandate independent peer review of actuarial models used by MGAs. Oversight depends on the due diligence of the carrier partner—the insurer that provides the paper and assumes the ultimate risk. But carriers vary widely in their actuarial capacity. A large Lloyd's syndicate may have a team of actuaries, but a smaller carrier may rely on the MGA's own numbers.

Lloyd's managing agents are subject to the Lloyd's Actuarial Function, which requires model validation and independent review. However, many MGAs operate outside Lloyd's, under delegated authority agreements with standard insurers. These agreements often include clauses about pricing adequacy, but they rarely specify the level of actuarial rigor required. As one regulatory expert put it, the system relies on market discipline—carriers are supposed to police their MGAs, but they have limited tools to do so.

The gap is especially wide for professional liability, which is not a regulated line in the same way as motor or workers' compensation. There are no standard filing requirements or rate approvals. The model that prices a professional liability policy can be as simple or as complex as the MGA chooses. The buyer—the insured—has no way to evaluate whether the premium reflects the true risk.

Some market participants argue that regulation should be tightened. The International Association of Insurance Supervisors has issued principles on model governance, but they are not binding. Others counter that more regulation would stifle innovation and raise costs. The MGA model thrives on flexibility, and requiring pre-approval of every pricing model would slow the market. The trade-off is that some models will be flawed, and some buyers will pay too much or too little.

Lessons for Buyers: What to Ask Before You Bind

For a risk manager or broker placing professional liability coverage, the story of this MGA offers practical lessons. First, ask about the pricing methodology. Is the model built by a single actuary or a team? Has it been independently reviewed? The answer may be vague, but the question signals that you are paying attention. Second, ask the carrier about its own actuarial involvement. Does the carrier have a dedicated actuary overseeing the MGA's book? If not, that is a red flag.

Third, compare loss ratios across peer MGAs. If one MGA consistently reports loss ratios below 30% while competitors are at 50%, it may be under-pricing risk. Low loss ratios are not always a sign of good underwriting—they can indicate inadequate pricing that will eventually correct. Fourth, consider whether binding authority is the right structure for your risk. Admitted market carriers are subject to more regulatory scrutiny and may have more robust pricing models.

Finally, ask whether the model has been stress-tested. A good model should simulate scenarios—a recession, a spike in claim frequency, a change in legal environment. If the MGA cannot provide a stress-test summary, the model may not be robust. The buyer does not need to see the spreadsheet, but they should understand the assumptions and the limits of the analysis.

This case is not an indictment of all MGAs. Many operate with strong actuarial governance and transparent pricing. But the story of the single-spreadsheet underwriter shows what can happen when speed outpaces rigor. The next time you receive a professional liability quote, it is worth asking: who built the model, and has anyone checked it?

Counter-Arguments and Trade-Offs: The Case for Speed

Not everyone agrees that the MGA's approach was reckless. Some market participants argue that the flexibility of the MGA model is precisely what allows niche risks to be priced at all. A fully validated actuarial model with peer review and regulatory filing can take months to develop and cost tens of thousands of pounds. For a small MGA writing a few hundred policies, that expense may be prohibitive. The trade-off is that some models will be less rigorous, but the alternative might be no coverage at all for certain professions.

Proponents of the MGA model also point out that even large carriers with sophisticated actuarial departments have made pricing errors. The difference is that carriers have deeper pockets to absorb the losses. An MGA's errors are more visible because the book is smaller and the impact is concentrated. In this case, the carrier ultimately bore the loss, not the MGA. The MGA's reputation suffered, but the carrier's balance sheet took the hit. This raises the question: should the carrier have done more due diligence? Perhaps the real failure was not the actuary's model but the carrier's oversight.

Another counter-argument is that the model performed well for two years. The two large claims that caused the loss ratio spike could be seen as bad luck rather than poor modeling. If the model had been stress-tested for a recession or a legal change, it might have been more conservative, but that conservatism would have made premiums higher and potentially driven away business. In a competitive market, there is pressure to price aggressively. The MGA's pricing was in line with competitors, and the loss ratio was initially better than average. The model was not obviously wrong until it was.

Finally, some argue that the buyer's responsibility is limited. The buyer cannot be expected to audit the MGA's pricing model. The buyer relies on the broker and the carrier to ensure the price is fair. If the buyer pays too much, that is a problem; if the buyer pays too little, the carrier may become insolvent and claims may go unpaid. But the buyer's primary concern is coverage, not actuarial accuracy. The lesson for buyers is to choose a broker and carrier with strong reputations, not to become an actuary themselves.

These trade-offs do not excuse the lack of peer review, but they contextualize it. The MGA market is built on trust and speed. When trust is broken, the system can correct itself through reputational damage and carrier oversight. The question is whether that correction happens quickly enough to protect policyholders.

Alternative Approaches: How Other Markets Handle Model Governance

Not all insurance markets rely on the same light-touch approach. In the United States, for example, property and casualty rates are often subject to prior approval by state insurance departments. While this does not apply to surplus lines or many professional liability policies, the regulatory culture is more interventionist. Some states require actuarial opinions for certain lines, and model governance is more formalized. In contrast, the London market has historically relied on self-regulation and market discipline.

Another approach is the use of model registries or centralized actuarial functions. Some large carriers maintain a central actuarial team that reviews all MGA models before binding. This adds cost but provides a layer of protection. In the case of the London MGA, the carrier had an actuarial function but did not exercise it for this book. A more proactive approach would have required the MGA to submit the model for review before the first policy was bound.

Technology also offers solutions. Some MGAs now use cloud-based pricing platforms that log all changes and allow for real-time auditing. These platforms can flag unusual assumptions or deviations from benchmarks. The actuary's Excel model, by contrast, was static and opaque. If the MGA had used a platform with built-in governance features, the lack of peer review might have been mitigated by automated checks. However, such platforms are expensive and may not be feasible for a small MGA.

The London MGA's experience is a reminder that model governance is not just a technical issue but a cultural one. The market must decide how much rigor it wants. If the industry continues to rely on single-actuary Excel models, it should expect more failures. But if it moves toward greater transparency and peer review, it may lose some of the speed that makes MGAs attractive. The balance is delicate, and each firm must find its own equilibrium.

This article is for informational purposes only and does not constitute professional advice. Readers should consult qualified insurance professionals for guidance specific to their circumstances.