In liability insurance, the worst kind of claim is the one that does not stop at a single policy limit. It jumps from one carrier to the next, climbing a reinsurance tower until it hits the same set of reinsurers multiple times from different directions. That is exactly what happened when a defective product lawsuit generated a single loss that exhausted the limits of three managing general agents (MGAs), each of which had placed different layers of the same reinsurance tower with the same group of reinsurers.

The incident, which occurred in the U.S. general liability market roughly two years ago, has become a case study in how MGA capital efficiency depends on tower structure—and how that structure can create correlated exposure that pricing models routinely underestimate. The claim, a large product defect suit involving a consumer good, triggered losses under three separate MGA policies. Each MGA had a distinct attachment point in a shared reinsurance tower. The first MGA's first-loss layer was exhausted, then the second MGA's quota-share layer was hit, and finally the third MGA's excess layer responded. Reinsurers that had written all three layers paid out on all three policies for the same underlying event.

The episode raises uncomfortable questions about accumulation risk in the MGA sector, where rapid growth has been fueled by reinsurance capacity. If a single claim can move through three MGAs in one tower, how many other towers harbor similar hidden correlations? And what does that mean for pricing, capital modeling, and regulatory scrutiny?

One Claim That Rippled Through Three MGAs

The claim originated from a defective product lawsuit filed against a manufacturer. The plaintiff alleged that a design flaw caused personal injury and property damage. Because the product had been distributed through multiple channels, the manufacturer had purchased general liability policies from three different MGAs, each specializing in a different segment of the supply chain: one wrote coverage for the manufacturer's primary operations, another wrote coverage for its retail distribution, and a third wrote coverage for its logistics contractors.

Unbeknownst to the MGAs, all three had placed their respective policies into the same reinsurance tower—a structure designed by a single reinsurance broker that worked with the MGAs independently. The tower had multiple layers: a first-loss layer of $5 million, a quota-share layer that ceded 50% of losses between $5 million and $15 million, and an excess layer attaching at $15 million. Each MGA's policy attached at a different layer. The primary MGA retained the first $5 million but ceded the rest. The second MGA's policy was written entirely within the quota-share layer. The third MGA's policy attached at $15 million and sat in the excess layer.

When the lawsuit settled for roughly $22 million, the loss cascaded through the tower. The first MGA's limit was exhausted at $5 million. The second MGA's policy, sitting in the quota-share layer, absorbed $10 million, half of which was ceded to reinsurers. The third MGA's excess layer then responded for the remaining $7 million. Because the same group of reinsurers had written all three layers—the quota share, the excess, and the aggregate cover that sat above the tower—they ended up paying on all three MGAs' policies for the same loss. The total reinsurance recovery from that single claim was roughly $14 million, spread across three separate policy numbers.

Rob Lewis, CEO of INTX Insurance Software, has noted that such events expose the fragmentation in reinsurance operations. In a recent interview, he argued that platforms must unify data, workflows, and intelligence across the economics of risk to address significant inefficiencies. The claim that moved through three MGAs is a textbook example of why that unification matters. Without a common claim ID or a shared view of the tower, no single MGA or reinsurer saw the full picture until the loss had already been paid.

How MGA Capital Efficiency Depends on Tower Structure

MGAs rely on reinsurance to write large limits without holding proportional capital. By ceding a portion of their risk to reinsurers, they can offer policies with limits that would otherwise require significant surplus. The efficiency of this model depends on the structure of the reinsurance tower: the number of layers, the attachment points, and the retention levels determine how much capital an MGA must hold and how much premium it can generate.

In a typical tower, the first-loss layer is the most capital-intensive, because it bears the highest frequency of claims. MGAs that write into that layer often need more surplus or a larger quota-share cession. MGAs that write into higher layers can operate with less capital, because the probability of attachment is lower. But the trade-off is that when a large loss does occur, it can pierce through multiple layers, exposing the reinsurers that sit across all of them.

The three MGAs in this case had built their capital models assuming that their portfolios were independent. Each MGA modeled its own loss distribution based on its own book of business, without considering that a single event could trigger claims under all three policies. The correlation was invisible because the MGAs did not share underwriting data or claim systems. Their reinsurance broker had placed the tower as a single structure, but the MGAs treated their placements as separate transactions.

Industry observers have noted that this is a growing risk as the MGA sector expands. According to some estimates, MGAs now account for roughly 30% of the U.S. specialty insurance market, up from about 20% a decade ago. Much of that growth has been fueled by reinsurance capacity, as traditional carriers have pulled back from certain lines and delegated underwriting authority to MGAs. But the capital efficiency that makes MGAs attractive also creates the potential for correlated losses that are hard to detect.

The broader industry trend is that tower capacity is being stretched. As more MGAs compete for the same reinsurance limits, towers become more crowded, and the probability that multiple MGAs share the same reinsurers increases. That means a single large loss could ripple through multiple MGAs' books, even if they are not explicitly linked by the tower structure.

The Reinsurance Tower That Linked Them

The tower at the center of this case was a multi-layer structure with a total capacity of roughly $50 million. The first-loss layer, $5 million, was placed with a small group of reinsurers that specialized in high-frequency, low-severity claims. Above that, a quota-share layer ceded 50% of losses between $5 million and $15 million to a broader panel of reinsurers. The excess layer, attaching at $15 million and running to $50 million, was placed with a different set of reinsurers, but several of them also participated in the quota-share layer.

The three MGAs had each negotiated their own placement within this tower. The first MGA wrote a $5 million policy that fit entirely within the first-loss layer. The second MGA wrote a $10 million policy that sat in the quota-share layer. The third MGA wrote a $7 million policy that attached at $15 million and fell into the excess layer. Because the tower was designed by a single broker, the reinsurers knew that the three policies were linked, but the MGAs did not.

When the defective product claim came in, it exhausted the first-loss layer. The second MGA's policy then responded, triggering the quota-share cession. The reinsurers that had written the quota-share layer paid 50% of the $10 million loss, or $5 million. The remaining $5 million was retained by the second MGA's carrier. But then the loss exceeded $15 million, and the third MGA's excess layer attached. The reinsurers that had written the excess layer—some of whom were the same as those in the quota-share layer—paid the full $7 million.

In total, reinsurers paid $5 million from the quota-share layer and $7 million from the excess layer, plus the first-loss layer had been fully ceded to the first-loss reinsurers. The same reinsurers that participated in both the quota-share and excess layers effectively paid twice for the same loss. Without a common claim ID or a unified system to track accumulation, the reinsurers did not realize the extent of their exposure until after the claim had been settled.

Rob Lewis of INTX has argued that such events are a symptom of fragmented reinsurance operations. In his view, the industry needs platforms that unify data, workflows, and intelligence across the economics of risk. The three-MGA claim is a clear example of why that unification is necessary. If the reinsurers had been able to see that the same loss was hitting three different policies in the same tower, they might have reserved differently or negotiated a different tower structure.

Data Silos Masked the Accumulation Risk

One of the key factors that allowed this hidden correlation to exist was the lack of data sharing among the three MGAs. Each MGA used a separate policy administration system, and none of them had a common claim ID that could link losses across their books. The reinsurers, in turn, received bordereaux from each MGA in different formats, with different data fields and different definitions of what constituted a claim.

The manual workflows that reinsurers use to process bordereaux made it difficult to spot the connection. According to Lewis, manual workflows cause significant inefficiencies in reinsurance operations. When a loss is reported under one MGA's policy, the reinsurer's claims team processes it in isolation. If the same loss is later reported under a second MGA's policy, there is no automatic flag to indicate that the two claims are related. The reinsurer may not discover the connection until the aggregate loss exceeds a threshold that triggers a manual review.

In this case, the connection was not discovered until after the third MGA's claim was paid. A senior claims analyst at the lead reinsurer noticed that the insured names and the date of loss matched across three separate bordereaux. By that point, all three claims had been settled. The reinsurer's internal review later found that the tower had been designed without any accumulation controls for cross-MGA claims. The broker had not flagged the potential correlation because the MGAs had not disclosed their full portfolio of policies to each other.

The incident highlights a broader problem in the MGA market: data silos are endemic. MGAs often operate as independent entities, even when they share the same reinsurance tower. They may compete for business in the same lines and territories, but they do not share underwriting data or claim histories. Reinsurers, for their part, have historically relied on the MGAs' own reporting to assess accumulation risk, but that reporting is often incomplete or delayed.

Some industry participants argue that the solution is to adopt unified platforms that give all parties a common view of the tower. Lewis has been a vocal advocate for this approach, arguing that reinsurance operations must evolve to remove constraining inefficiencies. But the cost and complexity of integrating multiple MGA systems into a single platform are significant, and many MGAs are reluctant to share proprietary data with competitors.

Pricing Models Underestimated Correlation in Liability Lines

Actuarial models that price MGA portfolios typically treat each MGA's book as independent. The assumption is that losses from different MGAs are uncorrelated, or at most weakly correlated, because they underwrite different segments of the market. But as the three-MGA claim demonstrates, that assumption can be wrong in liability lines, where a single event—such as a product defect—can trigger claims under multiple policies across different MGAs.

The problem is that product defect lawsuits often involve several defendants along the supply chain. The manufacturer, the distributor, the retailer, and the logistics provider may all be named in the same suit. If each of those entities has a general liability policy from a different MGA, and if all those MGAs sit in the same reinsurance tower, then a single lawsuit can produce correlated losses across the entire tower.

The pricing models used by the MGAs in this case did not account for that correlation. Each MGA modeled its own loss distribution based on its own historical experience, which did not include large coordinated losses. The reinsurers' models, in turn, assumed that the MGA portfolios were independent, because they had no data to suggest otherwise. The result was that the loss ratios for the tower as a whole spiked far beyond what the models had predicted.

Some actuaries have argued that the industry needs to develop new correlation parameters for liability lines, especially as supply chains become more complex. The traditional approach of using a single correlation factor for all lines of business is too crude. Instead, underwriters and actuaries should map the interconnections among insureds in the same supply chain and adjust their models accordingly.

But that is easier said than done. Mapping supply chains requires data that many MGAs do not have, and it raises privacy concerns. Moreover, even if the correlations were known, pricing them into the premium would require a level of coordination among MGAs that is rare in a competitive market. The three-MGA claim is a reminder that the industry's pricing models are only as good as the data they are built on—and that data often hides the most dangerous correlations.

Regulatory and Rating Agency Scrutiny Intensifies

In the wake of incidents like this one, regulators and rating agencies have begun to scrutinize MGA tower structures more closely. The National Association of Insurance Commissioners (NAIC) has issued guidance encouraging insurers to assess accumulation risk across all MGAs that share the same reinsurance tower. Some rating agencies have indicated that they will consider cross-MGA correlations when assigning financial strength ratings to reinsurers that write multiple layers of the same tower.

According to a recent article in Carrier Management, the use of chief risk officers (CROs) has risen in the MGA sector, as firms try to get a handle on these hidden exposures. CROs are now expected to map all MGAs that share a common reinsurance tower and to stress-test the tower with correlated large-loss scenarios. The three-MGA claim is often cited as a cautionary example in boardroom presentations.

For reinsurers, the regulatory pressure is twofold. First, they must demonstrate that they have adequate systems to detect cross-MGA accumulation. Second, they must hold capital against that accumulation, even if their models do not explicitly capture it. Some reinsurers have responded by tightening their underwriting guidelines for MGA towers, requiring that all MGAs in the same tower use a common claim reporting system.

But the regulatory response is still evolving. Some observers argue that the current approach—relying on voluntary industry action—is insufficient. They point to the growing complexity of liability exposures, such as those arising from self-driving cars, which could create new cross-MGA risks. As Carrier Management reported, self-driving car firms must address emergency vehicle interference, a liability risk that could involve multiple insureds—the car manufacturer, the software provider, and the fleet operator—each with a different MGA policy.

The rating agencies, for their part, have been cautious. They have issued warnings but have not yet downgraded any major reinsurer solely on the basis of cross-MGA correlation. That may change if another large claim moves through multiple MGAs in a single tower.

Practical Takeaways for Actuaries and Underwriters

For actuaries and underwriters working in the MGA space, the three-MGA claim offers several concrete lessons. The first is to map all MGAs that share the same reinsurance tower. That means identifying not just the direct counterparties but also the indirect ones that may write into the same layers. A simple spreadsheet can reveal connections that are invisible in a single MGA's books.

The second lesson is to use a common claim ID across all systems. If every MGA in a tower reports claims using the same unique identifier for the underlying event, then the reinsurer can automatically aggregate losses across policies. That requires coordination among the MGAs and the broker, but it is technically feasible with modern platforms. Lewis has argued that unified platforms can achieve this, but the industry has been slow to adopt them.

The third lesson is to stress-test towers with correlated large-loss scenarios. Actuaries should simulate events that trigger claims under multiple MGAs' policies simultaneously, using realistic assumptions about supply chain connections. Those stress tests can reveal whether the tower's capital is adequate or whether it is overexposed to a single event.

The fourth lesson is to adopt unified platforms that reduce manual fragmentation. As Lewis has said, reinsurance operations must evolve to remove constraining inefficiencies. That means moving away from bordereaux sent as spreadsheets and toward real-time data sharing through APIs. The cost of such platforms is significant, but the cost of a hidden correlation event can be much higher.

Finally, underwriters should ask their reinsurance brokers for a full view of the tower, including all MGAs that place business into it. If the broker cannot provide that view, that is a red flag. The three-MGA claim shows that the absence of that information can lead to losses that no one saw coming.

This article is for informational purposes only and does not constitute professional advice. Readers should consult qualified actuaries and reinsurance specialists for guidance on their specific situations.