12 Insurers Went Insolvent After Hurricane Andrew. After a Storm Ten Times Larger, One Did.

Mark Anquillare joined the company that would become Verisk — then called Insurance Services Office — the same year Hurricane Andrew hit, after starting his career at Prudential in Newark. ISO was looking to upgrade its financial systems and reporting, and the opportunity to work at a dynamic, central player in the P&C data ecosystem pulled him over. Almost 30 years later, he’s President and COO of Verisk Analytics, and the Andrew anniversary — recorded during hurricane season, close to Andrew’s actual anniversary — became the natural framing for the entire conversation.

In Episode 81 of InsurTechTalk, Mark and I covered how catastrophe modeling fundamentally changed insurer solvency outcomes over three decades, the specific property-level and portfolio-level decisions Verisk’s data actually informs, and his genuinely striking argument that insurance should eventually function like a modern trading platform.

About Mark Anquillare

Mark Anquillare is President and COO of Verisk Analytics, a data analytics and risk assessment company serving the insurance industry across the full value chain — underwriting, claims, catastrophe modeling, and core systems — alongside separate energy and ESG-adjacent verticals. Verisk’s insurance business spans roughly 31 lines and covers underwriting analytics, claims fraud detection (ClaimSearch), property repair cost estimation (Xactware), London market policy administration software, and extreme event/catastrophe modeling (Touchstone, formerly AIR’s product suite). He joined the company (then Insurance Services Office) the year of Hurricane Andrew, almost 30 years before this recording.

What Verisk Actually Covers

Mark’s plain breakdown of the business: roughly 80% is insurance-focused, spanning the full value chain — underwriting (pricing and loss cost data across 31 lines), claims (fraud detection via ClaimSearch, which aggregates industry-wide claims data), property repair cost estimation (Xactware, working directly with policyholders, adjusters, and claims departments), London market core systems software, and — the area the conversation spent the most time on — extreme event and catastrophe modeling via Touchstone. The remaining roughly 20% covers energy (aggregating information about underground assets — extraction cost, exploration value) and adjacent government-facing work.

ESG: More “S” Than “E” or “G”

Asked how Verisk engages with ESG specifically, Mark was candid that the company’s involvement skews toward the social dimension more than environmental or governance. Verisk’s catastrophe models have been adjusted to account for climate change trends, and the company holds meaningful data on commercial building consumption and carbon footprint relevant to environmental tracking. But its most active ESG involvement is in building social indices — data capturing social injustice risk, political risk, and demographic shifts across specific geographies, which insurers increasingly use in conjunction with their own policyholder-facing ESG frameworks.

Andrew to Today: The Insolvency Numbers That Actually Matter

This was the central, genuinely striking statistic in the conversation. Hurricane Andrew, roughly a $15 billion insured loss event at the time, resulted in 12 insurance company insolvencies. Mark’s comparison points, drawn directly forward: 2011 produced roughly $145 billion in insured losses with a small handful of insolvencies; 2017 produced roughly $150 billion in insured losses with just one insolvency.

His explanation for the gap between event size and insolvency outcome, given that 2011 and 2017 losses dwarfed Andrew’s roughly tenfold: catastrophe models fundamentally changed how prepared insurers and reinsurers actually are. Insurers today can quantify tail risk with genuine precision, price and reserve against it proactively, and — critically — catastrophe models function as a shared language between insurers and reinsurers, letting two separate companies converge on a common valuation of risk and place reinsurance far more effectively than was possible in 1992.

Why Florida Coastal Exposure Keeps Growing Anyway

I pushed Mark on a genuine tension: if catastrophe risk is this well understood today, why do insurers keep writing more coastal Florida exposure rather than retreating from it? His answer, delivered with a wry observation about American risk-taking culture (“America loves to take their most prized assets and put them in the most risk-prone areas”): growth in a well-understood risk category is a legitimate business strategy precisely because the risk is now genuinely quantifiable — insurers can grow into Florida coastal exposure with real underwriting guidelines and layered reinsurance protection calibrated to their actual risk appetite, rather than growing blind. The additional tools available today beyond traditional reinsurance — cat bonds and broader insurance-linked securities transactions — give insurers meaningfully more precise control over exactly how much risk stays on their own books versus gets ceded elsewhere.

Where Verisk’s Data Actually Shapes an Underwriting Decision

Mark got specific about what Verisk’s data actually informs at the individual property level, distinct from portfolio-level catastrophe modeling:

• Policy language nuance — the practical coverage differences between an HO3, HO4, and HO5 homeowners policy form, and what specifically is and isn’t covered under each
• Property-specific risk attributes — presence of a pool, number of stories, distance to coast
• Fire protection granularity that goes beyond “is there a local fire department” — his specific example: whether the actual hydrant serving a given property has sufficient water flow and pressure to meaningfully fight a fire at that specific address, not just a general presence/absence check

That property-level data then rolls up into portfolio-level decision-making: how much additional coastal exposure an insurer should take on given what they’ve already written, and where the internal guideline threshold sits for saying no to further concentration in a specific geography.

Aerial Imagery Down to Individual Inches

Mark’s description of where risk data acquisition is heading was genuinely specific: aerial imagery resolution has reached the point where individual pixels represent roughly a couple of inches — meaning an underwriter isn’t just confirming a pool exists, they can potentially identify a diving board or trampoline present on the property (both meaningful liability and injury risk factors most standard applications never ask about directly). Verisk also gathers granular fire-protection data at the building level (sprinkler system currency, protection quality), framed as serving two purposes simultaneously: making communities genuinely more resilient during major catastrophes, and making the specific properties within those communities measurably less risky to underwrite.

Individuals, Not Just Properties

Mark clarified that Verisk’s data granularity extends to the individual as well as the property. With opt-in consent and basic identifying information, Verisk can compile a person’s vehicle history, accident and moving violation record, and elements of financial history relevant to underwriting decisions — the personal-lines equivalent of the property-level granularity described above. On the claims side, that same granularity supports fraud detection: identifying whether a claim has been submitted twice, or whether specific parties involved (adjusters, attorneys, contractors) show patterns consistent with organized claims fraud rings the industry has flagged previously.

Partnering With, Not Just Competing With, Insurtechs

Asked directly how Verisk relates to the wave of newer insurtech data and analytics startups, Mark’s answer described a genuinely open partnership posture rather than pure incumbent defensiveness. Verisk’s Xactware/Xactimate platform (property repair cost estimation) maintains roughly 100 different partnerships with third parties supplying specific data inputs — measurements, material specifications, and similar detail feeding into a given repair cost assignment. His framing: Verisk effectively sources and aggregates on behalf of its insurance customers, functioning as connective infrastructure within a broader ecosystem rather than trying to own every data source itself. He noted this dynamic is most visible and mature on the claims side specifically, where he sees the clearest path toward further industry-wide automation.

Why Insurance Should Function Like a Modern Trading Platform

This was Mark’s sharpest, most quotable framing in the conversation. His illustration: buying Microsoft stock around the time of Hurricane Andrew meant calling a broker, waiting for manual execution on paper tickets, and paying a meaningful commission (his example: $35-50) for a genuinely clunky process. Buying the same stock today is instantaneous, often free, and essentially frictionless.

His argument: insurance — at least for standardized, commoditized coverage — should be moving toward that same trajectory, and he sees early signs of it in auto specifically. His vision for how Verisk fits into that transition: building products designed to interconnect not just with each other, but genuinely with third-party systems and insurers’ own internal infrastructure, primarily via microservices and APIs — letting any given insurer pull exactly the information they need in whatever form is easiest for their own systems to consume, rather than being locked into a single vendor’s proprietary format.

Four Priorities Driving Verisk’s Roadmap

Asked what’s next, Mark named four concrete priorities, largely driven directly by customer feedback:

• Better decisioning — more granular data and segmentation specifically to support accurate, fine-grained pricing and stronger fraud detection
• Digital engagement — meeting policyholders’ expectation of instant, mobile-native interaction: quotes generated instantly at the point of inquiry, direct participation in surveys and claims inspections via phone rather than waiting for a scheduled in-person visit
• Automation — reducing frictional cost industrywide by making the broader ecosystem (Verisk’s own products plus third-party and internal insurer systems) genuinely interconnected, since Mark considers cost reduction the primary driver behind the push toward greater automation, not novelty for its own sake
• International expansion — extending Verisk’s current US-centric footprint further into international markets, with Europe as the immediate near-term focus

Advice: Storm of the Century

Asked for a closing recommendation, staying on theme with the Hurricane Andrew framing that opened the conversation, Mark recommended Storm of the Century by Al Roker — covering the 1900 Galveston, Texas hurricane, historically one of the most significant and deadly storms in US history.

Key Takeaways

• Catastrophe modeling adoption produced a dramatic, measurable solvency outcome shift — 12 insurer insolvencies after Hurricane Andrew’s roughly $15 billion loss, versus a single insolvency after 2017’s roughly $150 billion in insured losses, a tenfold larger event
• Catastrophe models function as a shared analytical language between insurers and reinsurers, enabling more precise, mutually understood reinsurance placement than was possible before their widespread adoption
• Growing exposure in a well-understood risk category (like coastal Florida) is a legitimate strategy precisely because catastrophe modeling now makes that risk genuinely quantifiable, rather than a sign of imprudent underwriting
• Property-level risk data has become extraordinarily granular — down to identifying specific liability-relevant features (diving boards, trampolines) via aerial imagery, and hydrant-specific fire protection adequacy rather than generic fire department presence
• Verisk positions itself as ecosystem infrastructure rather than a pure walled-garden incumbent, maintaining roughly 100 active data partnerships feeding its Xactware repair cost estimation platform alone
• Mark’s trading-platform analogy for insurance’s future — instantaneous, low-friction, largely automated execution for standardized coverage — represents a genuinely different vision than most incumbent data providers articulate publicly
• Verisk’s roadmap priorities (better decisioning, digital engagement, automation, international expansion) are explicitly customer-feedback-driven rather than technology-push-driven, reflecting a data-provider’s view of where the broader industry is already pulling