The managing general agent model used to mean one thing: a small, hyper-specialized operation built around a single kind of risk nobody else wanted to underwrite. According to Matt Wolfe (pictured right), president and CEO of Aon Reinsurance Solutions Canada, speaking on a panel at a recent industry conference, that model is still alive, but it's no longer the whole story.
"We oftentimes saw it as a smaller, middle-sized company with an extremely specific niche," Wolfe said. He pointed to real examples from his own career: an MGA in the US that insured only volunteer fire departments, another that covered white-linen fine dining restaurants in just three states. Their value proposition was narrow and deep.
Read more: AI-assisted cyberattacks are the biggest wildcard for Canada's life insurers, analyst says
"They would do a lot better underwriting this business through loss control, value-add," Wolfe said, describing how those MGAs would advise clients directly, recommending floor mats for snowy climates or training restaurant staff on responsible alcohol service, specifically to avoid the claims those niches were prone to.
What still makes an MGA viable, Wolfe said, is the same discipline that built the model decades ago: the ability to consistently beat industry-average loss ratios. That matters more than ever, he said, because an MGA's structure adds cost layers an insurer wouldn't otherwise carry.
"You really do need to drive some real value if you're that MGA," Wolfe said. "If you're just going to offer me industry average loss ratios but say I want an additional 10% commission, potentially a processing commission, that math just doesn't work."
What has changed, according to Wolfe, is who owns the MGAs pursuing that model. Large broker-owned MGAs have become a defining feature of the space, he said, pointing to Encon (now Victor Canada), the well-established MGA owned by the Marsh Group, and Aon's own significant investment in its Lynx MGA operation.
"I think where you're seeing that [benefit] is potentially technology distribution, that's becoming another kind of value proposition," Wolfe said. The old pitch, he said, was built entirely around underwriting expertise: we'll underwrite this better than you can. The newer pitch increasingly leans on infrastructure. "You're rising to say, we are going to make massive investments in AI processing, analytics, cat modelling, large data, so that we can produce business for our partners."
The most significant shift on the horizon, in Wolfe's view, has nothing to do with underwriting talent at all. It's coming from capital markets, specifically the rise of what he called alternative capacity: money from hedge funds, pension funds, and other asset managers looking for a place to park enormous sums in search of yield.
"There's billions and billions of dollars of capital that needs to find a yield," Wolfe said. That capital has moved meaningfully into reinsurance over the past 15 to 20 years, he said, growing from essentially nothing to an estimated $110 billion to $120 billion of the total $700 billion to $800 billion reinsurance market.
Part of the appeal, Wolfe said, is that catastrophe risk is genuinely uncorrelated with the rest of a typical institutional portfolio, unlike the interconnected financial exposures that caused cascading failures during the 2008 financial crisis.
"The hurricane in the Atlantic doesn't know what the Dow Jones did that day," Wolfe said.
Wolfe expects that capital to keep moving closer to the original risk, beyond reinsurance and directly into insurance and MGA business. Forward-looking MGAs, he said, should already be thinking about how alternative capacity could become part of their own capital solutions in the next decade.
Asked where the MGA market goes from here, the panel's answers converged on a similar theme: AI is coming for the underwriting process itself, and the biggest constraint won't be capability; it'll be trust. Brady DeSantis (pictured left), vice president of underwriting solutions at Navacord, said the near-term application is straightforward: AI handling submission ingestion and filtering, but the harder question is what happens once AI starts making underwriting judgments directly.
Read more: The US border is doing more work than people assume in Canadian insurance
"The key to that would obviously be the auditability," DeSantis said. "If you're not able to know what that AI unit is doing, then there are several reasons that you're not up to it. Is it actually making sense, and is it following whatever direction we're trying to get it to go?"
Paul Jackson (pictured centre-right), CEO of Zurich Canada, offered a different read, suggesting MGAs might end up mattering more for their underwriting judgment than for their technology, even as automation spreads elsewhere in the value chain.
Wolfe's closing point tied both threads together: the scale of investment required to compete on AI may end up forcing consolidation among smaller MGAs and insurers alike, the same way broker consolidation played out years earlier.
"The painful part may be that the investment needed to compete in that kind of AI-driven landscape will force M&A," Wolfe said. "If you're a $50 million mutual, having the ability to make the investment in AI to compete against multi-billion-dollar worldwide firms is going to be challenging."
