Want More Trains? Then We Need to Talk About Insurance.

Sorry.

Six days ago, a very wonky problem just became a very real barrier to new passenger-rail service.

There are few phrases more effective at clearing a room than “passenger-rail liability insurance and indemnification.” I get it, really. I understand. You came here for trains, not insurance towers, contractual risk allocation, and Federal liability statutes. But please, stay with me, because this stuff can decide whether the train ever leaves the station. And your voice could be crucial to helping us solve this problem.

Here’s the simplest way I know to explain it. There are three different questions that tend to get mashed together. Liability asks how much injured passengers can recover. Indemnification asks which party — the host railroad, passenger operator, public agency or somebody else — ultimately has to bear the financial responsibility. Insurance asks where that party gets the money.

These are not the same things.

Federal law currently limits the aggregate awards to all rail passengers, against all defendants, arising from a single accident or incident. As of last Friday -- six days ago, September 4, 2026 -- that inflation-adjusted number jumped to $401.9 million. The law also expressly allows passenger-rail providers to make contracts allocating financial responsibility for claims.

Photo by Pat Moin on Unsplash

That second part is especially important. That’s because, as we all know, passenger trains mostly operate over track owned by freight railroads. The freight host, understandably, wants to know what happens if something awful occurs while somebody else’s passengers are riding over its railroad. Over time, the industry developed a kind of “grand bargain” that often relies heavily on indemnification rather than waiting until after an accident to spend years fighting about fault.

GAO looked at 33 commuter-rail/freight-rail agreements in 2009. Twenty-one were no-fault arrangements, ten mixed no-fault and fault-based provisions, and only two were fundamentally fault-based. In a typical no-fault structure, the passenger side might agree to take responsibility for its passengers, employees or equipment even if ordinary negligence by the freight host contributed to the accident; the host takes responsibility for specified freight-side risks.

There are perfectly sensible reasons for doing that. It can make claims easier to resolve, reduce litigation among the railroads after an accident, and give everybody a clearer idea of the risk they are accepting before the first train operates.

But there is an obvious next question: if I have promised to indemnify you, how do you know I can actually pay?

Enter insurance.

The host may require the passenger operator or public agency to demonstrate financial security through insurance, self-insurance, collateral or some combination. In the conventional commercial market, very large limits are assembled as a “tower”: one insurer takes a layer, another takes the next layer, another sits above that, and so on until the required limit has been reached.

And here is where the system has started to eat itself. For many commuter agencies, a tower can include 20 or more layers. Imagine what it would be like if you needed to carry 20 insurance policies for your house or your car?

The $401.9 million figure is a liability ceiling for passenger claims. It is not, by itself, a universal Federal command that every passenger operator go buy a $401.9 million commercial insurance policy. In fact, the statute separately specifies a minimum insurance-and-self-insurance requirement for Amtrak. But access agreements and indemnity provisions can turn that passenger cap into a practical benchmark for how much financial security a host wants to see.

So, a law that was intended to make passenger-rail risk finite and insurable can, through the contracts built around it, become the number used to size an increasingly difficult insurance placement. Why do I say “difficult”? Because the international insurance market is losing its appetite to chip in a layer to these towers. From London to Bermuda and everywhere in-between, some insurers are just looking at this and saying, “No thanks.”

That would be less troubling if the number tracked actual rail risk. But it just doesn’t. The FAST Act requires the passenger cap to be adjusted every five years using the Consumer Price Index. CPI can tell us a lot of useful things. It cannot tell us whether Positive Train Control reduced catastrophic-accident risk, whether an operator has an excellent claims record, whether a route has fewer grade crossings, or whether the global excess-liability market has decided it wants less railroad exposure this year.

Meanwhile, the insurance market has hardened. Carriers have reduced participation, agencies rely on multiple layers and international markets, premiums have climbed, and some operators have found it increasingly difficult to assemble complete towers. My view on this comes from my perch on the Surface Transportation Board’s Passenger Rail Advisory Committee, where one of my committee assignments is our liability committee. Our group’s liability work has been documenting exactly that disconnect: the statutory number keeps climbing while underwriting capacity does not necessarily climb with it.

There is another wrinkle. The passenger cap is not the entire universe of risk from a railroad accident. Federal law expressly leaves Federal Employers Liability Act (FELA) and workers’ compensation damages untouched, and contracts may address third-party injuries, property damage, environmental exposure, defense costs and other claims. So, when somebody says, “the cap is $401.9 million, therefore we need $401.9 million of insurance,” the right response is: insurance for which obligation, exactly?

This is why the problem hits new and smaller passenger-rail projects especially hard. A big established operator or well-funded agency can spread risk across a large system, maintain substantial self-insurance and buy into the market at scale. A new state or regional service may be trying to insure one corridor with no operating history and a much smaller balance sheet. The railroad may be ready. The stations may be ready. The funding may be ready. And the project can still run aground on the question of who stands behind a catastrophic loss.

And by the way, let’s be clear here: that’s a market-design problem, and not a reason to compensate injured passengers less. Nobody, most of all me, is arguing that we should shirk our obligations to anyone who’s hurt or worse in an incident.

The interesting alternatives therefore are not about making liability disappear. They’re instead about financing it differently. Public agencies can retain sensible lower layers themselves. Multiple operators can pool risks. Captive insurance companies or protected-cell structures can allow participating agencies to retain and share selected layers. Commercial insurers and reinsurers can then be used where their capacity adds the most value rather than assuming that every dollar of the tower has to be purchased the same way from the same shrinking market.

None of those ideas is magic. A captive needs capital. A pool needs governance and actuarial discipline. Catastrophic risk still has to be financed. Freight hosts still need confidence that an indemnity is worth more than the paper it is printed on. And every proposed solution has to preserve the central public-policy objective: if passengers are seriously injured or killed, money must be there to compensate them and their families.

But that’s exactly why just saying “insurance is expensive” is no longer a satisfactory answer...at least to me, anyway.

We have spent years diagnosing the problem. Operators say the towers are becoming untenable. Insurers explain, quite accurately, that capacity is constrained and this is what the market costs. Host railroads want financially secure indemnification. Public agencies want to run trains without putting an absurd share of their operating budgets into premiums. Everybody can be completely rational from inside their own box and still produce a system that prevents new service from happening.

For passenger-rail advocates, that’s the reason to care about this spectacularly unglamorous (some might even say narcoleptic) subject. Sometimes the thing standing between a community and a new train is not a missing locomotive, a bridge that needs rebuilding, or a track that needs upgrading. Sometimes it’s a spreadsheet with $401.9 million at the bottom, and not enough willing insurers above it. And six days ago, that problem got nearly $80 million worse.