1967#

California’s minimum auto liability was 15/30/5 — $15,000 per injured person, $30,000 per accident, $5,000 for property damage.

It was set in 1967.

It was still the law in December 2024.

Fifty-eight years. The number was written two years before Apollo 11, and it was still the legal floor when I started working on insurance documents. In January 2025 it finally moved, to 30/60/15, under a bill called the Protect California Drivers Act.

Sit with the arithmetic for a second. $15,000 in 1967 was real money — you could buy a house in much of the country for four or five times it. By 2024 it was an emergency room visit and a couple of nights on a ward. The statute never changed, so the protection it described quietly evaporated while the words stayed identical.

Nobody decided $15,000 was the right number in 2024. Somebody decided it in 1967, and then fifty-seven legislatures in a row didn’t revisit it. That’s not a policy; it’s a fossil.

And here’s the detail that convinced me this is systemic rather than a one-off embarrassment: when California finally fixed it, they didn’t index it. The new law schedules exactly one more increase — to 50/100/25 in 2035 — and then stops.

So the 2035 number will be a 2022 judgement about what 2035 should look like, and from 2036 onward the same erosion starts again, at the same speed, with nobody having decided anything. They replaced a stale number with a slightly fresher number and a timer, when the actual fix — tie it to an index and let it move by itself — was available and free.

The part I have to retract#

Here’s where I have to correct myself, and I’d rather do it in public than quietly.

When I first worked through this, I came up with what felt like a satisfying explanation, and I’ve repeated it confidently: the floor is low on purpose. Raising the compulsory minimum means forcing the poorest drivers to buy more insurance, which prices some of them out entirely — and a driver with no insurance is worse for victims than a driver with inadequate insurance. So legislatures keep the floor low deliberately, trading adequacy for participation. A real dilemma, not cowardice.

It’s a tidy story. It has the shape of a hard truth. I went looking for the evidence and it isn’t there.

States that have raised their minimums in recent years saw premium increases smaller than the national average, and no increase in uninsured drivers — in most of them the uninsured proportion actually fell. The pricing-out effect is a prediction, made repeatedly, that does not appear to have materialised.

Two honest caveats, because I’d be doing exactly what I just did wrong if I skipped them:

• The clearest research making this case comes from an association of plaintiffs’ lawyers, who benefit directly from higher limits. That’s an interested party and you should read it as one.
• But the other side’s own numbers don’t rescue the pricing-out story either. The insurance industry’s research body attributes variation in uninsured rates mainly to unemployment and general affordability, not to where a state sets its minimum. And the awkward case for everybody: New Hampshire doesn’t compel insurance at all and has one of the lower uninsured rates in the country. Whatever drives people out of the insurance market, it isn’t primarily the size of the mandate.

So I was wrong, and specifically I was wrong in the most seductive way available: I found an explanation that made a bad outcome sound like a difficult choice. That’s a very comfortable place to stop thinking.

So what is actually holding the number down?#

Something much less dignified, and I think it’s a straight consequence of the seam.

The compulsory coverages — bodily injury and property damage liability — are third-party. They do not protect you. They protect strangers, from you.

Now ask the political question. Who lobbies to raise them?

Not drivers: raising the minimum costs them money and buys them nothing they can feel. Not insurers, particularly: higher mandated limits mean higher premiums but also higher exposure, and the market for optional coverage above the floor is where the margin is anyway. Not the beneficiaries: the people this money protects are, by definition, people who haven’t been hit yet. They don’t know who they are. They can’t form a constituency. There is no association of future accident victims.

The only coverage the state compels is the one coverage nobody who pays for it will ever benefit from — so it has no defenders, and it rots.

Meanwhile the coverages that protect you — collision, comprehensive, uninsured and underinsured motorist — get bought voluntarily, sold enthusiastically, and updated by market pressure. Nobody has to legislate demand for those. The seam predicts which numbers get maintained and which get abandoned.

That’s the answer, and it’s worse than a trade-off. A trade-off implies someone weighed it.

What the neglect actually costs#

Two numbers, from the industry’s own research body, for 2023:

• 15.4% of U.S. drivers were uninsured — more than one in seven. Ranging from 5.7% in Maine to 28.2% in Mississippi.
• 18.0% were underinsured — carrying coverage, just not enough. Ranging from 4.6% in D.C. to 49.7% in Colorado, where half of insured drivers can’t cover what they might do.

Put them together:

Roughly one in three American drivers is either uninsured or underinsured.

And both rates have been climbing since 2020.

This is the thing to understand about the second number. Being underinsured is not primarily a choice drivers made — it’s a definition that moved. A driver who buys exactly what their state requires is, in Colorado, quite likely to end up in the underinsured bucket. They complied. The floor is what failed.

Which closes the loop on the last article’s ending. The coverage that actually protects you from all of this is underinsured motorist coverage — first-party, optional, bought by you. You are buying private protection against the consequences of a public number nobody updated.

The other design#

I live in British Columbia, which solved this differently — worth looking at not because Canada is better, but because it makes the American choices visible as choices rather than as the way things must be.

BC’s mandatory third-party liability limit is $200,000. One combined limit, no per-person cap, on every registered vehicle.

Line that up against the American numbers, including the new ones:

California’s new, doubled, hard-won per-person minimum is 15% of BC’s floor.

And the structural piece matters more than the number. In BC, basic insurance is bound to vehicle registration and provided by a single public insurer. You don’t license a car without it. There’s no monthly policy to lapse out of, no separate purchase decision to defer, no enforcement problem to solve — the coverage and the licence plate are the same transaction. Several other provinces run variants of this.

That eliminates the entire uninsured-driver category by construction rather than by enforcement. It is not that Canadians are more law-abiding. It’s that the option to be uninsured was designed out.

There are real costs to that model, and I’ve written about one of them — BC also removed the right to sue, so your compensation is now determined by an adjuster you cannot appeal to a court. That’s a genuine loss, and I don’t want to present a public monopoly as a free lunch. But it does demonstrate the point: a $200,000 mandatory floor is achievable. It isn’t a fantasy number. It’s a jurisdiction fifty kilometres north of Washington State.

What I’d take away#

The state minimum is not advice. It is not a considered view about what you should carry, and it isn’t a floor that anyone is watching. It’s the residue of a decision made in a year you may not have been born in, eroded by six decades of inflation that nobody was assigned to notice.

Treat it accordingly — both when you buy your own coverage, and when you find out how much the person who hit you was carrying.

And the general lesson, the one that isn’t about cars: when a rule protects people who can’t identify themselves in advance, expect it to decay. There’s no constituency for a number that only matters to whoever gets hurt next.

California’s was frozen for fifty-eight years, and the fix has a 2035 timer instead of an index.

Notes and sources

• California: minimum liability of 15/30/5 was in place from 1967 and raised to 30/60/15 effective for policies renewing on or after 1 January 2025 under Senate Bill 1107 (the Protect California Drivers Act), signed in September 2022, amending Vehicle Code §16056. The statute schedules a further increase to 50/100/25 in 2035. It contains no inflation index.
• Uninsured and underinsured rates: Insurance Research Council, Uninsured and Underinsured Motorists: 2017–2023, published 2025 — 15.4% uninsured countrywide in 2023 (Maine 5.7% low, Mississippi 28.2% high) and 18.0% underinsured (D.C. 4.6% low, Colorado 49.7% high), with roughly one in three drivers falling into one category or the other. Both rates rose after 2020. The IRC measures these from the ratio of uninsured/underinsured motorist claims to bodily injury claims.
• On raising minimums: the finding that states raising minimums saw below-average premium changes and no increase in uninsured drivers comes from research published by the American Association for Justice, an association of plaintiffs’ lawyers — an interested party in this debate, and I’ve flagged it as such in the text. The counterweight is that the industry’s own research body attributes uninsured-rate variation primarily to unemployment and affordability rather than to mandate levels, and that New Hampshire, which has no compulsory insurance requirement, records a comparatively low uninsured rate. I don’t think anyone has a clean causal story here; what I’m confident about is that the pricing-out claim is weaker than I presented it as being.
• British Columbia: basic third-party liability under the mandatory Autoplan coverage is capped at $200,000 for most vehicles, with higher limits for buses, taxis and limousines, per the Insurance (Vehicle) Regulation. Basic coverage for non-exempt vehicles is available only from the public insurer; extended limits above $200,000 and physical-damage coverage may be bought from it or from private insurers.
• Other state minimums cited (New York 25/50/10, Pennsylvania 15/30/5, Florida’s absence of a bodily injury liability requirement) as of 2026. State minimums change; verify current figures rather than relying on this article.

I’m learning this in public, and this article contains a correction to something I asserted confidently in the previous one. I’d rather be corrected than confident — if something here is wrong, please tell me.