The Great British Insurance Swindle

Some frauds are perpetuated by villains. The more durable variety are perpetuated by systems, and are thus more difficult to prosecute, because no individual hand can be shown to be in anyone’s pocket at any particular moment. The most durable variety of all are perpetuated by the victims themselves, who have mistaken their chains for jewellery and will fight you if you try to remove them.

Let us discuss the No Claims Discount.

Every year, approximately thirty-five million British motorists renew their car insurance, and virtually every one of them will receive a document asserting that they have accumulated a given number of years without making a claim, and that this entitles them to a reduction of anywhere from twenty-five to seventy-five percent on their premium. This is presented as a reward for virtue: you drove without incident; the insurer gratefully acknowledges the fact. The scheme is spoken of, in marketing literature and in the assured tones of the comparison website, as though it were a straightforward bargain between responsible driver and prudent insurer.

It is nothing of the sort. But you try telling that to Derek.

Derek (and you know Derek, everyone knows Derek) is the brother-in-law who arrives at Christmas lunch having apparently spent the preceding eleven months conducting independent actuarial research into the motor insurance market. He has Opinions about NCD. He has, with the forensic intensity of a man who has watched three YouTube videos on the subject, developed a system. He will explain it to you between the starter and the main course whether you request this or not. Never claim for anything under a thousand pounds, he says, with the solemn authority of a man delivering the Sermon on the Mount. You’re gaming it, you see. Playing the long game. He has nine years’ NCD. He will die with nine years’ NCD. He has raised it with his children.

The insurance industry, to its very partial credit, is aware that Derek is wrong about almost everything, and is also aware, to its commercial horror, that it cannot tell him so. Derek’s faith in the NCD system is not merely erroneous; It is structural. It is load-bearing. The entire edifice of British motor insurance consumer psychology rests on the assumption that the discount is the thing, that the number of years is what matters, that the card in the drawer represents something real and portable and earned. Several insurers have, in the privacy of their strategy sessions, contemplated replacing NCD with more sophisticated continuous claims history models. These conversations reliably founder on the same rock: focus groups. The moment a researcher suggests pricing customers without reference to their NCD years, respondents react as though someone has proposed removing their kidneys. The product, whatever its actuarial failings, has been successfully marketed as an entitlement. You do not take away British entitlements. You simply continue quietly not delivering on them.

The Numbers That Damn It

The NCD scheme is an asymmetric ratchet — slow to earn, catastrophically fast to lose. A claim-free year advances you one step toward maximum bonus. A claim throws you back by two to five years’ worth of accumulated restraint. As British roads have become considerably safer, the entire policyholder population has migrated upward toward maximum bonus. A calibrated model separating good drivers (five-percent annual claim frequency) from bad ones (fifteen percent) shows that at maximum bonus level, where roughly ninety percent of a modern portfolio resides, the premium discount delivered is approximately three percent. Three percent, to distinguish a pool of heterogeneous risks. The system is, at precisely the point where it applies to most people, actuarially inert. It has, in the manner of a stopped clock, achieved a state of perfect uselessness.

The industry’s defenders will cite predictive power, and a gradient-boosted model run on nearly eight hundred thousand policies does indeed rank NCD level as the second most important predictor of future claim frequency. But the signal is nonlinear and interactive, and the standard pricing model fails to capture it cleanly, producing a plausible-looking multiplier that mixes the true effect with noise in proportions you cannot reliably disentangle. A young urban male driving a high-group vehicle will, entirely without reference to his individual behaviour, tend to generate more claims and therefore tend to occupy a more adverse NCD class; an insurer that then applies an NCD surcharge to a base premium already inflated for youth, urbanity, and vehicle group has priced the same risk twice. Derek does not read the actuarial literature, and it is Derek’s credulity that makes the double dip possible.

There is a further distortion that Derek would recognise as his own master-strategy. The NCD penalty is independent of claim size, so every rational policyholder suppresses small claims rather than trigger a step-back worth several years’ elevated premium. Estimated claim frequency falls. Estimated severity rises. NCD level partly reflects non-reporting propensity rather than underlying risk. The insurer has created, through its own pricing mechanism, a contaminated signal, and then cited that signal as justification for the mechanism. This is circular reasoning dressed in the language of actuarial science. It is also, incidentally, what Derek calls “gaming the system” — when he is in fact the system’s most cooperative accomplice.

The Protection Racket

Now we arrive at what is, in the specific context of the British market, the most egregious deception of all. And I use that word not for rhetorical effect but because no milder word is adequate to the facts.

NCD protection products, available to policyholders with four or more years’ discount for an additional premium, purport to shield the policyholder’s NCD level after a claim. The FCA does not prohibit them. Comparison websites promote them assiduously. And they are a near-perfect consumer fraud, in that they protect precisely the component of the post-claim premium increase that is smallest, while leaving the larger component entirely untouched.

The British motor insurance market operates a dual-channel architecture. The final premium is the product of two independent mechanisms: a claims loading applied to the base premium, and then the NCD discount applied to the result. A single at-fault claim will, over five years, generate approximately £900 in NCD loss and approximately £2,800 in claims loading, the latter persisting for three to five years regardless of whether the NCD is protected or not. NCD Protection shields the £900 and leaves the £2,800 entirely exposed. The industry markets the smaller number and conceals the larger one, which is a tolerably precise definition of a mis-sold financial product. It is also, one notes with grim admiration, an extraordinarily elegant trick: the consumer buys protection against the visible loss, never suspecting that the invisible loss is more than three times larger. A stage magician would be proud.

When NCD level is decoupled from actual claims history by the protection product, the NCD integer ceases to mean what the pricing model was calibrated to assume it meant. A policyholder with five years’ protected NCD and two recent at-fault claims is not the same risk as a policyholder with five years’ unprotected NCD and a clean record, but the NCD level reads identically. The insurer knows this, of course, because it has access to the Claims and Underwriting Exchange, which retains six years of granular claims history including fault determination, claim type, settlement amount, and even incidents that did not result in a formal claim. The CUE database, with its thirty-four million records, is where the real pricing happens. NCD is, in the industry’s own internal practice, a consumer-facing simplification and a portability mechanism. The actuarial heavy lifting has long since moved elsewhere. The gap between what insurers tell policyholders their premium depends upon and what it actually depends upon is not a rounding error. It is the product.

The Willing Victim

Here is the peculiar tragedy of the British motor insurance consumer, who has been comprehensively deceived and would very much prefer to remain so.

A number of insurers has the data, the models, and the actuarial sophistication to price entirely without reference to the NCD integer, using the granular claims record directly. NCD level and underlying claims data are near-collinear: NCD is roughly a deterministic function of the at-fault claims history that insurers already hold in full. The NCD integer is retained not because it is actuarially superior, but because it is the figure on which consumers are trained to focus; It is the decoy. Sophisticated insurers know it is the decoy. They use it anyway, because to remove it would be to tell thirty-five million Dereks that the document in the kitchen drawer is a marketing construct rather than an actuarial instrument, and that the nine years he has been so careful to accumulate represent, at the level of actual risk discrimination, approximately nothing.

This information would not be well received.

No Claims Discount is presented to the British public as a transparency mechanism: a legible summary of their risk history, a fair reward for safe driving, a portable asset earned through prudence. It is, in practice, a consumer-facing simplification that decouples the number policyholders are shown from the analysis that actually prices them; a double-counting mechanism that surcharges risks already embedded in the base premium; a signal contaminated by its own incentive structure; and, in its protected form, a product sold as a shield while quietly leaving the largest post-claim costs entirely unguarded.

The industry and the actuarial literature has known most of this for thirty years. What the industry has lacked is not knowledge but incentive. The system persists not because it is believed, but because it is useful to believe — and a customer base that has internalised the deception so thoroughly that it will defend it against correction is not a problem to be solved. It is an asset to be managed. There is a word, in the literature of confidence tricks, for the mark who cannot be told they have been fooled because they have too much pride invested in their own cleverness. In the motor insurance industry, that word is: customer.

Derek would like to speak to you about his renewal strategy. He’s getting a very good deal, he thinks.