Few pieces of safety equipment have been studied as intensively as the automobile seat belt, yet the history of its adoption is less straightforward than the familiar public-health narrative suggests. Although three-point belts were fitted to production cars in Sweden as early as 1959, drivers in most countries ignored them for well over a decade, and manufacturers, fearing that visible safety devices would remind customers that driving was dangerous, did little to promote their use. It was not until the 1970s, when Australia and several European nations made wearing a belt compulsory, that usage rates rose sharply. Legislators justified these laws with laboratory crash tests demonstrating that belted occupants were far less likely to be thrown against the windshield, and early statistics appeared to confirm that fatalities among vehicle occupants were falling.
This consensus was disturbed in 1975, when the economist Sam Peltzman published an analysis arguing that federal safety regulations in the United States had produced almost no net reduction in traffic deaths. His explanation, which came to be known as risk compensation, held that drivers who feel better protected unconsciously adjust their behavior, driving faster and following more closely, so that part of the engineering benefit is spent on convenience rather than survival. Peltzman further claimed that the burden of this adjustment fell disproportionately on people outside the vehicle: while occupant deaths declined, pedestrian and cyclist fatalities in his data rose enough to offset the gain. The proposition was provocative precisely because it implied that a measure everyone regarded as unambiguously beneficial might merely redistribute danger.
Subsequent research has neither vindicated nor entirely dismissed Peltzman's thesis. Studies that tracked individual drivers before and after belt laws took effect generally found only modest changes in speed, and large-scale reviews conclude that mandatory belt use has saved tens of thousands of lives, an outcome that risk compensation, even if it exists, did not cancel out. Nevertheless, the behavioral effect has appeared with surprising regularity in other domains: cyclists wearing helmets have been observed to ride faster, and motorists have been shown to pass helmeted riders more closely than bare-headed ones. Anti-lock braking systems, introduced with great expectations in the 1980s, produced smaller reductions in crashes than engineers had projected, apparently because drivers equipped with them braked later and drove more aggressively in wet conditions.
What the seat belt episode ultimately illustrates is less a flaw in any particular device than a recurring weakness in how safety policy is evaluated. Engineers test equipment on crash dummies, which do not change their habits in response to protection, and regulators tend to assume that the public will behave as those dummies do. Human beings, however, treat safety as one resource among many and are inclined to trade some of it for speed, time, or comfort whenever the opportunity arises. The lesson is not that protective measures are futile, since the historical record shows they are not, but that their real-world benefits will almost always be smaller than laboratory figures promise, and that policymakers who ignore this gap will consistently overestimate what technology alone can accomplish.