A cycling path in a city is widened and given improved lighting and a physical barrier separating it from vehicle traffic. The safety intervention is sound and evidence-based. The city expects a reduction in cyclist injuries.
Some reduction in injuries occurs. But it is smaller than expected. An analysis finds that average cycling speeds on the route have increased, and riders are less frequently checking intersections as carefully as they did before the barrier was installed. The improved infrastructure made cyclists feel safer - and they responded by cycling somewhat less cautiously.
The intervention worked. But some of its benefit was absorbed by behavioural adaptation. This is risk compensation.
What Risk Compensation Means
Risk compensation - sometimes called the Peltzman effect - is the tendency for people to take greater risks when their perceived safety increases.
The core mechanism is that people have a preferred level of risk. When a safety intervention reduces the felt level of danger, the gap between actual and preferred risk produces a behavioural response: people adjust their behaviour to restore something closer to their preferred risk level. The safety intervention shifts the level; behaviour shifts back.
This does not mean safety interventions are ineffective. It means their benefits may be partially offset by behavioural responses - and understanding that response is necessary for designing interventions that deliver their intended impact.
How It Works: The Mechanism
The mechanism is a psychological shift in perceived risk. Safety measures reduce the subjective sense of danger. Lower perceived danger lowers the motivational pressure to maintain the caution levels that were previously adaptive.
This is not irrational behaviour. In many situations, taking more risk when you are better protected is a reasonable response - the same outcome can be achieved at lower cost, or a better outcome can be achieved at the same cost. The problem for safety policy is that the behavioural adaptation partially or fully offsets the intended effect of the intervention.
The degree of compensation depends on how directly the safety measure reduces perceived risk. Measures that visibly increase protection - seatbelts, helmets, barriers - tend to produce more compensation than measures whose effect is less perceptually salient. Behavioural adaptation is stronger when the change in felt safety is clear.
Why This Matters
In road safety, risk compensation is one reason why mandatory seatbelt legislation has not always produced the full reduction in fatalities that pure mechanical benefit would predict. Seatbelted drivers sometimes drive faster and follow more closely - not as a deliberate choice, but as a consequence of feeling more secure. The net effect may be better safety for seatbelt wearers while increasing risk for other road users who are not protected by the same measure.
In sports and recreation, protective equipment adoption can increase the aggressiveness of activity. Helmet use in contact sports has been associated with increased intensity of head contact in some studies - players who feel protected may accept more collision risk.
In public health and workplace safety, risk compensation can appear when workers are given protective equipment that reduces the immediate physical danger of a task. The equipment reduces injury risk from the specific hazard it addresses, but may reduce overall caution in ways that expose workers to other hazards.
The Common Misunderstanding
Risk compensation is often taken to mean that safety measures are counterproductive or that providing protection is pointless. This seriously overstates the effect.
The reality is that risk compensation is typically partial, not complete. Safety interventions continue to produce net benefit in most well-studied cases, even when some behavioural offset occurs. The point is not that seatbelts make driving more dangerous overall - they don't - but that designers of safety interventions should model behavioural responses in advance rather than assuming behaviour will stay constant.
A second misunderstanding is that risk compensation applies equally across all contexts. Individuals and groups vary substantially in how much their risk-taking responds to perceived safety changes. The effect is real and documented, but it is not uniform.