Governments and other regulatory bodies should ''over-invest'' in safety when there is public uncertainty about such widely publicised risks as terrorist attacks and infectious diseases. That is the central conclusion of research by François Salanié and Nicolas Treich, published in the April 2009 issue of the Economic Journal.

Their study provides an economic rationale for a systematic ''over-regulation'' of risks when people''s beliefs about risks are distorted – whether they are too pessimistic about the risks, as is typically the case with potential flu epidemics or acts of terrorism; or too optimistic, as with home safety, car seatbelts and tolerance of food additives.

Citizens often hold beliefs about risks that systematically differ from the beliefs held by leading experts. These differences in risk perceptions may be related to the well- documented tendency of citizens to overestimate risks with small probabilities, or risks that have received significant media attention. Examples include hazardous wastes, drinking water pollution, terrorist attacks, mad cow disease or swine flu.

There is also evidence of a great deal of variability in spending on public prevention across different risks. And putting together these two bodies of evidence suggests that risk regulators respond to citizens'' beliefs about risks.

Regulators may respond to citizens'' beliefs by political opportunism, or simply because they have similar perception biases. It may be argued, however, that regulators'' responsiveness is an expression of well-functioning deliberative democracies, and is in a sense perfectly legitimate. But is it economically desirable? Salanié and Treich''s research suggests that it is.

Whether or not regulators should respond to citizens'' beliefs raises a puzzling question. This question can be illustrated by the fable of Happyville, first described by Paul Portney in a 1992 article called ''Trouble in Happyville'', published in the Journal of Policy Analysis and Management.

Happyville is a society in which people believe that the drinking water is contaminated but all experts agree that it is harmless. What''s more, any attempts to inform people that the water is safe have no effect. People remain anxious and they urge the Director of Environmental Protection to invest in a water cleanup technology.

What should the Director do? Even though the Director is convinced that citizens'' beliefs are erroneous, the decision of whether to invest in a water cleanup technology is a difficult one. Consumer sovereignty arguments support the Director's decision to purchase a cleanup technology. This would improve welfare since worried people would ''feel'' protected.

But regulatory decisions should be driven by risk-related facts, not by citizens'' uninformed worries. There is an obvious opportunity cost here, as the money spent to prevent a phantom risk could be used for the prevention of other risks, and therefore lead to ''statistical murders''. Yet this argument is controversial as well, as it relies on a paternalistic view of citizens'' preferences. It presumes that the regulator knows better than the citizens about what is good for them.

This new study proposes a behavioural economics approach that relates risk perception to risk regulation. It analyses the optimal paternalistic risk policy. It recognises that citizens may make faulty choices due to their biased beliefs, and that a regulator may rationally anticipate these choices and may want to curb them.

Consider the fact that thousands of tourists cancel their travel plans after a plane crash, a terrorist attack or an outbreak of an infectious disease, with the associated negative consequences for the economy. This analysis suggests that a paternalistic regulator should rationally respond to these events by increasing investments in safety compared with the case where there are no risk misperceptions so as to mitigate these negative consequences.

Similarly, a regulator should raise standards for food safety to prevent a drop in consumption after bad news about health-related food quality. In other words, a paternalistic regulator should over-invest in safety so as to ''encourage'' pessimistic citizens to increase their risk-exposure since this exposure is too low in the first place.

But the most striking result is that a more stringent prevention policy may be always warranted, both in a society with pessimistic or in a society with optimistic citizens. Indeed there are two effects: the ''encouragement effect'' for excessively pessimistic citizens, and a ''protection effect'' for excessively optimistic citizens.

The protection effect allows the regulator to reduce risks when citizens are prone to take too much risk. Examples may include home safety prevention, compulsory seatbelts and tolerance standard for pollutants or food additives. The protection effect may overpower the encouragement effect when citizens are optimistic and vice-versa when citizens are pessimistic.

So the study provides an economic rationale for a systematic over-regulation of risks when people''s beliefs about risks are distorted, no matter whether the direction of distortion is towards pessimism or optimism.

''Regulation in Happyville'' by François Salanié and Nicolas Treich is published in the April 2009 issue of the Economic Journal.