Risk warnings are everywhere. You can find them on cigarette packets, medicine bottles, children’s toys and power tools. You see them on TV, in magazines and on the train. You might even notice them on fluffy pillows and greetings cards. But where did they come from? Why are there so many? And do they actually work?
In advertising, risk warnings highlight the possible negative consequences of a purchase or (mis)use of a product. They provide information which ideally helps a potential customer to weigh up whether the product is suitable for them. Warnings are relatively easy to mandate and they hand responsibility to the consumer to know what is best for them. For these reasons, they have been a common policy tool of governments and regulators around the world over the last few decades.
Risk warnings are most effective when products cause high impact, high probability harm, especially when this is not well known. But when they are used indiscriminately and intensely, they may have the potential to backfire.
Research around smoking labels has shown that graphic, larger and more comprehensive warnings on packets increase knowledge of health risks and encourage smokers to quit. However, the evidence for this mostly comes from smokers’ own self-reports. In fact, a lot of existing research on risk warnings in advertising relies on consumers’ testimony of the impact or lack of impact of the warning in a laboratory, rather than observing their actual behaviour in the real world.
In financial services, Cox and de Goeij (2016) found in an experiment on investment prospectuses that risk warnings increased participants’ risk perception by about 5% and decreased the inclination to search for more information by 12%. All well and good. But the messenger seems to be important: references to regulatory approval (in this case, automatic, rather than implying safety or quality) increased willingness to invest and decreased perceptions of risk.
There are many reasons why risk warnings might fail to achieve their objectives. There may be faults with their design, meaning that they are not attention grabbing, well timed, or well understood. Anyone who has ever heard terms and conditions read at high speed over the top of music and other sounds at the end of a radio advert will recognise the impact of this. Or they may not take account of human psychology. Consumers have a limited supply of attention and overproliferation of examples can lead to “warning fatigue” and an inability to identify the highest risk options.
For example, a recent Californian law requires all products containing one or more of a list of more than 900 chemicals to display a warning, such as: 'WARNING: This product contains a chemical known to the state of California to cause cancer'. This has resulted in warnings in supermarkets, petrol stations and restaurants, and on birdhouses and even coffee, which has been mocked by commentators. Research collaborators in the US – Lisa Robinson, W. Kip Viscusi and Richard Zeckhauser – argue that the ubiquity of risk warnings leaves consumers unable to distinguish between danger from 'wolves' (high impact, high probability risks from activities like smoking) and danger from 'puppies' (common but usually low-impact accidents like slipping on a wet floor). The danger is that consumers treat high-risk activities like low-risk ones because 'it’s all the same'.
The Financial Conduct Authority, in collaboration with Warwick Business School and the University of Bath, has published research from 10 experiments. These investigate the effectiveness of risk warnings on social media posts for 10 commonly advertised financial products, including car insurance, pensions and payday loans.
The results show that including risk warnings on initial character-limited social media posts (tweets), compared with including them only on product webpages, reduced consumers’ preference, shopping around and understanding of risks. This ultimately led to them choosing less suitable products.
Unsurprisingly, tweets with risk warnings were less attractive to potential customers. They were then less likely to be clicked on and their webpages explored relative to tweets without warnings. Even when viewing a social media feed where all tweets contain warnings, participants were less likely to shop around compared with a feed where all tweets are not compliant. Ultimately, tweets with warnings were less likely to be chosen than those without.
This might be appropriate if products are very risky or likely to be unsuitable. To rule this out, participants in the final two experiments were given a scenario in which they needed to meet specific needs and levels of risk, while avoiding certain outcomes. Only one product type from a range of options was suitable for the consumer in each scenario, because of product features which were stated in the risk warning. In these experiments, including a risk warning in the tweet reduced the likelihood that participants chose the most suitable product by 7-14%, probably due to reduced shopping around.
The results are consistent with arguments about warning fatigue. There are two hypotheses. Either a glut of warnings made all options look unfavourable, or they made all options look similar, even though the products carried different levels of risk.
In fact, a recent proliferation of research on disclosure has increasingly found that policymakers tend to overestimate the effectiveness of simply warning or informing people. From the FCA’s own research into savings accounts and compensation letters, to the US Consumer Financial Protection Bureau’s CARD Act disclosure nudge and displaying recommended calorie intake in restaurants, the positive effect of improving existing information tends to be small to non-existent.
When you take human psychology into account, this is understandable. While constantly bombarded by attention-grabbing adverts, brilliant blogs and cute cat videos, people don’t notice, or avoid reading, information they find boring, especially if it contradicts their existing beliefs.
So what are the alternatives? Well, firstly, we should use risk warnings only where the evidence shows us they are effective – usually for high-impact, high-probability risks that are less well known. And we should test risk warnings in different contexts to understand how they might perform, especially against other, stronger measures, like sales restrictions. We should also explore any potential long-term educational effects of risk warnings, which can help us to identify the situations where risk warnings have value.
We can also consider the effective use of defaults and nudges, as popularised in Richard Thaler and Cass Sunstein’s 2008 book, Nudge. While people may not take the time to consciously read information, they will respond to cues such as the ordering of options. And if they fail to make a decision, they will be better off if the default they end up with is a good one.
Regulators could go one step further and put the onus on advertisers and companies to prove that their customers understand key features of the products they buy. Lauren Willis, Professor of Law at Loyola Law School, argues in favour of performance-based consumer law. This advocates 'confusion audits', in which firms test how well their customers understand what they have bought – and allows regulators to restrict sales or 'hidden' fees if they don’t. Precedent includes regulation of US medicines, where producers must prove that customers are using the medicine correctly according to its risks before it is allowed to be sold over the counter.
In financial services there is a mass of disclosure, built up in international, European and domestic law over many years, often based on difficult circumstances or case law. Risk warnings undoubtedly have an important place in the policymaker’s toolbox. But our research warns against overestimating the effectiveness of disclosure remedies. As we consider new or review old regulation, we need to make sure we can filter out the noise so that consumers can hear the most important messages.