Beyond disclosure for high-risk investments: slow down and think

Research Published: 02/02/2022 Last updated: 02/02/2022

This article by Lucy Hayes, Cameron Gilchrist, Max Spohn and Cherryl Ng at the FCA explores our research into risk warnings, decision points and self-certification, published in January 2022.

 

It’s never been so easy to invest your money – anyone can do it online in just a few clicks. Low interest rates and the increasing social media hype are driving more and more people in this direction. Over 1 million UK adults (6% of all UK investors) increased their holdings in high-risk products, or purchased new high-risk products during the first 7 months of the Covid pandemic – a time of increased consumer vulnerability.

However, making the right choice can be complicated – promises of high returns come with significant risks, including losing the entire investment. Investors surveyed recently by researchers for the FCA seemed to significantly underestimate these risks, with 45% of new self-directed investors not aware that ‘losing some money’ was a potential risk of investing.

There are restrictions on access to high-risk investments. For example, consumers have to self-certify as either ‘high net worth’ (wealthy enough) or ‘sophisticated’ (experienced enough) before they can access certain high-risk investments (like speculative illiquid securities). However, the FCA has revisited this regime due to a concern that too many customers do not understand the impact of self-certifying, or may just be clicking through without properly considering whether the investment met their needs.

In addition, the FCA has recently launched a campaign targeting young investors and is also proposing changes to help strengthen the journey into these investments so consumers better understand the risks. In order to inform this drive to help consumers maker better investment decisions, we carried out a programme of behavioural research.

Decision points

Driving better choices

Our research used behavioural science to improve consumer understanding of the risks involved in high-risk investments, and reduce the numbers of people self-certifying as high net worth or sophisticated to more realistic levels.

We focused on how to achieve this through the use of ‘decision points’ – steps added to the consumer journey and designed to interrupt automatic behaviours. Decision points aim to put people back into a more deliberative frame, giving them time to pause, read and reflect.

One way to do this is by providing information to consumers at a critical point in time, which we refer to as ‘disclosures’ in this article. Disclosures can not only help people understand financial products, but also draw attention to neglected considerations like risk, perhaps deterring some consumers from buying products that are unsuitable for them. Some of us have tested and shown the effectiveness of disclosures in the past, including risk warnings about the potential losses from minibond investments and on fees involved in online investments. We built on this in the current work by creating behaviourally designed risk warnings that are relevant to other high-risk investments such as crowdfunding and cryptoassets.

Another way to introduce decision points is by using ‘positive frictions’, which are processes designed to slow people down and make them consider their actions more carefully. In a recent study, Twitter users were forced to open links and articles before tweeting them, with the aim of reducing the spread of misinformation. In another study, consumers were shown warning messages on their phones when pre-set credit card spending limits were reached, which had to be clicked away to spend more, resulting in lower spending levels.

Experiments

Testing interventions to promote investor deliberation

We conducted 3 online experiments that brought together these ideas, with a representative sample of several thousand participants.

In experiment 1 we used a range of behavioural principles to redesign risk warnings. The standard risk warning typically found on financial promotions is ‘your capital is at risk’, which is often in the small print. This risk warning does not clearly convey the potential losses to consumers and is often perceived as ‘white noise’ by investors.

We tested this standard warning against a new warning that contained clearer information about the specific risks (including that consumers could lose all of their money), which was presented more saliently (bold text, red background). We combined this with a link to more information and made it clear that reading this information would be quick. The behavioural framings we used included a social information framing, which drew attention to the fact that high-risk investments are actually not that popular and loss aversion – the principle that people feel losses much more strongly than equivalent gains.

Figure 1 below shows our baseline or ‘control’ condition with the standard risk warning, and our loss aversion treatment beneath it.

Figure 1: Risk warnings tested in experiment 1

Risk warnings tested in experiment 1 (control version)
Risk warnings tested in experiment 1 (loss aversion treatment)

Risk warnings on financial promotions are only one step in the consumer journey and we wanted to test the impact of disclosures and positive frictions at other stages. In experiment 2 we introduced an FAQ style disclosure about the key investment risks, combined with several positive frictions such as tick boxes for consumers to confirm they understood some key risks, and free text boxes.

See the examples we used for experiment 2

In experiment 3 we tested interventions to the current process of self-certification. The positive frictions we introduced here included active choice checkboxes for consumers to confirm which criteria they met, free text boxes for consumers to provide evidence that they met these criteria, and the imposition of a time delay before consumers could submit their certification statement. We also simplified the text to make the purpose of self-certification easier to understand and added an additional screen where consumers had to declare that they accepted the risks associated with self-certification.

See the examples we used for experiment 3

Results

Some friction, some disclosure

Across the board, we found that our behaviourally informed disclosures enhanced participants’ understanding of risk because of increased salience, simplified text and clearer explanations. Our disclosures in experiments 1 and 2 also reduced the chance of participants recommending the investment to a friend.

Figure 2 shows the results when our behaviourally informed disclosures from experiment 1 are tested on a fictional crowdfunding product. It shows that the likelihood of correctly answering a comprehension question is significantly increased by all our treatments, compared to the control. No single individual treatment is more effective than the others, however.

Figure 2: Average likelihood of correctly answering a comprehension question for the crowdfunding product, across all treatments (experiment 1, sample size: 6,618)

Experiment 1 results (crowdfunding product)
- The p-value (p) is a measure of statistical significance, where values larger than 0.05 are considered not to be statistically significant: ***p<0.001, **p<0.01, *p<0.05

- The 95% confidence interval is displayed at the top of the bars representing our treatments

 

Figure 3 shows the effect of interventions from experiment 2. Again, all the treatments significantly increase the likelihood of correctly answering a comprehension question correctly. Our analyses show that the FAQ disclosure ('summary info'), rather than the positive frictions ('active click', 'active input' and 'personalisation'), was the main driver of the increased understanding. Note that our consultation paper (CP22/2) refers to all of the interventions in this experiment as 'positive frictions'.

Figure 3: Average likelihood of answering a comprehension question correctly, across all treatments (experiment 2, sample size: 4,008)

Experiment 2 results (comprehension question)

Figure 4 shows the results from experiment 3 on self-certification. We found in a separate survey that 16% of participants reported they were eligible to certify (based for example on their income), so we took this as our baseline. In the control group, self-certification levels were over 50%. Our interventions were successful in reducing this, although a much larger proportion than 16% still self-certified.

The treatments in experiment 3 were additive, meaning that each intervention built on the preceding one. Therefore, whilst all treatments in Figure 6 appear effective, we can infer by reading the bars from left to right that the reduction in self-certification levels was driven by the inclusion of the ‘active’ treatment and the ‘evidence’ treatment, and not by the additional interventions embedded in ‘simplify’, ‘time’, and ‘risks’.

Figure 4: Proportion of people self-certifying across all treatments (experiment 3, sample size: 8,094)

Experiment 3 results (self-certification)
- The 0.16 line indicates the 16% of participants we would expect to self-certify in our experiment 

 

Some of our frictions in experiment 3 actually caused people to drop out of the process altogether. This attrition was most likely to occur where we asked participants to provide short supporting statements if they want to self-certify (‘evidence’) or forced participants to acknowledge the risk of self-certifying in order to continue (‘risks’). This was an unintended (although perhaps predictable) outcome. Despite this attrition, a clear majority of participants (77%) still completed the experiment when asked to provide evidence declarations. As with all participants in our study, we do not know if the people who dropped out really did meet the criteria for ‘high net worth’ or ‘sophisticated’.

It is possible that dropping out of the process could be a good outcome for some consumers. After all, if they found these additional checks too onerous or frustrating to complete, then arguably it’s unlikely that they were in the deliberative frame of mind required to invest in a high-risk investment with a substantial chance of losing all their money.

Discussion

Financial decision making and friction

The stakes for high-risk investments are high, and our research informed the range of measures the FCA proposed in January 2022 to protect consumers.

Our experiments show that behaviourally informed reminders and information still have an important role to play in bringing attention to and helping consumers understand investment risks.

The value of positive frictions has been demonstrated in other contexts and we were somewhat surprised not to see larger positive effects in experiment 2. This may have been due to the risk warnings and information already doing a lot of the work to encourage people to stop and reflect on the risks.

We also suspect that positive frictions, designed to interrupt a decision-making process, might alter people’s behaviour through many channels, not just by inducing reflection and in turn a fuller comprehension of relevant information. In experiment 3, our most effective friction was where we asked people to provide evidence for their self-certification. We believe this may have reduced self-certification for several reasons. It may have made it harder for people to knowingly incorrectly certify whilst still maintaining a positive self-image, or whilst being able to plausibly deny that they had not incorrectly certified. The intervention may also simply have worked because the process made it harder for people to self-certify.

Furthermore, the investment settings in our experiments were hypothetical and so lacked the high stakes of real-world investing. A possible extension would be to test positive frictions in the field, where we could see clearer effects because investors would be making decisions with their own money. There are likely to be many other areas where positive frictions could help introduce moments of reflection during other key financial decisions.

Read more about experimental design, the interventions we tested, and the results in our 3 Research Notes.