Paragraph 14 of the discussion paper notes that partly automated processes have been observed across the three sectors “in the provision of some banking products (e.g. mortgages, loans, and savings accounts), where the consumer uses the tool to obtain advice on the products up to a certain point”.
When discussing the provision of automated advice it is important that the distinction between providing regulated financial advice and information does not become blurred – for example, where a consumer is using a platform provider’s ‘best buy’ list, it may not be clear to them that any investment decisions they make are at their own risk. If a consumer is using a service which is merely a mechanism to facilitate a sale (i.e. execution only), this is also not regulated advice. Any automated service which does not deliver a personal recommendation based on the characteristics of the customer should make it clear to the consumer that it is not an advice service.
The Panel believes that providers could do much more to make it clear whether their service is regulated advice or not: consumers are likely to use the common meaning of the word ‘advice’, which can have serious implications if they buy an unsuitable product through an automated service.
Several automated advice services claim to offer advice when in fact they are not authorised to do so. One example is www.moneyfarm.co.uk which says on its homepage that “our advice is totally unbiased”. In its FAQs section (https://www.moneyfarm.com/uk/faqs), however, it says that it does not give investment advice, as it is a discretionary investment manager. There is a similar lack of clarity on other automated advice websites. This underlines the need for clearer signposting of the protection available to investors, and for additional risk warnings on non-advised sales.
In addition, an automated advice process should not allow for an immediate sale, but give the consumer time to consider the personalised recommendation made. There should also be an option for the consumer to speak to a qualified advisor. This would provide an opportunity to spot an erroneous input, skewed risk and suitability assessments or flawed recommendations.
Moreover, automated advice which is marketed as ‘independent’ should adhere to the same standards as face-to-face independent advice, notably the ban on commission for independent intermediaries introduced by Directive 2014/65/EU (MiFID II) and the requirement to consider a sufficient range of different product providers’ products.
Automated advice tools in the investment sector typically have low minimum investment levels and most recommend portfolios of low-cost passive investments, such as ETFs. As the number of people using automated advice models grow this may begin to impact on how markets behave. For instance, if large numbers of investors are in passive investments they will be more likely to see the value of their portfolios fall in line with markets, with no active management to mitigate against that loss of value. That may exacerbate the impact of a downturn on certain sectors and stocks.
In the UK, there are firms that are trying out new and innovative business models, including some regulated investment and pensions advice through a mixture of telephone, email and Skype. Two market leading providers have developed fully automated, fully regulated, advice models – LV= and Just Retirement. These models provide a personal recommendation signed off by Level 4 qualified advisers. The firms also accept full liability for the advice provided and consumers are covered by the Financial Ombudsman Service and Financial Services Compensation Scheme.
Money on Toast provides a regulated restricted advice service which focuses solely on investments and does not take a complete overview of a consumer’s personal circumstances. This comprises a questionnaire designed to get a view of the individual’s investment understanding and risk tolerance. It gives a grade for risk and then asks further questions about risk capacity, after which the risk grading may be adjusted. The investor is then sent a suitability report and recommendation.
Other services are closer to the discretionary advice model, from specialists such as Money on Toast and Nutmeg to large investment firms including Brewin Dolphin and Hargreaves Lansdown. Investors using Nutmeg, an execution-only service, can choose the level of risk to take, provide details of their assets and take questionnaires covering affordability, risk and capacity for loss. They can invest for specific goals and select different risk levels for each of those goals to produce a specific outcome. Users can contact a support team which can answer questions about the process, but can’t give advice.
The extent to which automated advice services establish investment objectives, suitability and risk can vary significantly between firms.
The ease with which consumers can access online services can be appealing. Services that aim quickly and efficiently to recommend an investment portfolio using online questionnaires can save consumers time and effort.
While the Panel accepts that the use of automated services to deliver advice could reduce costs to both firms and consumers, we have not yet seen any convincing evidence that costs for traditional delivery methods for advice are causing gaps in the availability of advice to different groups of consumers. Moreover, the potentially lower costs come at a price identified in the discussion paper: where advice is fully automated, limitations inherent in the design of the advice process with no or very limited human intervention may result in inappropriate recommendations to the customer, potentially resulting in mis-selling.
We believe it is misleading to focus only on the costs of automated advice. It is also important to take into account the costs consumers incur if they buy a product. The total cost of ownership once a consumer enters into a transaction following advice (automated or not) are often well above the 1% ‘headline’ figure, once costs including the wrapper, ongoing management costs and ancillary fees such as exit charges are taken into account.
In the Panel’s view, the more complex the product the more challenging it becomes for providers to build a fully automated advice service which delivers the same quality as a human adviser. Therefore, the benefits of a fully automated advice delivery mechanism are likely to be relatively higher for straightforward retail banking and general insurance products, and lower for complex investment or long-term insurance products.
An automated advice process that is subject to less stringent consumer protection standards than traditional methods of delivering advice (e.g. face-to-face) offers financial institutions an incentive to offer automated advice, as the potential liabilities for unsuitable recommendations are less costly. At the moment it is possible to offer a reduced-liability automated service for execution-only. Extending that to automated advice would allow firms to market a broader offering but still restrict their liabilities. It is therefore being seen by UK banks as a low-cost, low-liability route back into mass-market investment advice, allowing them to sell their own products in large volumes.
To avoid creating regulatory arbitrage, it must be clear which type of advice is subject to regulatory standards so that the consumer is aware whether they are protected or not if things go wrong.
Yes. The Panel is concerned in particular about the following risks, which to some extent are inherent in an automated advice model:
• The risk of a systematic flaw in the design of the algorithms underpinning the automated advice services, whether deliberate or accidental. This could, for example, exclude certain product categories or providers from any potential recommendations. The expertise required to calibrate the algorithm is such that it may be only a handful of people who know when there is a flaw so errors may take longer to identify and resolve;
• Risk profiling methods have been prone to error in the past, as shown by the FCA in its work on clarifying the boundaries between advice and ‘non-advice’ . In addition, research by the US Securities and Exchange Commission in 2015 found that many of the assumptions and recommendations made by online portfolio planners didn’t properly account for changing individual and/or market circumstances. Poor risk profiling and/or flawed metrics could result in widespread poor outcomes and unchecked systematic ‘misbuying’;
• Erroneous consumer input could skew the outcome of the advice session, for example a wrongly entered date of birth or income level. There is also evidence that people don’t understand their own appetite or capacity for risk. A Finametrica study based on more than 500,000 investors who have taken its psychometric risk test found that 21.2% incorrectly estimated their true risk tolerance by a ‘significant margin’. Another 17.2% were incorrect by between 11 and 20 points and 4% got it wrong by more than 20 points .
• Investors may be ‘shoehorned’ into convenient but unsuitable portfolios, particularly when the service offers a limited range of generic pre-defined portfolio options. Accordingly, the Panel believes that any automated advice service should include some form of human intervention to check the data being entered and the recommendation produced by the automated system. This is a necessary ‘security check’ to ensure the advice is suitable for the customer.
The paper touches on the potential for automated advice usage to create a ‘herding’ risk. However it doesn’t appear to cover the ‘panic’ risk that may be inherent in an automated process. In other words, where there is no human adviser to provide insight or reassurance there is a distinct possibility that investors will make poor decisions in the event of volatility or a bear market. How investors react in the absence of the ‘hand-holding’ element of face-to-face advice remains to be seen.
Understanding emotions around investment risk – and managing them - is where much of the value lies in face-to-face financial advice. Automated advice at its current evolutionary stage is inevitably unable to perform that role so the emotional and psychological behaviour of consumers when interacting with these tools is still an unknown.
Typically, consumers looking for long-term growth will need to consider riskier investments. This will lead them away from core bank savings products to higher risk investments, such as property and stocks and shares.
Yes. There are other factors to consider, however. For instance, the value of the sector may ultimately lie in the data produced by automated services. The data collected in risk profiling especially can enhance our understanding of consumer behaviour, helping the industry provide better products and services for consumers.
However, on the downside, the assessment doesn’t sufficiently address the risk of automation further entrenching financial exclusion. In the UK alone there are still around six million people without access to the internet (particularly in rural areas, and among low income and vulnerable households).
In the UK (and in other markets) the evolution of automated advice will feature the entry into the market of the country’s large retail banks. The concern is that high street banks view automated advice as a route back into mainstream advice, providing a cheap, low liability way of distributing their own investment products. The history of investment mis-selling by those banks raises concerns over the potential for serious and widespread consumer detriment.
The fact that the regulatory and common definition of ‘advice’ do not coincide continues to be a source of confusion to consumers, but whether a service constitutes regulated advice is of crucial importance in ensuring that consumers receive suitable advice and make well-informed purchasing decisions.
Automated services offered online which are merely platforms for transacting a sale, such as many price comparison websites, can ‘masquerade’ as advice providers, without having any liability for the decision taken by the customer on the back of their recommendation.
In the Panel’s view, automated advice should be subject to the same standards as other forms of financial advice. Advice is either regulated or it isn’t, and providers should make this clear to customers up front. If the advice is not regulated there should be a clear warning to consumers that, if they buy a product or otherwise act on the advice, they have no recourse to alternative dispute resolution or redress. This will bring much-needed clarity to the market, especially giving rapid growth in the provision of automated advice services.