The AIC agrees with the potential benefits described for automated advice. It has the potential to widen consumer access to advice as well as providing it at a lower cost. Automated advice may also make it easier for firms to provide cross-border transactions, though many consumers may prefer to deal with familiar, domestic companies. Consistency may be one of the outcomes, but that also raises questions about the algorithm used and the absence of human judgment (see comments below). For example, automated advice may be able to cope relatively well with comparatively simple investment scenarios. It may be less able to cope with the concept of building or adding to an existing portfolio and spreading risk across a variety of assets (which present different levels of risk) within that portfolio.
The AIC agrees with the potential benefits described for firms. Automated advice may provide access to a wider pool of consumers and may have benefits in terms of the service provided. In addition, the ability to provide automated advice could potentially enable firms to provide services across borders more easily. However, as the EU has noted in its Green Paper on retail financial services, there are still barriers to persuading consumers to access cross-border services. These include language barriers, differing provision of complaint and compensation mechanisms etc.
Some of the benefits to firms, however, may be at the expense of consumers if automated advice directs consumers towards decisions that are convenient or beneficial for the firm rather than the optimum solution for the consumer (see comments below). Automated advice should not be used to limit investment options. In the UK, for example, independent advisers are obliged to consider all types of retail investment products which are capable of meeting the investment needs and objectives of a retail client.
The AIC agrees with the potential risks to consumers that are identified.
The most significant risk for consumers is whether an automated system can replicate the judgment and level of tailored advice that a professional adviser can provide. Advice has to take account of the specific situation of the individual. Different consumers will have a wide range of factors that affect their choice of products. These may include factors like their age, tax considerations, appetite for risk, their existing portfolio etc. Whilst factors such as age may be straight-forward to include in an automated system, it may be considerably harder to take into account all the relevant considerations for pension planning. Alternatively, where individuals hold a portfolio of investments, a relatively more complex or risky product may be suitable as part of the portfolio when such an investment would not be suitable if it was the only investment held by the investor. This is where the judgment and knowledge of an adviser may be very hard to replicate.
As the paper notes, the quality of ‘advice’ that is produced will depend on the quality of the tool. Given the range of investment products that are available, there must be a risk that consumers will be channelled into specific groups of products. This may be an inevitable result of the way that any system is put together, but it creates risks that the advice will not meet standards such as those of MiFID II which requires advice to look across the market. This could result in consumers not being given the best range of options and some types of investments being overlooked. Any inherent bias in the system is unlikely to be apparent to consumers who will be unaware that the advice given may not take account of the whole range of products that would be suitable for them. Automated advice should not be allowed to alter the status of independent advice.
It is also correct to say that automated advice could raise issues of cost transparency. Costs to the consumer could easily be hidden within such an arrangement. It will create challenges for regulators and others to ensure that an automated system is meeting requirements such as those of MiFID II which require transparency about costs.
There are different risks that arise in the three sectors.
Many insurance and banking products can be more standardised than investment products: non-life insurance cover for example. Most householders are likely to purchase broadly similar insurance cover for their house, although they will be able to choose from a range of features and levels of cover. The features may be comparatively straight-forward to explain to consumers and the factors that would make a product right for that consumer, may be far less complex than an investment product. Different mortgages will provide similar benefits to consumers, although may have different terms or fees. In addition, in the UK for example, there is extensive online information and comparison sites that provide consumers with information to help with their decisions.
Investment products can be considerably more diverse, with a huge range of investment strategies available and a wide range of different features and risks that can be difficult for consumers to understand (such as some of the features of structured products or synthetic products that try to replicate the performance of other assets). The impact of any investment decision may not be apparent to the consumer until years after the purchase.
Investment products, as a whole, are less uniform than insurance and banking products. The wide range of investment products and accompanying features will make it harder to create automatic advice that is able to take account of both the nature of investors’ needs and the range of products that could be suitable for an individual investor.
The AIC agrees with the potential risks set out for those firms that offer automated advice. As the paper notes, there are potential risks for both consumers that use automated advice and firms that provide it. In particular, there may be risks for consumers if the system produces biased advice (see Q 14 above).
The risks from automated advice do not just extend to those firms that are offering automated advice.
As discussed above, there are risks for consumers in the way that automated tools are set up and their ability to look across the whole of the relevant market. There is also a risk that automated tools will not be able to consider the most suitable investment or that they will reflect any investment bias of the firm operating the system. For example an automated tool might favour UCITS products over other rival investments, or tracker mortgages over fixed rate mortgages.
This creates risks for product manufacturers that do not control distribution of their products. They could find that their sources of distribution are reduced if automated models do not consider all products, or are inherently ‘biased’ towards certain classes of product. This would be an undesirable outcome for both consumers and those firms where distribution of their products could be affected.
There are potentially risks for institutions arising from all three sectors. For example, if a consumer buys inadequate or inappropriate household insurance and then suffers a major loss, this brings with it the possibility that the firm will be held responsible for giving unsuitable advice.
However, as set out in Q16, the range of products and the range of factors that may impact upon what makes an investment product suitable, is likely to make the risks of selling investment products on an automated basis, higher than for banking and insurance products.
The AIC agrees with the assessment of the potential evolution of automated advice. Consumers in the UK are becoming used to conducting financial transactions through a variety of means – including the phone. This engagement with financial services without any human involved in the process is likely to encourage the adoption of services such as automated advice. There are already firms in the UK that will provide financial advice on securities on an automated basis.
As the paper notes, automated advice has the potential to bring benefits to both firms and consumers. Consumers can benefit through wider availability of potentially cheaper advice. Firms will gain if automated advice encourages additional consumers to use financial services products.
The risks to consumers and to firms that do not control the distribution of their products will need to be carefully considered. Consumers need to be confident that an automated model is able to deliver genuinely independent advice and that there are no hidden constraints or a bias towards certain outcomes within the model.
Ensuring the independence of the automated model will help to prevent some products from being shut out of the market to the detriment of both firms and consumers.