Response to joint Comitteee Discussion Paper on automation in financial advice
Go back
As an example, the selection of funds is one of the major aspect of financial investment advice and this activity cannot be achieved with a robot. Another example is strategic and economic analysis; there too the use of algorithms may contribute to the work of prevision, but human intervention will always be necessary in order to check the results of algorithms and to take into account data and events which are not predictable.
Automated advice should not give pace to the possibility for some actors to cope at a good price with the new requirement of investors’ protection regulation as far as ‘clicks’ are much an easier way to get through these requirements than asking questions to the client during a rendezvous.
In order to maintain a good level of investor protection, it will probably be necessary to introduce some kind of qualification and authorization for ‘robot-advisors’ with due diligence requirement that will permit to check the quality of algorithms. In addition, it is important to prevent investors from having the illusion of receiving a financial advice whereas, in fact, what is given to them does not rely on a serious basis; also is it important to protect them from the risk of biased advice which would be provided by unregulated actors acting in connection with some product manufacturers.
The description of automated financial advice tools provided in page 14 is quite relevant but it only refers to data related to the investor. As useful they may be, they will not be sufficient in order to provide good advice. Economic and financial analysis is also necessary, as well as analysis of the intrinsic value of the available instruments. For example:
- if one must achieve an investment during a 14 years period (for example before retirement) perhaps will it be more profitable for him to invest successively in different funds during 5 to 7 years instead of investing in a single fund during14 years;
- the choice between stocks or bonds will have to take into consideration the level of market prices and the level of interest rates, etc.
In addition we observe that, quite often, retail investors have not a precise view of what they wish beyond an increase of their fortune combined with a minimum risk… Therefore, when a default option is proposed – as it is the case with the French PERCO for example – it uses to be adopted by a vast majority of retail investors.
Apart from that sector, for years Credit Agricole group has also used an algorithmic tool which matches the profile of the client and its financial needs with a great variety of investment products.
See the description of our Group in the executive summary.
ii) the service you provide (e.g. to what extent do you integrate human interaction in the tool you provide?);
We are presently working on an online investment advice service which will provide funds allocations. Each recommended allocation will have to fit with the client profile and with his savings project. Allocations are produced by a system which is based on Harry Markowitz’s theory (efficient frontier, diversification, etc.) which extracts an allocation from a funds’ universe defined for the client with an opened architecture principle i.e. a set of funds internally and externally managed. The algorithm has been framed by a fund selection manager who has performed a fund categorisation before setting the various allocations. Hypothesis underlying these allocations are regularly reviewed by the management team. The client’s profile results from a suitability test. In case answers to the suitability test would not be coherent, an adviser of Amundi would get in touch with the client.
When accepted by the investor, the allocation will be subject to ongoing review and some alerts will be sent when necessary to the client offering the possibility of talking with an adviser.
iii) the nature of your clients;
This service is free and proposed to existing clients.
iv) your business model;
We observed that a large part of our retail clients’ savings are invested in money market funds; through this offer of digital advice we wish them to shift to other funds which may provide better returns.
v) who developed the automated tool (i.e. an external company or developed internally?);
As explained, it has been developed internally.
vi) the size of your activity and/or forecast activity?
This service will probably have a progressive development.
We do not believe that digital advice will lead to any important cost reduction and we do not see this point as a main advantage. In fact, good financial advice requires human intervention at a given point of the distribution process. For years, when e-business began to develop, banks were wondering whether to maintain their branch network. The experience has shown that the combination of a website for payments in line, account consultation etc, with a good service and advice in local branches could optimize client relationship. We do believe that the same thing will occur with automated financial advice. In fact, only a human advisor is able to provide truly personalized advice, evaluating a consumer’s priorities against the complexity of family histories, guide a consumer through unexpected life events, explore the full range of products and choices with a consumer, consider all accounts, products, investments, etc. a consumer holds, and have a dialogue with the retail investor which will be crucial in particular during times of financial panic or euphoria. In addition, people with more savings use to be elder clients often retired, who like coming into the branch.
Benefit 2 (§ 31/32):
We do not share the view that many clients would feel not wealthy enough in order to ask for financial advice; at least it is not the case in France. French Banks have very large networks spread all over the territory and cope with mass market clients who feel comfortable in the branch where they have their account. Nevertheless, digital access may provide an alternative in some rural areas.
Benefits 3 (§34):
In the present state of things and due to fiscal disparities, cross border benefits will be quite thin. In addition, the language is an obstacle, at least in southern Europe. At this stage, it is important to distinguish between pure advice tools and transactional websites providing advice before investing on line. Of course it may be quite easy to provide cross border pure advice, but when coming to transaction, it is much more complex for investors – and even impossible – to use different cross border providers, due to the fact that these providers are not connected with a sufficient number of depositaries. This blocking point could be more easily overcome in the field of individual pension products which are not subject to the same running than funds (no need for securities account). We hope that the PEPP initiative will provide the legal framework to ease the cross border distribution of these plans.
Benefit 4 (§35):
We do believe that this advantage is key, and it will be all the more interesting for clients if they are provided with simulators and calculators that will help them to get some kind of valuation of their future savings according to various scenarios. But we do believe that after this first ‘weekend step’, most of retail clients will prefer to have an exchange of views with their financial adviser and will very often need some additional explanation and precision about some feature of the products and funds.
Benefit 5 (§36 & 37):
Algorithms allow for some kind of standardization of the advice and as explained earlier, most networks use advising tools; the success of Harvest’s Big Expert is an illustration of this and various internal tools have been developed by most banking networks. The alliance of these tools with a good human expertise provides best results, easing the initial phase of suitability and the storage of useful information about the situation and characteristics of each client.
Benefit 6 (§38):
In the context of savings plan, real time information may be detrimental rather than beneficial, inducing short-termism. For example, when facing extreme volatility in financial markets as it is presently the case, real time information would generate unnecessary stress and could lead to very damaging decisions.
Benefit 7 (§39):
Rather than for the client, it will allow the adviser to keep easily more data in relation to the client and its savings in order to provide consistent advice during the whole period of products detention.
- Digital advice with simulators may introduce some kind of play or leisure in the investment decision, thus facilitating wide spread financial education.
- It may introduce a certain geographical equality between well distributed areas and more desertified regions.
- More autonomous retail clients will find direct access to products with a minimum human intervention at a better price. But nowadays, this is not true for funds as explained supra in our reply to Q6, benefit 3.
With respect to what has been said earlier, this benefit is not obvious because it would not be relevant to get rid of human advice; digital advice tools do not exclude the need of human advice or intervention. It can only reduce this need in certain circumstances thus allowing for some kind of thrift.
When looking for the quality of advice, we do believe that digital tools may be more useful when employed by advisers as mentioned in the executive summary, instead of having the client left alone in front of his computer. In fact, the greatest part of retail investors have a long term perspective; in this context the longer the investment will be, the lower the relative cost of initial advice. What most matters is the quality of this advice. Thinking that robots may provide better advice than experienced professionals would be somewhat promethean. This being said, we can admit that digital tools may provide good support for the following up of the achieved investment.
Benefit 10 (§42 & 43):
Amundi does not agree with this benefit. As far as financial is anything but an exact science, we do believe that the potential for differences due to human interpretation is a good thing. The risk of short-termism already pointed must also be thoroughly considered. Last but not least, the experience shows that the origin of each new financial crisis is always different and difficult to predict.
In another perspective, digital distribution may provide investment firms and asset managers with a direct access to new retail clients that they do not access through external distributors or intermediaries.
With respect to §54, we would be somewhat doubtful with regards clients’ needs. In our view, it does not make much sense to identify a saving product aiming precisely at the payment of children’s studies. A lot of fuss has been done around this topic of clients’ needs which is more marketing argument than serious financial approach. With four main criteria, it is possible to fit with specific needs of clients: length of sparing, search for revenue or accumulation, level of risk. It is important not to make the confusion between investment objectives and the intended use of savings.
We also agree with the analysis provided in §59.
For what is of the tailored recommendation, it is important to stress on the fact that financial advice always incorporate a part of assumption linked to the future of financial markets, which is a very uncertain exercise, may it be the result of human analysis or of algorithms.
In France, the risk identified in §66/67 is limited due to the supervision of the CNIL.
Risks of §68/71 linked to unsuitable decisions because of limitations or assumptions or errors within the tool are further arguments in favor of digital tools combined with human advice. In the case of websites providing direct advice to retail investors, the financial expertise of experienced fund managers within the firm operating these online tools is essential.
The risk of haste identified in §74 can only be coped with the help of human intervention.
We consider that the “herding risk” mentioned in §77 would clearly occur in case of widespread automated advice and we also share the concern of a potential extinction of human financial advice in favor of digital linked to new regulatory burden. In fact, it would be much easier for digital operators to comply with all suitability and information requirements introduced by MiFID II and PRIIPs: some rapid clicks will allow the client getting through without reading any KID or other “boring” documents and the investment will be concluded…
This being said, the ESAs should consider how to prevent potentially miss-selling due to automated advice. Some kind of qualification and authorization would probably be necessary for automated advice driven actors in order to provide the good level of retail investors’ protection. It will be important to avoid the use of these tools by unregulated actors who could mislead retail investors as it has been the case with the offer of forex or CFD products.
In addition, it will be important to clearly distinguish between financial digital advice and tools that merely guide investors to make their own investment decisions on a well-informed basis. This topic is somehow treated in MiFID as it may be compared to the distinction between products placement with financial advice and execution only service; it should be taken into consideration when looking for a specific regulation.
1. Do you agree with the assessment of the characteristics of automated financial advice tools presented in this Discussion Paper? If not, please explain why.
Amundi is of the opinion that digitalization may help to improve advice in so far as it is combined with human advice. The mix between human intervention and digital tools may be quite different according to the type of actor; but we do not think that good advice could be 100% digital. For example, it seems quite obvious that a robot adviser will never be able to provide and maintain a tactical portfolio allocation of the same quality as what would be provided by an experienced professional.As an example, the selection of funds is one of the major aspect of financial investment advice and this activity cannot be achieved with a robot. Another example is strategic and economic analysis; there too the use of algorithms may contribute to the work of prevision, but human intervention will always be necessary in order to check the results of algorithms and to take into account data and events which are not predictable.
Automated advice should not give pace to the possibility for some actors to cope at a good price with the new requirement of investors’ protection regulation as far as ‘clicks’ are much an easier way to get through these requirements than asking questions to the client during a rendezvous.
In order to maintain a good level of investor protection, it will probably be necessary to introduce some kind of qualification and authorization for ‘robot-advisors’ with due diligence requirement that will permit to check the quality of algorithms. In addition, it is important to prevent investors from having the illusion of receiving a financial advice whereas, in fact, what is given to them does not rely on a serious basis; also is it important to protect them from the risk of biased advice which would be provided by unregulated actors acting in connection with some product manufacturers.
The description of automated financial advice tools provided in page 14 is quite relevant but it only refers to data related to the investor. As useful they may be, they will not be sufficient in order to provide good advice. Economic and financial analysis is also necessary, as well as analysis of the intrinsic value of the available instruments. For example:
- if one must achieve an investment during a 14 years period (for example before retirement) perhaps will it be more profitable for him to invest successively in different funds during 5 to 7 years instead of investing in a single fund during14 years;
- the choice between stocks or bonds will have to take into consideration the level of market prices and the level of interest rates, etc.
In addition we observe that, quite often, retail investors have not a precise view of what they wish beyond an increase of their fortune combined with a minimum risk… Therefore, when a default option is proposed – as it is the case with the French PERCO for example – it uses to be adopted by a vast majority of retail investors.
2. Are there any other relevant characteristics of automated financial advice tools?
Some automated financial advice tools allow for more regular information to the investor as well as for the adjustment of an allocation in case the volatility of markets increases, in order to maintain the initial level of risk accepted by the client.3. Are you aware of examples of automated financial advice tools being used in the banking, insurance, and/or securities sectors? Please provide examples, giving details of their operating process.
Yes indeed: in the securities sector, online transactional websites have developed with great success and they often provide some kind of advice online for investors in connection with their personal situation.Apart from that sector, for years Credit Agricole group has also used an algorithmic tool which matches the profile of the client and its financial needs with a great variety of investment products.
4. Do you offer/are you considering offering automated financial advice tools as part of your business model? If so, please briefly describe: i) what type of entity you are, e.g., long established, start-up, a product provider, an intermediary; ii) the service you provide (e.g. to what extent do you integrate human interaction in the tool you provide?); iii) the nature of your clients; iv) your business model; v) who developed the automated tool (i.e. an external company or developed internally?); and vi) the size of your activity and/or forecast activity?
i) what type of entity you are, e.g., long established, start-up, a product provider, an intermediary;See the description of our Group in the executive summary.
ii) the service you provide (e.g. to what extent do you integrate human interaction in the tool you provide?);
We are presently working on an online investment advice service which will provide funds allocations. Each recommended allocation will have to fit with the client profile and with his savings project. Allocations are produced by a system which is based on Harry Markowitz’s theory (efficient frontier, diversification, etc.) which extracts an allocation from a funds’ universe defined for the client with an opened architecture principle i.e. a set of funds internally and externally managed. The algorithm has been framed by a fund selection manager who has performed a fund categorisation before setting the various allocations. Hypothesis underlying these allocations are regularly reviewed by the management team. The client’s profile results from a suitability test. In case answers to the suitability test would not be coherent, an adviser of Amundi would get in touch with the client.
When accepted by the investor, the allocation will be subject to ongoing review and some alerts will be sent when necessary to the client offering the possibility of talking with an adviser.
iii) the nature of your clients;
This service is free and proposed to existing clients.
iv) your business model;
We observed that a large part of our retail clients’ savings are invested in money market funds; through this offer of digital advice we wish them to shift to other funds which may provide better returns.
v) who developed the automated tool (i.e. an external company or developed internally?);
As explained, it has been developed internally.
vi) the size of your activity and/or forecast activity?
This service will probably have a progressive development.
6. Do you consider the potential benefits to consumers to be accurately described? If not, please explain why.
Benefit 1 (§31) :We do not believe that digital advice will lead to any important cost reduction and we do not see this point as a main advantage. In fact, good financial advice requires human intervention at a given point of the distribution process. For years, when e-business began to develop, banks were wondering whether to maintain their branch network. The experience has shown that the combination of a website for payments in line, account consultation etc, with a good service and advice in local branches could optimize client relationship. We do believe that the same thing will occur with automated financial advice. In fact, only a human advisor is able to provide truly personalized advice, evaluating a consumer’s priorities against the complexity of family histories, guide a consumer through unexpected life events, explore the full range of products and choices with a consumer, consider all accounts, products, investments, etc. a consumer holds, and have a dialogue with the retail investor which will be crucial in particular during times of financial panic or euphoria. In addition, people with more savings use to be elder clients often retired, who like coming into the branch.
Benefit 2 (§ 31/32):
We do not share the view that many clients would feel not wealthy enough in order to ask for financial advice; at least it is not the case in France. French Banks have very large networks spread all over the territory and cope with mass market clients who feel comfortable in the branch where they have their account. Nevertheless, digital access may provide an alternative in some rural areas.
Benefits 3 (§34):
In the present state of things and due to fiscal disparities, cross border benefits will be quite thin. In addition, the language is an obstacle, at least in southern Europe. At this stage, it is important to distinguish between pure advice tools and transactional websites providing advice before investing on line. Of course it may be quite easy to provide cross border pure advice, but when coming to transaction, it is much more complex for investors – and even impossible – to use different cross border providers, due to the fact that these providers are not connected with a sufficient number of depositaries. This blocking point could be more easily overcome in the field of individual pension products which are not subject to the same running than funds (no need for securities account). We hope that the PEPP initiative will provide the legal framework to ease the cross border distribution of these plans.
Benefit 4 (§35):
We do believe that this advantage is key, and it will be all the more interesting for clients if they are provided with simulators and calculators that will help them to get some kind of valuation of their future savings according to various scenarios. But we do believe that after this first ‘weekend step’, most of retail clients will prefer to have an exchange of views with their financial adviser and will very often need some additional explanation and precision about some feature of the products and funds.
Benefit 5 (§36 & 37):
Algorithms allow for some kind of standardization of the advice and as explained earlier, most networks use advising tools; the success of Harvest’s Big Expert is an illustration of this and various internal tools have been developed by most banking networks. The alliance of these tools with a good human expertise provides best results, easing the initial phase of suitability and the storage of useful information about the situation and characteristics of each client.
Benefit 6 (§38):
In the context of savings plan, real time information may be detrimental rather than beneficial, inducing short-termism. For example, when facing extreme volatility in financial markets as it is presently the case, real time information would generate unnecessary stress and could lead to very damaging decisions.
Benefit 7 (§39):
Rather than for the client, it will allow the adviser to keep easily more data in relation to the client and its savings in order to provide consistent advice during the whole period of products detention.
7. Are you aware of any additional benefits to consumers? If so, please describe them.
We see three additional benefits:- Digital advice with simulators may introduce some kind of play or leisure in the investment decision, thus facilitating wide spread financial education.
- It may introduce a certain geographical equality between well distributed areas and more desertified regions.
- More autonomous retail clients will find direct access to products with a minimum human intervention at a better price. But nowadays, this is not true for funds as explained supra in our reply to Q6, benefit 3.
8. Do you see any differences in the potential benefits arising for consumers in each of the banking, insurance and securities sectors?
We would be of the opinion that benefits can more easily arise for people investing in securities markets – in fact there are already many transactional websites in this market – and then for insurance. Lending activities do not fit so easily with digital processes.10. Do you consider the potential benefits to financial institutions to be accurately described? If not, please explain why.
Benefit 8 (§40):With respect to what has been said earlier, this benefit is not obvious because it would not be relevant to get rid of human advice; digital advice tools do not exclude the need of human advice or intervention. It can only reduce this need in certain circumstances thus allowing for some kind of thrift.
When looking for the quality of advice, we do believe that digital tools may be more useful when employed by advisers as mentioned in the executive summary, instead of having the client left alone in front of his computer. In fact, the greatest part of retail investors have a long term perspective; in this context the longer the investment will be, the lower the relative cost of initial advice. What most matters is the quality of this advice. Thinking that robots may provide better advice than experienced professionals would be somewhat promethean. This being said, we can admit that digital tools may provide good support for the following up of the achieved investment.
Benefit 10 (§42 & 43):
Amundi does not agree with this benefit. As far as financial is anything but an exact science, we do believe that the potential for differences due to human interpretation is a good thing. The risk of short-termism already pointed must also be thoroughly considered. Last but not least, the experience shows that the origin of each new financial crisis is always different and difficult to predict.
11. Are you aware of any additional benefits to financial institutions? If so, please describe them.
In line with what has just been said, CRM digital tools provide a thorough and exhaustive knowledge of clients avoiding loss of information and fostering the quality of human advice.In another perspective, digital distribution may provide investment firms and asset managers with a direct access to new retail clients that they do not access through external distributors or intermediaries.
13. Have you observed any of these potential benefits to financial institutions? If so, please provide examples and describe the kind of benefit that has accrued.
The success story of Charles Swab’s website in the US is well known. Nevertheless, US securities markets are much more developed than European markets and north American retail investors are used to invest in listed stocks taking much more financial risks than European citizens. In addition, this success is based in a large extent on ETFs which are broadly subscribed by retail investors in the US, which is not the case in Europe.14. Do you agree with the description of the potential risks to consumers identified? If not, explain why.
Yes we agree with the analysis of ESAs which provides additional arguments in favor of a combination of human and digital advice.With respect to §54, we would be somewhat doubtful with regards clients’ needs. In our view, it does not make much sense to identify a saving product aiming precisely at the payment of children’s studies. A lot of fuss has been done around this topic of clients’ needs which is more marketing argument than serious financial approach. With four main criteria, it is possible to fit with specific needs of clients: length of sparing, search for revenue or accumulation, level of risk. It is important not to make the confusion between investment objectives and the intended use of savings.
We also agree with the analysis provided in §59.
For what is of the tailored recommendation, it is important to stress on the fact that financial advice always incorporate a part of assumption linked to the future of financial markets, which is a very uncertain exercise, may it be the result of human analysis or of algorithms.
In France, the risk identified in §66/67 is limited due to the supervision of the CNIL.
Risks of §68/71 linked to unsuitable decisions because of limitations or assumptions or errors within the tool are further arguments in favor of digital tools combined with human advice. In the case of websites providing direct advice to retail investors, the financial expertise of experienced fund managers within the firm operating these online tools is essential.
The risk of haste identified in §74 can only be coped with the help of human intervention.
We consider that the “herding risk” mentioned in §77 would clearly occur in case of widespread automated advice and we also share the concern of a potential extinction of human financial advice in favor of digital linked to new regulatory burden. In fact, it would be much easier for digital operators to comply with all suitability and information requirements introduced by MiFID II and PRIIPs: some rapid clicks will allow the client getting through without reading any KID or other “boring” documents and the investment will be concluded…
15. Do you consider there to be any risks to consumers missing? If so, please explain.
Yes, we believe that on-going engagement could be missing in the context of digital advice. In case of change in the customer’s financial situation, there may be no adequate means to enable the automated tool to get aware of it. The onus would be on the customer to re-engage with the automated tool and re-input the updated information. In the current human advice model, advisers have the duty of keeping in touch with their clients.17. Have you observed any of these risks causing detriment to consumers? If so, in what way?
Automated advice tools are still at an early stage of their development and it is very difficult to already give an answer to this question.18. Do you agree with the description of the potential risks to financial institutions identified? If not, explain why.
In line with all what has been said supra, we do not consider R15 as being a risk: human intervention will always be required for good financial advice.22. Would you agree with the assessment of the potential evolution of automated advice? Please provide your reasoning.
Yes we agree. Firms will develop their digital business but we are of the view that this development will not be equal throughout Europe.This being said, the ESAs should consider how to prevent potentially miss-selling due to automated advice. Some kind of qualification and authorization would probably be necessary for automated advice driven actors in order to provide the good level of retail investors’ protection. It will be important to avoid the use of these tools by unregulated actors who could mislead retail investors as it has been the case with the offer of forex or CFD products.
In addition, it will be important to clearly distinguish between financial digital advice and tools that merely guide investors to make their own investment decisions on a well-informed basis. This topic is somehow treated in MiFID as it may be compared to the distinction between products placement with financial advice and execution only service; it should be taken into consideration when looking for a specific regulation.