ARTIFICIAL INTELLIGENCE NEWSLETTER– AUGUST 2019
Digital advisory services, often colloquially referred to as the “Robo-Advisory” services, has been a topic of interest in the last few years, particularly from the perspective of investor-protection and regulatory outreach.
While there are many definitions and connotations to the term, in our view the most comprehensive is the one proposed by Paolo Sironi.
Sironi goes on to argue that first generation robo-advisors show atleast one of the below if not all facets:
- To be digital businesses, with automation and technology at their core;
- To confirm by and large to passive investment and diversification principles;
- To institutionalize automated portfolios rebalancing and tax optimization;
- To use attractive self-assessment approaches which attempt to personalize investment decisions to individual goals and behavior;
- To display high degree of business focus.
Sironi also suggests that the typical workflow of a first generation robo-advisor would be, as laid down in Table 1 below.
GLOBAL LEGAL STATUS
Holding this workflow in thought, what forms an essential argument is whether any or all steps of this workflow can be regulated. A few regulators across the globe have tried to introduce regulations, frameworks and guidance notes to tackle the issues surrounding the provision of robo-advisory services. We can take quick look at how various jurisdictions have approached this concept.
In the United States a draft of Algorithmic Accountability Act of 2019 has been introduced in the parliament and awaits consensus. The United State Securities and Exchange Commission (SEC) had also issued a guidance update, as early as in February, 2017 to potentially explore the possibility of bringing in robo-advisors under the ambit of its existing Advisers Act. The key areas of consideration of this guidance were:
- The substance and presentation of disclosures to clients about the robo-adviser and the investment advisory services it offers;
- The obligation to obtain information from clients to support the robo-adviser’s duty to provide suitable advice; and
- The adoption and implementation of effective compliance programs reasonably designed to address particular concerns relevant to providing automated advice.
The SEC later also brought two enforcement actions against the firms using robo-advisors. In the famous Wealthfront Advisors case, SEC settled with a tax-loss harvesting programme which claimed to “create tax benefits for clients by selling certain assets at a loss that, if realized, to offset income or gains on other transactions, thereby reducing clients’ tax liability in a given year.” The firm was eventually found guilty of falsely claiming to “monitor” all client accounts, to avoid any transactions that might trigger a wash sale/ tax evasive activities. Another firm, The Hedgeable Inc on the other hand was found guilty by SEC of creating indices that were created not from the actual data but “approximation” of performances of mere 4% of total number of data sets.
But what is interesting is that SEC has carefully picked up on “monitoring” and “approximation”, the two concepts that are not only relevant to robo-advisory, but have also marred the future of Artificial Intelligence at large, quite a few times in the past. For instance, with reference to above table, specifically in terms of cognitive/ automated self-assessment, the risk of approximation by robo-advisors or similar solutions, posses a serious threat to ‘Execution’ of a transaction, hence imbalancing the entire ecosystem.
In July, 2018 the European Securities and Markets Authority (ESMA) published guidelines (“EU Guidelines”) for the suitability requirements under MiFID II which considered the use of robo-advisory services for the first time. The guideline defines robo-advisory to mean a “provision of investment advice or portfolio management services (in whole or in part) through an automated or semi-automated system used as a client-facing tool”. The guidelines are applicable to all the firms providing the services of investment advice and portfolio management services and targets suitability assessment requirements using robo-advisors.
The EU Guidelines target more on the means of suitability assessment, KYC and proportionality. As per the guidelines the entity using robo-advisors is also required to make full disclosure to clients about ratio of human interaction vis-à-vis use of algorithm while providing investment advice or portfolio management services.
ABU DHABI GLOBAL MARKET
On the home turf, in April of this year the financial regulator of the Abu Dhabi Global Market – the Financial Services Regulatory Authority (FSRA) issued a Supplementary Guidance on Authorization of Digital Investment Management Activities or the “Robo-Advisory” to regulate the activities in this realm (“ADGM Guidance”).
The ADGM Guidance, defines digital investment management (“DIMs”) to mean “investment management services using algorithm-based tools that require limited/optional human interaction between clients and the provider of digital investment management services”. It excludes asset management or advisory activities that rely on algorithm-based tools solely for the purposes of back-office support services.
ADGM Guidance also assesses that DIMs may fall within the categories of fully digital, or hybrid models. Further, it exempts the companies operating solely as technology providers from its ambit.
DIMs operating from and within the ADGM are required to obtain a Financial Services Permission (FSP) in order undertake any of the regulated activities related to robo-advisory. Again, FSRA has assessed that DIMs may fall in one of the following categories of regulated activities:
- Advising on investments or credits;
- Arranging deals in investments;
- Managing Assets;
The FSRA although exempts Custodians licensed as such from ADGM, from obtaining an FSP in the event they are found ‘arranging deals’ for the clients.
Prudential Capital Requirements
FSRA has prescribed the capital requirements for where an entity provides a Digital Management Services for each of the abovementioned category of regulated activity. The applicable Capital Requirement is the higher of the Base Capital Requirement and Expenditure Based Capital Minimum. Where a Digital Investment Manager undertakes a combination of activities, the highest prudential category shall be applied. Please refer to below Table for more clarity.
Base Capital Requirement
Expenditure Based Capital Minimum
|Managing Assets||$250,000||–Holding Client Assets: 18/52nds of Annual Audited Expenditure
–Otherwise: 13/52nds of Annual Audited Expenditure
|Advising on Investments or Credit Arranging Deals in Investments Arranging Custody||$10,000||6/52nds of Annual Audited Expenditure|
The FSRA has although provided a relief for the DIMs carrying out the activity of managing assets.
FSRA has recognized that the higher prudential requirements for ‘managing assets’ are owing to the (i) inherent risks associated with the investments wherein the investment managers shall have the discretion to make investment decisions for the clients; (ii) the operational and logistical complexities of holding client assets. It also recognizes that, the discretion of such an investment manager shall ultimately lead to ‘Rebalancing’ of clients’ portfolio which may also lead to volatility and higher trading volumes.
Although, in the event such DIMs are advising and working on business models wherein:
(i) product offering is limited to liquid, non-complex and diversified passive investments; (ii) the discretion is limited to portfolio rebalancing- which does not include purchase of new investment products not included in the portfolio initially intended by the client; or (iii) the investment manager does not hold client’s assets; then, the prudential requirements may be lowered to the higher of $10,000 or 6/52 of Annual Audited Expenditure.
Algorithm and Technology Governance
As in the past, FSRA has yet again delved into the nitty-gritties of the technology- this time by laying down principles for algorithm governance by the DIMs. In accordance with the existing rules on management, system and controls, and other core principles- FSRA requires the DIMs to, ensure, at all times (i) Qualifications and competency of the staff supervising the functioning of algorithm (ii) adequacy, explainability, and traceability of documentation on data and assumptions on which the algorithm is based. (iii) efficient management and maintenance of the algorithm and model; (iv) continuous monitoring and reviews of the model.
FSRA also provides additional guidelines the requirements for technology in relation to treatment of information and contingencies on disruption.
Similar to requirements under MiFID II, FSRA pursuant to Rule 3.4.2. of its Conduct of Business Rules requires the DIMs to undertake suitability assessment to assess whether an investment recommended or executed on a discretionary basis is suitable for the client.
To ensure the same the ADGM Guidelines lists down various benchmark for such assessments by way of ‘Risk Assessment Questionnaires’ which shall contain questions for assessing the proportionality of the said investment/recommendation with client’s financial appetite, and intention and objective as mentioned in the portfolio. What is pertinent to note is that the ADGM Guideline, makes a provision for ‘know-out’ questions, which shall help DIMs in rejecting the prospective clients whose investment horizon, liquidity needs or other circumstances are misaligned with the investments offered through the technology.
ADGM Guidelines also mandates that adequate disclosures be made to the client about the products offered through the specific model and the degree of human interaction and supervision involved in the process of investment and rebalancing.
The effort to regulate a robo-advisory services by various regulators around the world is definitely a laudable effort. Having said that most regulators have currently tried to fit in the robo-advisors under the ambit of their existing regulation. However, Robo-advisors may not always be paralleled with human advisors. As we have noted in our previous Newsletter the dynamics of placing of legal liability in terms of absolute and vicarious liability may be very different when it comes to an algorithm.
The way the regulatory landscape around use of robo-advisors has evolved, the dynamics around investor protection and investor rights may also need a revisit. The principle of Caveat Emptor may have to be looked at with a fresh perspective, as today’s investors are possibly more aware than what they were a century before the advent of AI.
Having said that, till there is a clarity on resting of the legal liability in case of an AI/algorithm (since it is not a legal person, as yet), any steps towards balancing emerging tech vis-a-vis investor protection would be a welcome move!
 Sironi identifies ‘Robo-Advisors’ as “automated investment solutions which engage individuals with digital tools featuring advanced customer experience behavior towards rudimentary goal-based decision making, conveniently supported by portfolio rebalancing techniques using trading algorithms based on passive investments and diversification strategies. These digital businesses differentiate by degree of passive management, depth of investment automation, interaction between human advisors, and level of self-assessment, as well as target clientele.” Sironi, Paolo. Fintech Innovation . Wiley , 8.
 The principle that the buyer alone is responsible for checking the quality and suitability of goods before a purchase is made.
Authored by Akshata Namjoshi (Senior Associate).