Regtech Financial Advice Files Demonstration and Symposium

Event overview (Sydney 22 August 2019)

ASIC Deputy Chair Daniel Crennan QC giving the commencement address at the Financial Advice Files Demonstration And Symposium

Problem statement and event design

Australia’s financial advice industry has been under considerable scrutiny following the Royal Commission into Misconduct in the Banking, Superannuation and Financial Services Industry (Financial Services Royal Commission). This has led to an increased focus by ASIC and licensees on how licensees can better monitor and supervise their representatives in the provision of personal financial advice.

To test whether an adviser has provided compliant advice to their client, a review of the client file is required. To date, this has been a manual, resource-intensive task. This impacts the number of reviews a licensee can undertake and the timeliness of those reviews.

This event set out to investigate whether regtech can provide licensees and regulators with a better solution so that reviews could be done in a more efficient and effective manner. With this in mind, the purposes of the problem statement were to:

explore how regtech could be used by licensees to make file reviews more efficient and accurate
identify opportunities and limitations for licensees associated with using regtech to assist in file reviews
highlight regtech tools for licensees currently available or in development
demonstrate how regtech tools may also be able to help ASIC in day-to-day reviews.

ASIC selected six regtech firms from a set of applicants — Advice Regtech, Flexprod, IBM Research, IRESS, K&L Gates, and TIQK — to demonstrate their solutions to the problem at a symposium on 22 August 2019. These demonstrators were selected from a set of applicants based on the strength of their solutions.

ASIC regtech FA files dataset

To assist the regtech firms in demonstrating their solutions, ASIC provided a dataset of 20 synthetic financial advice client files that contained approximately 60 documents in total, across all the files (including file notes from client meetings, a ‘fact-find’ document, and a Statement of Advice (SOA) document) across varying formats.

For the purpose of this exercise, demonstrators were advised that they should consider the legal requirements that apply to financial advice products and services as detailed in relevant legislation and regulatory guidance. Additionally, demonstrators were provided with some key risk indicators, which (to the extent that they were identified in the synthetic client files) might indicate non-compliant or high-risk advice.

Problem-specific findings

Many of the demonstrators focused on processing larger volumes of client files aiming for the highest possible levels of accuracy.

Using combinations of natural language processing (NLP), rules-based expert systems, proprietary artificial intelligence (AI), and application programming interface (API), demonstrators were able to identify a range of potential compliance issues in the financial advice files provided by ASIC. Whilst the results provided by the demonstrators could not be verified as part of the limited nature of the trial, the presentations were promising.

Some demonstrators went beyond ASIC’s provided dataset to include hundreds of additional real SOAs in their platforms. These demonstrators illustrated on an aggregate basis how their applications attempt to identify and assess potential misconduct from large numbers of financial advice files.

The demonstrators had a limited time to work with the dataset. This meant some demonstrators found it difficult to break-down Portable Document Format (PDF) documents contained within the dataset, and extrapolate data contained within tables and images. The difficulty of processing this unstructured data was exacerbated by varying file formats and terminology.

Some demonstrators were able to rate the level of potential risk to clients contained in various advice documents. When combined with real-time or near real-time tracking, this would enable them to capture, flag and investigate potential misconduct in financial advice prior to it being provided to a client, or within hours of being sent to a client.

General observations

Monitoring accountability more efficiently
Regtech can help reduce compliance costs and provide an initial review of financial advice. However, responsibility remains with advice providers and advice licensees for the outcomes of these models.
Regtech can help rebuild trust and confidence in the financial advice sector
Regtech has potential to help the financial advice, and wider financial services industry, regain consumers’ confidence by building automated checking processes into the provision of financial advice and other services.

Challenges faced

Complex solutions require higher data volumes
Access to larger datasets would assist with ML in identifying risk factors, vastly improving results.
Standardisation of datasets
Some observers advocated for industry and regulators to introduce standardised models, glossaries, and/or practices for financial advice file documents to assist in the implementation of regtech tools.

Live Stream and Presentations

Agenda

Introduction by Mark Adams, Senior Executive Leader, Strategic Intelligence and Co-Ordinator, Innovation Hub, ASIC (26:48)
Welcome to Country by Aunty Ann Weldon, Metropolitan Local Aboriginal Land Council (31:40)
Commencement Address by Commissioner Daniel Crennan QC, Deputy Chair, ASIC (41:48)
Event Overview by Kate Metz, Senior Executive Leader, Financial Advisers, ASIC (50:41)
Demonstrations (56:55)
Click the providers' names below to view their presentation slides:

Commentator Observations & Recap of Demonstrations (3:47:38)
Panel Discussion covering "Regtech - Experience and Potential" (4:14:10)
Closing by Mark Adams, ASIC (5:14:26)

For more information about ASIC's regtech series, follow the links below:

ASIC regtech initiative series 2018-19

ASIC