The reality of roboadvice

Share

The impact of digital innovation has been felt in many parts of the global economy and the financial planning sector is not immune. Recent turbulence in the advice industry, coupled with an increasingly engaged and digitally-aware public, has created the perfect environment for digital disruption. Technology has a key role to play in improving the availability and consistency of financial advice and one area in particular that has been receiving a lot of interest is the use of roboadvice.

While the term robo-advisor could be taken to imply a robot or algorithmic digital tool designed to perform all of the tasks of an adviser, Australia has yet to see a roboadvice tool which comes close to offering the full services of a traditional financial adviser. The scope and sophistication of financial planning software and online calculators is increasing, but they are still essentially support tools or algorithms. If we look to the US, where the phrase robo-advisor was coined and where they have had the most success, we see that the majority of these services focus on portfolio construction and rebalancing. However, this activity is a very narrow subset of what is usually referred to in Australia as advice – that is, assessing an individual’s financial position and proposing holistic strategies to improve that position – so we are a long way from replacing human advisers with machines.

Benefits and limitations of the robo-advice model

Complex advice algorithms have many benefits, but they also have their limitations. To illustrate the challenges in bridging the gap between algorithmic roboadvice tools and the more holistic work of a financial adviser, we have explored a typical, seemingly simple, advice scenario.

Suppose an advice client currently has a mortgage on their family home and is considering the following two options:

  • Making prepayments on the mortgage to pay if off sooner; or
  • Entering into a salary sacrifice arrangement to build up their superannuation savings.

To provide insight into the level of complexity required to deal with what appears to be a relatively simple question, we illustrate the results of our algorithmic calculations in the figure below.

SN-AS-EY-Pic1-161015Note: This chart was constructed using many different assumptions about various client types and economic assumptions and should not be relied upon for financial advice. It is provided for illustration only. For example, we have assumed that client age is a proxy for remaining mortgage term.

As the chart shows, there is no one-size-fits-all solution. For many people – such as those who are closer to paying off their mortgage and are on higher marginal tax rates – paying off their mortgage with free cash flow may not be the optimal strategy. They could stand to be in a better financial position at retirement if they were to salary sacrifice this free cash flow instead. Conversely, younger individuals on lower marginal tax rates may be better off financially if they elect to prioritise paying off their mortgage.

Not a trivial calculation

Even by restricting our analysis purely to the objective of maximising net wealth at retirement, arriving at the best solution for the client is not a trivial task and requires the exploration of multiple scenarios to arrive at an appropriate strategy. While this can be achieved with commonly available planning tools, it is a time-consuming process, especially if a high degree of accuracy and consistency is required. This provides an opportunity for ‘next-generation’ algorithmic tools that can perform the mechanical operations quickly and in a way that gives the adviser confidence in the accuracy and consistency of the results. The adviser would then be able to generate a reliable strategy and talk the client through it in one sitting.

However quick and accurate they may be, algorithms on their own are not enough as there are many variables that must be addressed, some of which are subjective. For example, the algorithm used to generate the output above does not capture liquidity preferences, the risk of breaks in employment, possible changes in salary, bequest motives or other sources of uncertainty, such as a potential spike in interest rates. While some of these considerations can be addressed by developing smarter algorithms, others require higher level thinking. An example of this would be factoring in an individual’s preference to reduce their leverage as quickly as possible, to achieve greater peace of mind. This is more than a numerical optimisation exercise; it requires human-like intelligence.

Who will be the winners?

What does this mean for the future of robo-advisors? We expect to see the development of greatly enhanced algorithmic tools to support advisers, with benefits including:

  • Speed and efficiency of advice
  • Reduced cost to serve and increased proportion of the population serviced by the advice industry
  • Increased consistency of advice and the potential to enhance documentation and record keeping
  • The retention of advice data in readily-accessible digital formats to assist with compliance functions, client engagement and trend identification.

As for the term roboadvice, while great for headlines, it is a little unhelpful when it comes to understanding the reality of the advantages that automated algorithms can bring to the advice industry.

We are still many years away from robo-advisors having sufficient artificial intelligence to replace financial advisers. However advisers do need to acknowledge that they are part of a rapidly changing industry which is adopting algorithmic tools of increasing sophistication. This is both a great opportunity, as well as a threat to those unable to adapt quickly. Early movers who take advantage of these advances in technology will attract more clients, increase productivity, drive down costs and serve previously unadvised segments of the market.

As with many technological advances, the ultimate winners are likely to be the end consumers. With such a large portion of the population currently unadvised, and no let-up in the complexity of our financial system, this can only be a good thing.

 

Steven Nagle is a partner in EY’s Oceania financial services practice. Anthony Saliba is a manager in EY’s Oceania actuarial services practice.

The views expressed in this article are the views of the authors, not Ernst & Young. The article provides general information, does not constitute advice and should not be relied on as such. Professional advice should be sought prior to any action being taken in reliance on any of the information. Liability limited by a scheme approved under Professional Standards Legislation.

Share
Print Friendly, PDF & Email

, , ,

19 Responses to The reality of roboadvice

  1. Peter Pontikis October 23, 2015 at 8:52 PM #

    So it seems its multi-dimensional experiential challenge. Complementary at their best. all distrusted at their worst and lots of scenario’s in between. And as with “choice” theory, I suspect people will apply Occam’s razor even if as a heuristic bias would yield sub-optimal choices. all that doesn’t mean we should try to do better. Good comments regardless – thank you All!

  2. Paolo Sironi October 23, 2015 at 8:51 PM #

    I believe the key element is to provide a consistent graphical experience about investment decision-making. Doesn’t matter if process is fully automated, as long as the problem is unbundled into simpler statements such a what-if, that easily maps the reasoning of a human investors. In a Robo-4-Advisor case the investment experience and the digital experience overlap.

  3. Greg Einfeld October 23, 2015 at 8:51 PM #

    Yes Peter that is right. Both humans and computers have their respective advantages. What Robo advice brings to the table is additional choice which has got to be a good thing.

  4. Peter Pontikis October 23, 2015 at 8:50 PM #

    In the client’s eyes – it comes down to trust doesn’t it? Trust the human “advisor” or the “Algorithm”.

  5. Paolo Sironi October 23, 2015 at 8:49 PM #

    Thanks for the insights. You might also appreciate our “probabilistic scenario proposition” about ROBO-RETIREMENT which we are showcasing using IBM Risk Analytics. It helps creating intuitive what if analysis based on cumulation / de-cumulation strategies.
    See here recent LinkedIn post:
    https://www.linkedin.com/pulse/dynamic-retirement-planning-what-robo-advisors-wealth-paolo-sironi

  6. Amy Ciolek October 21, 2015 at 9:38 PM #

    The consistent outcome of robo is comforting. The assumptions just need to be correct as in any volume system with limited human intervention. Good article. Thanks

  7. Charles Moore October 21, 2015 at 9:37 PM #

    If your answers to ASIC are like in a separate thread (basics not addressed at all), then the result will be like most “marketing” focused activities..
    My concern is the “overselling” what computers can actually do, and the current scope of the Robo “Advice”, but if it is actually free and does not have any human attached ( like the Sumsung advice I finally managed to disable) then probably don’t care..
    But if it is being sold as actual “Advice tailored to meet the specific requirements of an individual with full KYC”, then still technically waffle.. no amount of marketing will change this fact..
    I have no vested interest, in robo advice, but as a technologist need to ensure this stuff is “factual” not just marketing waffle with no basis…
    PS: ASIC typically investigate “risk: related products or services..

  8. Greg Einfeld October 21, 2015 at 9:36 PM #

    Like with everything in this world, ASIC might not get it 100% right (whatever right means) 100% of the time. But they are certainly trying and from what I have seen they are making good progress. e.g. they have set up a Digital Finance Advisory Committee and an Innovation Hub Taskforce. I met with them recently and I was impressed with the level of engagement, the quality of their questions and their insights. It is up to us as industry thought leaders to help ASIC come up to speed.

  9. Charles Moore October 21, 2015 at 9:35 PM #

    @greg we still disagree..
    ASIC is a slow moving entity.. and has dropped the ball on numerous occasions..

  10. Greg Einfeld October 21, 2015 at 9:34 PM #

    Thanks Angus for the UK perspective. In Australia the financial advice regulator, ASIC, is engaging actively with robo-advice businesses to ensure the robo advisers are delivering appropriate advice.

  11. Angus Goldie October 21, 2015 at 9:33 PM #

    In the UK one of the things to test is what the ombudsman would make of a robo advisor case it’s the ultimate test of customer outcome.

  12. Himanshu Dhawan October 18, 2015 at 10:40 AM #

    Definitely yes, a robo can do this. Just a matter of using rite data at rite time.

  13. Bernard Lunn October 18, 2015 at 10:40 AM #

    In the real world Robos target the Unadvised (assets too small to interest Advisers) so it is Robo or nothing. As Robos move up the wealth curve into Adviser territory the question is whether the certain loss of capital from higher AUM fees is compensated for by the uncertain possibility of higher returns. Bogle and Buffet say no and they seem like smart guys.

  14. Anon (name changed by Editor) October 17, 2015 at 2:12 PM #

    Agree, great points Greg.
    On 2 occassions we have not had any clearly explained modelling from our fee for service adviser who uses one of the established adviser planning tools – just a lot of unexplained and unchecked data modelling our pension models so we have lost confidence in the planning tool, whoever they are.

  15. Steve Nagle October 16, 2015 at 6:03 PM #

    Greg,

    Maybe we are just using language differently. Indeed we used a complex algorithm to determine the results shown in our chart. It did, as you suggest it might, compare financial outcomes achieved by different patterns of salary sacrifice versus paying off the mortgage, and chose the optimal one for different cases. However we’re calling that an “algorithmic tool”, not a robo adviser. The point in drawing this distinction is to explore how far “roboadvisers” have got in replacing human advisers. We have developed many such algorithms and are firm believers in their value. However a process does not in my lexicon meet the definition of advice until it uses an understanding of the fuller financial objectives of a client to recommend a course of action to meet these objectives. One day this will happen, and computers will also pass the Turing test at high percentages, but we are a few years off this in roboadvice.

    We are supportive of how such tools with increasing sophistication improve the advice process by performing the mechanical operations quickly and at a low cost. Some of the advantages, which we listed, are:
    >Speed and efficiency of advice
    >Reduced cost to serve and increased proportion of the population serviced by the advice industry
    >Increased consistency of advice and the potential to enhance documentation and record keeping
    >The retention of advice data in readily-accessible digital formats to assist with compliance functions, client engagement and trend identification.

    You asked the question “which would a client prefer”. If all they want to do is answer the simple question we used as an illustration, and they are confident that there is no other matter they need to consider, then many would be happy with the result of this algorithm. Before long I suspect they will be able to get this sort of information for free, making them extra happy. Others may need, or will not be certain that they do not need, more holistic advice. They can have both via a human adviser using the same algorithmic tool to provide part of the answer quickly and at low cost. That’s why we believe the “winners are likely to be the end consumer”.

  16. Paul G October 16, 2015 at 11:00 AM #

    Great points Greg.
    I am sure a sophisticated roboadvice model will enable assumptions and inputs to be varied by the user. A list of pros/ cons/things to consider/risks could also be generated.

    I suspect roboadvice isn’t just a threat to advisers who use inappropriate rules of thumb. Advisers who perform the complex calculations through software such as xplan, coin or midwinter are also under threat. I would expect in time that these complex planning calculators will be available directly to consumers at little or no cost. No longer will there be information asymmetry whereby advisers have the knowledge and access to complex planning calculators and the general public do not.
    I would expect the role of adviser to change from the ‘source of knowledge and strategies’, to more of a sounding board and 2nd opinion for increasingly educated clients.

  17. Greg Einfeld October 16, 2015 at 9:21 AM #

    As I understand it, the authors are saying:
    1. For some people it is better to pay off your mortgage first, whereas for others it is better to salary sacrifice.
    2. Therefore there is no rule that satisfies all clients.
    3. Therefore a computer can’t be programmed to make this decision.

    I would agree that it would be inappropriate for a computer to have a simple rule that applies across all clients. But computers can be far smarter than that if they are programmed correctly. This type of question is a perfect question for a robo-adviser to answer. And there are numerous ways to tackle the problem.

    One method is to say:
    1. What would my financial position look like if I salary sacrifice
    2. What would my financial position look like if I pay off my mortgage
    3. Which is better?

    Of course there is more to it – we need to consider changes in income, changes in expenditure, investment earnings, tax etc. We also need to recognise that it isn’t black and white. It isn’t a choice between 100% salary sacrificing or 100% paying off the mortgage. In some situations the best option might be 50/50. Or 80/20.

    The concept of “better” is important too. Better doesn’t necessarily mean higher wealth. It should take into account other factors specific to the individual such as their goals. Salary sacrifice might result in higher wealth. But that isn’t helpful if they are saving to buy a bigger house.

    A computer is well placed to consider all these factors if it is programmed correctly.

    The authors have taken a different yet equally valid approach. Ironically, they have performed a calculation (no doubt using a computer) that could be coded and used by a robo-adviser.

    Let’s consider this in a different manner. Suppose a client goes to a human adviser and asks the question “should I salary sacrifice or repay the mortgage?”. What response do they usually get? In my experience some advisers will always say “salary sacrifice”. Others will always say “pay off the mortgage”. Some will have a rule of thumb e.g. “if you are under 40 then pay off the mortgage, otherwise salary sacrifice”. Very few will pull out the spreadsheet and perform the complex calculations that sit behind the chart above.

    So if I was a client, what would I prefer? The human who uses an inappropriate rule of thumb, or the computer that analyses the problem correctly? In this case the computer wins hands down.

  18. Matt October 16, 2015 at 6:18 AM #

    It is exactly the human failings of financial advisers that roboadvice will be successful in addressing. As we know from the results of ASIC’s regular shadow shopping exercises, the level of inconsistency in the quality of human provided advice makes roboadvice the way of the future.

    Roboadvice also has the added advantage that unless the model / algorithm / formula is deliberately gamed, you will eliminate the human bias of placing the clients interests after the financial advisers own interests.

  19. Alex October 15, 2015 at 11:36 PM #

    There are some very skilled planners in specific areas but I would want to pay them like a plumber – ie fee for service, not a series of obscure asset based fees for the rest of my life

    So roboadvisor sites could potentially do a very tiny bit of a planner’s role a lot better and a lot cheaper – not replace planners.

Leave a Comment:

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Register for our free weekly newsletter

New registrations receive free copies of our special investment ebooks.