Lessons for roboadvice in Centrelink debacle

Share

‘Robo’ has rapidly become the prefix of choice for anything to do with automation or robotics. The biggest local political story of the Australian summer was the use of Centrelink’s computers to chase overpayments to welfare recipients. It was a ‘robo-debt’ campaign, featuring ‘robo-assistants’ to answer questions. An estimated 169,000 robo-debt letters had been sent by January 2017, with over a million more expected in coming years. It was a political fire-storm with thousands of the least wealthy Australians receiving debt notices in the days around Christmas.

The problem was not the attempt to curtail the burgeoning welfare budget. That is a necessary step at a time when 11% of the population draws a full or part age pension, and increasing numbers will continue to do so for decades to come. The problem was the method. The Centrelink computers were sending out 20,000 debt letters a week, the same number previously sent out each year.

Computers replaced humans

Centrelink usually relies on self-reporting for welfare claims, with clients maintaining their own records on the MyGov website with checks at personal meetings in Centrelink offices. If a client found some temporary part-time or casual work, then nothing for a few months, this could be explained across the desk and adjustments made accordingly. By taking account of the wider set of circumstances, verified personally by humans, there were fewer reported breaches.

Centrelink offices are crowded places at the best of times, with long waiting lines where clients take a number as if they are buying bacon pieces from Woolworths. They wait an hour or two to be called, even if they have an appointment, and sit in front of an officer who is obliged to ask a wide range of questions. What have you done to find full-time employment? Have you looked for voluntary work? How many hours of part-time work have you done? Let’s update your assets and liabilities, and have you gifted any money to anyone? The client then signs the form as a legal document, verifying everything is accurate.

It is much like a financial adviser collecting data for a Statement of Advice.

The downside of this personal, more equitable approach is the massive cost of Centrelink officers sitting across a desk chatting to each person. Instead, in the robo-debt campaign, Centrelink trawled through the income records of its clients at the Australian Taxation Office (ATO). The government argues it is the only way to claw back overpayments in the age pension, budgeted at $1.1 billion. Amid the claims and counterclaims, it’s likely that a minority of the letter recipients owe a debt.

The relevance to roboadvice

Roboadvice comes in many shapes and sizes, and we already have robo 1.0, 2.0 and 3.0 as new developments occur every week.

In the enhanced versions, call them 2.0 and 3.0, robo is an adjunct to a full-service offering, a more efficient way to collect data and record preferences as part of a complete face-to-face relationship. Elsewhere, digital services advise clients on a range of investments available to meet their risk tolerance and goals in a more cost-effective way, offering market commentary and delivering alternatives to a wider audience than traditional advice affords. A human is available for a chat.

In the US, the leading robo-advice providers have recognised this need as they move beyond the basic robo. According to FinancialAdviserIQ (a Financial Times service):

“New York-based Betterment, an early pioneer in online investment advice, says it’s joining a growing number of fund providers and independent wealth managers who are turning to ‘real-life’ advisors to help manage client accounts.

On Tuesday Betterment launched two new services. Its ‘plus’ offering charges 0.4% and lets clients of its basic ‘digital’ platform make one call a year with a human advisor. The all-robo option’s fees are being streamlined to 0.25% a year.

Betterment also says it’s now going to provide a ‘premium’ service for 0.5% where clients can call human advisors on an unlimited basis during weekdays. To boot, the company says each advisor will be fully licensed and expected to hold CFP designations.”

What about robo 1.0?

However, the version most people identify as ‘roboadvice’, the fully ‘automated or robotic’, remains robo 1.0, a relatively simple end-to-end investment sales platform. These basic robo models ask a few questions about risk, income and assets, and recommend a simple, ETF-based portfolio.

This simple robo approach is most exposed to the Centrelink comparison. They use an algorithm to select a suitable portfolio, but it’s based on a cursory glance at the overall circumstances of the investor. The process leads to a product sale without knowing if it is suitable for the long-term goals of the client which comes from a detailed discussion of personal circumstances.

In the Centrelink analogy, the algorithms look at a small part of the overall picture, creating a potential for incorrect assessment. The automation delivers snapshot ‘advice’ to the masses who cannot afford to see a full-service planner.

As Wade Matterson of leading consultants, Milliman, said in late 2016:

“Many automated advice providers are simply replicating the increasingly outdated traditional advice process, which places an investor’s risk tolerance at its apex and delivers product-led solutions … a more nuanced approach to risk profiling will include different components such as risk aversion (the flip side of risk tolerance), risk capacity (the financial ability to endure losses) and risk need (the amount of risk needed to likely achieve goals).”

Examples of potential robo shortcomings

Consider some examples of the shortcomings of many roboadvice processes:

1. Mortgage versus super

The basic robo models do not deliver financial planning, but rather online investment selection. Most cannot answer basic ‘advice’ questions. Advice does not start with which ETF to invest in, but questions such as:

“Can you tell me whether I should pay off my mortgage or invest the money into super?”

2. Coverage of the full picture

Imagine you meet a financial adviser and ask how you should invest $20,000. You tell her that you have total investible assets of $2 million, you own your home, you expect to retire in five years and you are reasonably risk tolerant. What type of portfolio should be assigned to the $20,000?

Based only on this information, it does not matter. Without knowing how the other $1.98 million is invested, or more about long-term goals and income needs, who cares how $0.02 million is allocated? The human adviser should address the preferred outcome for the entire portfolio.

Most of the robo 1.0 models only ask a few questions about risk tolerance to determine how much of the $20,000 to allocate to riskier assets (equities, property) versus defensives (term deposits, bonds), and then recommend a portfolio. There is no recognition that it might be allocating even more to an asset class that is already overweight in the rest of the portfolio.

4. Imbalance in Australian indexes

A major selling point of most online investment products is the low cost (although this often disguises all the expenses). To meet a price point, the portfolios usually consist of index ETFs. But there is a problem in Australian indexes.

The local market is heavily weighted to a few companies, mainly banks. In fact, the Financials Index excluding property trusts (ASX:XXJ) comprises over half of the S&P/ASX200, with the big four banks making up 30%. Some ‘value’ or ‘dividend’ ETFs have 75% allocated to financials. Not only does performance depends on one highly leveraged sector, but it’s likely that investors already hold the major banks directly in the rest of their portfolio.

A warning not a failing

In time, improvements in technologies such as artificial intelligence and data mining will take roboadvice to another level. The robo 3.0 models already provide human planners with tools which show clients the outcomes of various choices, and the human element remains important.

Often, the adviser is as much a coach and educator as a financial planner. It will be a while before the online roboadviser shows that level of empathy, as Centrelink is finding as it rolls out its robo-debt and robo-assistants.

 

Graham Hand is Managing Editor of Cuffelinks.

Graham will be speaking at the Australian Shareholders’ Association’s 2017 Securing Your Investing Future Conference in Melbourne on 15 May with his session: Is there an Uber awaiting wealth management?

Share
Print Friendly, PDF & Email

, , ,

One Response to Lessons for roboadvice in Centrelink debacle

  1. Gary M February 9, 2017 at 11:56 AM #

    Advice has nothing to do with it. Robo sites are sales tools to sell product. Nothing more, nothing less.

    It is only ‘advice’ in the sense that if you walk into a Ford dealer to buy a mid-sized Ford sedan (or whatever) which you’ve already decided you wanted to buy, the only ‘advice’ you want and get is “what colour would you like?”.

Leave a Comment:

*

Register for our free weekly newsletter

New registrations receive a free copy of our ebook, Cuffelinks Showcase 2016.