I am with French when it comes to robo-advisors

“Do the Brits continue to be ill-informed?”

As a standalone statement it is definitely very political. But bear with me and I will explain myself.

For this post, I looked around for an image for Systemic Risk. My inclination comes from being a graphic novel aficionado and having flirted with the idea of graphic storytelling while living in the birthplace of Marvel and Drawn and Quarterly.

Should I have picked explosives, or an earthquake, or simply the year 2008? None of these capture the essential element of “Systemic Risk”, which is an event whose consequences will require Central level intervention because the stakes are too high to leave it to the market forces.

Even though it has been 10yrs since the last major systemic event, the memory hasn’t dissipated especially behind the walls of Central banks, Regulators, Treasuries and the such.

Brits behind the walls sniffing around for signs of systemic risks

Jan 2016

Lord Alain Turner spiked Twitter impressions with his media comments in early 2016 around the systemic risks of the P2P lending space. As is customary in the media, his remarks were singled out.

“The losses which will emerge from peer-to-peer lending over the next five to 10 years will make the bankers look like lending geniuses,” said Turner on a BBC program.

Lord Turner, ex-chairman of the FSA, reversed his thinking before year end at the Fall London Lendit conference that I personally attended. He said that MPLs (market place lenders) may prove to be the spare tire in the next crisis. He claimed that peer-to-peer lending platforms could actually help reduce the risk of a future financial crisis. I agree with him in that the MPL sector is more than a spare tire already, in terms of mitigating the “Too big to Fail” risk that was the nightmare of the 2008 financial crisis. I also agree with Lord Turner that the multiple and complex risks of the broad, predominantly over-the-counter lending market are very similar within the “Castle” of the regulated financial system and within the “Fintech valley” of the self-regulated (partly) Altfi scattered ecosystem. From cyber security issues, to due diligence, to legal structuring, etc; these are structural issues that are inherent no matter what the delivery process is.

The conversation around this topic is live on the Fintech Genome and insights are welcome.

If one is sniffing around for the exact same kind of 2008 systemic risk (i.e. too big to fail) then I say that you wont find out coming out of the MPL space. Of course, black swans include those risks that we haven’t thought of.

Jan 2017

Carney, the governor of the Bank of England, showcased his concern around robo-advisors posing systemic risk to the financial system at a recent G20 conference meeting.

“Robo-advice and risk management algorithms may lead to excess volatility or increase pro-cyclicality as a result of herding, particularly if the underlying algorithms are overly-sensitive to price movements or highly correlated,” Carney told the conference.

Scalable Capital UK chief Adam French, decided to publicly defend the sector through an open letter which you can read on their site (instead of media excerpts).

I agree with French that for now, the money managed by robos globally (standalone and from incumbents) are clearly less than1% of managed assets. Honestly, the Customer Acquisition Cost (CAC) has proven to be very high and the growth rates of standalone robo-advisors are nowhere close to showing signs of overtaking the asset management industry. Actually, what is growing faster is the leapfrogging of incumbents into the robo space. Brokerage houses are the main adopters simply because their business is experiencing a type of second world war threat (first WW was the online brokerage digitization wave); followed by incumbents like Vanguard, Blackrock, ING, etc that are adopting variations of the business model parallel to their business as usual.

Current evidence shows, that

The majority of robo-advised AUM is by incumbents not by standalone robos. So, leave those kids alone.

Why voice concern for assets managed through robos, since the processes used don’t differ from those used by conventional practices?

  • Algorithms used are no different than those used by conventional managers or financial advisors. Mostly MPT based and in some case risk attribution (like Scalable Capital).
  • The ingredients are basic and liquid and used by conventional asset managers and financial advisors.

Would or should there be an central concern, if we woke up one day and realized that flows of funds shifted into actively managed vehicles that were using predominately passive low cost financial structures? I don’t think so.

Such kind of “herding” has been inherent in the system and more so, as the asset management business is consolidating and we live in a world that a few Blackrocks, Vanguard, etc can survive. Independent asset managers of medium size are struggling to stay afloat as standalone businesses. Did you not notice the recent move from Calamos (NASDAQ: CLMS) one of the few remaining public independent asset managers, that has been taken private?

Shouldn’t the attention be on those entities managing the majority of AUM through basic MPT asset allocation frameworks?

Look for concentrations within incumbents not at standalone robos. So, leave those kids alone.

I’ve always thought that robos could and should differ from the conventional way of servicing smaller amounts of wealth, in two main ways:

Batch One:

(a) Reduced Cost and transparency,

(b) Risk-adjusted performance transparency,

(d) fiduciary duties

Batch Two:

(a) Goal-based investing

Reduced and transparent costs have been the marketing strength of the robo-advisory business model. This is of no concern to central authorities, on the contrary.

The fiduciary responsibility or lack thereof, is of course the main concern. To put this in very simplistic terms, as long as you leave your money to be managed (i.e. not withdraw) the robo-advisory business model entail less incentive conflicts (i.e. a financial advisor selling a fund etc).

It is a simple algorithm (asset allocation – e.g. x% equities, y% fixed income, z% commodities etc) or a more sophisticated one (risk attribution and maybe coupled with momentum analysis; tax loss harvesting may also be incorporated). The only human or emotional decision, which is no different in the digitized asset management delivery or the conventional, is to liquidate part of whole of the portfolio or to Top Up.

As the robo-advisory business model is moving to a hybrid (auto asset allocation + Call a human for an extra cost), I really don’t see the difference in terms of the incentives and the fiduciary duty.

Central authorities should start thinking of mass-market, true AI-driven, asset management. Now that is a tricky one. It has been around for the elite, through the Bridgewater likes. What if technology makes it a viable business model for the mass? How can we think of Deep-Learning (a specific AI application) being used for managing our 10k or 50k savings? The fiduciary responsibility lies with whom?

Add to your radar screen Deep learning applications to asset management, as a future source of concern. In the meantime, please leave these kids alone.

If you want to understand more on AI, ML, Deep learning, listen to this primer because “Software is eating the world”.

Robo-advisory businesses will evolve and grow up and in some ways look more like their old-fashioned ancestors. My concern is that they have not taken the next step in terms of Transparency. They started with the Cost Transparency and reduction campaign. But the bottom line of the financial business is to create wealth. This means offering performance.

Lowering costs of execution, advise, custody, reporting etc is the first step. This is actually low-hanging fruit in 2017. Now, where is the performance? Where is the Risk-adjusted transparent performance?

Why are robo-advisors not reporting online, transparent risk adjusted Actual performance? Simply moving model portfolios online, isn’t an innovation. Simply measuring internally, performance and risk isn’t an innovation.

Contributing actual performance statistical data towards the development of some kind of Robo-Index, that can be used for internal improvement and for offering transparency to end-customers; is the way to go.

Are you a robo-advisor managing assets and want to know more? Drop us a note in the comments below.

Daily Fintech Advisers provides strategic consulting to organizations with business and investment interests in Fintech & operates the Fintech Genome P2P Knowledge Network. Efi Pylarinou is a Digital Wealth Management thought leader.


  1. Intransparency, illusion and institutional herding are the true problems

    The Governor of the Bank of England is wrong on this point, but so are others:
    Carney, the Governor of the Bank of England says in January 2017: „Robo-advice and risk management algorithms may lead to excess volatility or increase pro-cyclicality as a result of herding, particularly if the underlying algorithms are overly-sensitive to price movements or highly correlated.“

    The first problem is intransparency:

    Even most so called professionals apparently do not know how sensitive asset allocation algorithms are to different inputs (usually different forecast e.f. for risks, returns of and dependencies between assets) as well as to small changes to the models/algorithms themselves. Simulations and scenario analysis might help, but how to correctly select the „right“ simulation or scenario for an asset allocation?

    The second problem is „illusion“:

    If professionals know about these sensitivities and do not communicate them, then they are creating an „illusion of truth“ which may also be called lying in some cases.

    The third problem is institutional herding or the perceived „job risk of institutional asset managers“:

    Most interesting is the asset allocation behavior of institutional investors. US/UK investors on the one hand and e.g. German investors on the other hand may have the same institutional consultant. This consultant should use the same asset allocation model and the same data inputs for all his clients. Asset allocations therefore should look very similar. But in reality US/UK institutional investors usually have much higher equity allocations than German institutions. Regulation is not the explanation for the difference since regulation would allow German institutions much higher equity exposures than they have today. In my opinion institutional herding ist the main problem. Studies for the US have shown, that US pension funds in general do not like to deviate from the allocations of their peers. That also seems to be the case for German institutional investors. If regulators want to reduce the risk of herding, they should encourage institutions to deviate from peer group asset allocations. Pension fund asset managers should no longer be afraid to lose their job if they deviate from the allocations of other pension fund managers.

    What can Robo-Advisors do?

    I can only speak for my company Diversifikator GmbH:

    – We developed a „pseudo-optimizer“ for our webpage to show the illusion which can be created by noble-prize winning approaches to asset allocation.
    – We thoroughly document our robust optimization and forecast free asset allocation (and ETF/stock selection approach) which starts from the world capital stock (the aggregated asset allocation of all investors worldwide or the „most passive“ allocation possible). Individualization is possible with a „decision tree“ approach: Investors who want to invest according to ESG criteria get other portfolios than conventional investors. Investors who have negative expectations on bond investments can select portfolios without bonds etc.. Lower risk portfolios are created with the addition of „Cash“ to the portfolios.
    – We show our daily performance and even all backtest data in a free data download so everybody can control the data.

    • Thanks Dirk for sharing your insights.
      Could you clarify whether Diversifikator GmbH manages assets? or are you like AdviseOnly in Italy that offers subscription based model advice?

      • Dear Efi, technically we are a provider of model portfolios and therefore do not manage assets. Investors can follow our model portfolios and should pay us based on the assets invested. The main model is B-B, though: We offer fully automated white label robo-advice solutions to regulated partners. As of today, such platforms are available in German, English, French. italian and Dutch. Greetings, Dirk

  2. My 2c – as both a FinTecher and a former Bank of England watcher.
    It was a strange contribution by Mark Carney as
    a) Pensions regulation is the biggest source of herding (and this is presumably his responsibilty)
    b) the risk in robos is mainly caused by the ETFs they hold.

    By definition the ETF complex is more likely a real source of risk within the system. The systematic risk is contained in the ETF underlyings which represent 99%ish of the robo-advice offering. The BIS and various others have been looking at this issue for at least five years now, but the otherwhelming variety of ETFs available and questions on their liquidity in times of market stress are not being asked loud enough. Whilst we expect more regulatory capital to be necessary for the providers – we’d like more guarantees on liquidity, particularly as the standard shock absorbers in a crisis have or are dissappearing.
    At RiskSave we move beyond simple ETFs to direct holdings in the underlying – we expect others to follow suit, and in this way some of this ETF-Risk should be contained. The herding issue is for Mr Carney.

  3. Well done. Systemic risk is always out there, as financial markets are part of a complex system, that behaves like a complex system, with crisis and so on. Almost surely, the main source of risk, the “hot point”, won’t be centered around roboadvisors: as you said, roboadvisors use more or less the same methods and tools used by conventional managers or financial advisors. The problem, in truth, is that the roboadvisory movement is still immature (see https://www.linkedin.com/pulse/roboadvisors-less-more-nothing-better-short-guide-donts-zenti), and roboadvisors tend to use Paleolithic tools (a short, informal survey: https://www.linkedin.com/pulse/roboadvisors-like-commodore-vic20-apparently-according-raffaele-zenti). Basically, and quite frankly, I think there’s a lot of bullshit around…

Leave a Reply

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