Robo Investment Advisors and Psychometrics

by George Hatjoullis

This new breed of investment advice is rather interesting. Essentially, having assessed your risk preference through a set of simple questions, your funds are invested in a portfolio that matches the assessed risk preference. If the provider identifies 10 categories of risk preference then it will be managing ten portfolios and each risk preference class will own a slice of one portfolio. The entry investment is small but as entry and management involves certain fixed costs, the fees come down with size of investment. In order to assess the RIA model, and provide a benchmark for my DIY activity, I have invested a small amount (£10k) with one RIA and am adding £500 a month). I have chosen a 6 risk preference on a scale of 1-10 (“I am willing to lose some to make some”), though my DIY portfolio is more cautious. The fee for this size is 0.95%.

The first problem I encounter in comparing with my other activities is the rate of return. According to my online page my all time return is 5.32%. This is clearly not the return as it is the money gain after fees divided by the portfolio value at the end of the period. The fact that money is continuously invested understates this return calculation. The correct return is given by the time-weighted rate of return. Basically the return is calculated each time money comes in (or goes out) but before the money is added (subtracted) and then compounded (ie. a geometric mean). The provider has promised to provide such a calculation in due course but one wonders why firms do not include it in the software set up. The second problem I encounter is that my all time return differs from my year-to-date which, as I started in March, seems a bit odd. I assume the calculation is based on the UK tax year.

The target population was clearly ISA and SIPP investors. To invest with a RIA outside of a tax wrapper creates the need for some potentially complicated information to be provided to the investor for the purpose of completing self-assessment tax-returns. The version that I received for the last year was not wholly satisfactory. The problem arises because of the range of investments used and the often complex tax treatment of these investments. I may have to abandon my experiment if this becomes too untidy. The RIA seems best suited to ISA and SIPP investors where tax reporting is not required.

The range of investments used is a surprise. I am presently invested in 16 different vehicles, which is impressive for £14k. Apart from cash and physical gold they are ETF vehicles. Looked at like this the 0.95% fee does not look that bad. This RIA has an algorithm that selects from a large universe of investments to provide a risk profile that it is judged that I have chosen. It allows me access for a very small investment. I could not replicate this portfolio with £14k.

So what other potential problems arise? The main problem is me, or more precisely, you. People do not understand risk and this self-report system can lead to problems. Whatever people say up front, one does not know their risk preference until things go wrong. If I had invested £14k and the portfolio value was £12k how would I feel and how might I react? In my case I would not react at all. I set the programme up to yield a sum of 100k in 11 years. My £500 a month is now being invested at lower prices and I have a long way to go. Other investors might become alarmed and stop their monthly direct debits and worse still withdraw. A more robust risk preference assessment system is probably advisable for the provider.

Risk preference assessment is probably the main weakness of the RIA. The population of investors it is targeting is the least sophisticated and the most likely to either not understand its own risk preference or be in denial about it. This could result in inconsistent behaviour with considerable loss to the investor. In my experience people have much lower risk tolerance than they are willing to admit or understand. The problem is exacerbated by the ease of observing daily portfolio valuations. In the days when you got one valuation a year the window for overreaction was much narrower, especially as the process of divesting was cumbersome. Today you can watch value decline daily and the exit switch is just a click away.

The RIA community should be (maybe is) investing in psychometric risk assessments. I am not aware if any formats already exist but they need to be created if not. Psychometrics has established itself in a wide range of assessment roles. Employers use them widely as part of candidate selection. These systems are not a panacea but correctly constructed they can provide useful insights into personality. Risk preference is a personality type.