Two ETF Retirement Portfolio – Sometimes simplicity is still the best

The previous article on building a two-ETF retirement portfolio struck a chord with readers of this blog.

To summarise the approach, using a cheap low-cost broker with access to trading in the US markets, you can select a large ETF that covers global stocks and a similar global ETF for bonds in equal proportions. You can produce a retirement portfolio that can have over 90% chance of surviving 40 years at a reasonable withdrawal rate of 4%. The odds of success increase if you start retirement after age 65 and can supplement your expenses using CPF Life payouts.

In the back-tests done earlier, the global stock ETF is proxied by the Vanguard Total World Stock Index Fund ETF (“VT”) and the global bond proxied by Vanguard Total World Bond ETF (“BNDW”).

One point of dissatisfaction with the model is the specific asset allocation of 50:50.

Would it be possible to improve the portfolio performance with the latest and most significant investment approaches of the best hedge fund manager around the world?

One insight we have on stock and bonds ETFs is that the majority of risk that you are taking on as an investor is concentrated on the equity ETF. After attending a week’s worth of quantitative investment courses with Coursera, I was able to finally produce a diagram that can break the risk down by ETF, which I can display here.

This is the key insight of the risk-parity approach popularised by Ray Dalio’s Bridgewater Associate hedge fund.

The question is whether we can improve the performance of our retirement portfolio if we can equalise the risk contribution of both the stock and bond ETFs. Without going through the mathematics and engaging in the arcane process of quadratic optimisation, the reader should know that this should intuitively mean taking on a larger allocation into the Bond ETF.

This is exactly what I spent my weekend on:

Building a risk-parity portfolio builder on and back-tester using Python’s Jupyter Lab that can take on multiple combinations of local stocks.

If we choose an asset allocation that equalises the risk contribution, the asset allocation shifts dramatically:

More than 80% will have to be allocated to bonds to adopt this approach to investing.

Will a 2 ETF portfolio equalised for risk really do better?

The next step would be to test such a strategy and compare it to a 50:50 approach. Would a system that equalises the risk (ERC) outperform a simple 50:50 equal-weighted strategy (EW)?

Sadly, to my disappointment, the strategy underperforms not just in terms of returns, but also in terms of risk-adjusted performance (Sharpe ratio).

You can even witness the exact point in time by which the risk-parity strategy dies – it outperformed before the Covid-19 crash in March and then underperformed, after that, the lower allocation to equities prevents the portfolio from ever catching up with the 50:50 portfolio. Perhaps this reflects the recent woes of the Bridgewater Associates portfolio.

The results have been bitterly disappointing for me as I spent the more significant part of the week engineering the tool for ERM students to refine their stock allocations using this risk-parity approach. I have witnessed similar underperformance observing risk-parity portfolios consisting of local banks and blue-chip REITs.

While this is an elaborate and mathematical exercise, what are the critical lesson for the retail investor?

• Simplicity can be quite useful. It’s quite hard to beat an equal-weighted strategy in practice.  
• Just because a technique is sophisticated does not mean that it is suitable for the retail investor.

More importantly, the hand of the DIY investor has never been more fantastic. Choosing a discount broker may mean paying just around 0.08% per trade, and you can keep 2-3% in annual expenses that you would have paid to an investment professional. 

I will be trying to build other features to my asset allocation tool, such as a global minimum variance portfolio.

It is also perhaps time to start thinking about what the “opposite” of a risk-parity portfolio should look like.