Which asset allocation works best? A rear-view mirror back to 1926

I am a quant analyst by training and sometimes I come across a paper that simply makes me salivate. Not because it is full of recipes of good food but because it does something very simple, yet very informative. Such as the paper of Mikhail Samonov and Nonna Sorokina that tested modern asset allocation approaches for the period from 1926 to 2020.

First, they collected as much long-term asset class returns as they could find. Then, where possible, they extended the data back to 1926 and constructed portfolios along some common asset allocation approaches:

• US 60/40: The workhorse benchmark portfolio of 60% US large cap stocks and 40% US bonds.

• Global 60/40: The same as above but using global equities and global bonds (unhedged in US dollars).

• Diversified 60/40: The same as above but splitting equities and bonds up into smaller sections including small caps, emerging markets, high yield bonds, plus 6% commodities. This portfolio is constructed to represent a typical portfolio you would get from a financial advisor.

• Risk parity proxy: One third of the total portfolio risk (volatility) is allocated to US large cap equity, US government bonds and commodities.

• Endowment proxy: This portfolio replicates the average weights of modern-day endowment funds and in particular includes 13.5% private equity, 9.3% venture capital, 20% marketable alternatives (e.g. hedge funds), and 11.5% real assets.

• Factor-based portfolio: a combination of 70% of the US 60/40 portfolio and 30% invested in 15 factor indices (2% each, covering 8 equity factors, 2 currency factors, 2 bond factors and 3 commodity factors).

• Dynamic allocation: This is a dynamic allocation between US large cap stocks and US bonds targeting the same volatility as the US 60/40 portfolio over the last five years and using trailing 11-month returns as return estimates. It is a very simple approach to dynamic asset allocation used to test whether these approaches can reduce drawdowns without giving up too much return.

There are too many details to cover in the paper, so have a look for yourself if you are interested, but here are some of my main takeaways:

1. Diversification works: As if this needs any proof, but it is nice to see that the more diversified a portfolio gets, the better it is compared to the US 60/40 starting point. The Diversified 60/40 portfolio has about the same return (8.7% p.a.) as the US 60/40 portfolio but lower volatility (10.4% vs. 11.6% for the US 60/40). And while the more diversified portfolio doesn’t necessarily have less drawdown in a crisis, it gets out of the hole significantly faster than a less diversified portfolio. The diversification benefits increase if we move towards a risk-parity portfolio or an endowment-style portfolio.

2. The lowest drawdown risk exists in the dynamic portfolio allocation indicating the importance of adjusting portfolio weights in changing environments to reduce losses.

3. The highest return is shown by the endowment portfolio, which in no small part is due to the high returns in alternative asset classes like private equity and venture capital. However, this higher return comes at an important cost. Drawdowns and time under water for endowment portfolios are substantially bigger than for more conventional portfolios. If you don’t have a very long time horizon of several decades and an iron discipline to stick with the strategy in what can be years and years of underperformance, these portfolios are not for you.

4. Mixing systematic factor exposures in a portfolio can be a good middle ground between less diversified 60/40 portfolios and the more complex endowment portfolio. And they have the lowest drawdowns and time under water of all static portfolios. So, if you are not into dynamic asset allocation, why not look into a combination of many different factor portfolios from value to momentum, low volatility, etc.?