Risk parity – the quant favourite struggled through 2022

Asset allocation is tough. After decades of research, I don’t believe there’s a portfolio model that investors can follow blindly. And if a model promises exactly that, then investors should be very careful indeed.

Risk parity is an asset allocation model with an exclusive focus on the allocation of risks. The basic idea is that a diversified portfolio should aim to take similar amounts of risk across multiple market factors. When looking at asset allocation, risk parity argues that capital weights and pound exposures are a distraction, and investors should focus on risk instead.

That’s an important insight, and risk parity has taught us a lot about the importance of understanding the risk drivers in a portfolio. Risk parity is a favourite model among quantitative analysts (quants), and has a loyal fanbase which preaches its advantages.

Unanswered questions

A few years ago, John Southall and I delved into the risk parity world. Quants have applied the concept of risk parity to many portfolio types, but the most relevant for me is risk parity multi-asset funds.

While the main concept of diversification and the focus on risks are sound, we argued that risk parity leaves some important questions unanswered. And as a result, a naïve application of risk parity in a multi-asset context is likely to struggle.

The summary paper we published highlighted:

1. Rewarded risks: Risk parity doesn’t tell us what risks are rewarded and make sense to hold in a portfolio. In fact, some of the popular risk parity models seem to assume that any broad asset market risk (often extracted by quant techniques such as principal component analysis or linked to the broad asset classes equity, bond duration, credit etc) must be earning a risk premium and that these risk premia are similar in size once adjusted for their risk level. This didn’t seem sensible to us.
2. Backtest focus: Too often, investments are assessed mainly on historic performance. But history doesn’t repeat. In many cases, the positive historic performance in risk parity was due to the favourable bond returns – yield levels coming down over time and bonds performing well when equities sold off. That tailwind was more likely to stop when interest rates approached the zero bound. Our paper showed that a backtest rebased to the yield curve in 2015 (‘Risk parity (prospective)’) would have looked much worse, with sharper drawdowns and lower trend growth.
3. Different flavours: There are many versions of risk parity¹ – which unfortunately give relatively different results in terms of asset allocation and performance. Again, this highlights that the risk focus is really just one piece in our quest for great asset allocation.

Theory and practice

Fast forward to 2021 and 2022 – a tough period for most investors. Equities and bonds were selling off at the same time, and bonds that usually show a lower range of outcomes compared with volatile equities delivered significant losses.

What was bad for ‘regular’ multi-asset funds was a litmus test for risk parity funds. I have updated the historic analysis from 2015 to the end of 2022, and the results aren’t pretty.

As we predicted, the potential downside in a basic risk parity model was significantly higher than suggested by the backtests – I get to a drawdown of up to 30% in 2022 when extending the UK data series (see above). This is in line with the Bloomberg US data below, which shows a higher drawdown of their risk parity proxy compared with the 60:40 balanced portfolio proxy and a big performance gap of over 20% opening over that period.

Luckily, I think some risk parity managers had been tweaking their allocations after losses observed in the previous drawdown periods at the end of 2018 and 2020 and had tamed their bond allocations.

If anything, I think that highlights the points made in our analysis: risk parity is a valuable concept, but like other asset allocation approaches, it requires subjective judgement and ongoing oversight.

We haven’t found the holy grail of asset allocation yet.

1. In our paper, we highlighted naïve risk parity, maximum diversification, quantitative factors and qualitative factors. We could have added equal risk contribution, hierarchical approaches, etc.