Risk parity: Variations on a theme
Perhaps the only pursuit in the institutional investment world that is sure to outlast the current search for yield is the quest for the optimal asset allocation.
Risk parity has augmented standard portfolio theory in recent years. The approach involves allocating assets according to risk rather than to euro amounts. Studies showing that it can generate superior returns over the long term have enhanced its popularity.
The catch is that the approach tends to result in relatively large proportions of portfolios being shifted into fixed-income investments. Some analysts are questioning its viability with bond markets facing a protracted period of low rates, while equities are pushing to record levels. They are suggesting a revised version that uses assumptions about potential future returns to allocate assets.
The capital-asset-pricing model tends to push investors towards holding a standard market portfolio. They can then use leverage to fine-tune the targeted return according to their risk preference. But the popularity of risk parity strategies has sparked a new line of thinking. In broad terms, the result is that institutions following the approach invest more money in low-risk assets than high-risk ones. This ensures that equities and bonds contribute the same amount of risk to a portfolio.
Risk parity ignores return projections. But leveraging fixed income at a time when benchmark bond yields are negative in many cases, while simultaneously decreasing weights to equities when stocks were rallying to all-time highs, seems a recipe for delivering sub-par returns. That has led some to seek ways of incorporating return estimates, such as the Sharpe ratio, into a modified parity approach. The returns-based challenge to risk parity strategies has sharpened the debate. It has refined understanding of how risk parity works and how it handles changing conditions.
The case to revise risk parity to a ‘Sharpe parity’ approach, in which the weighting to each asset class is proportional to the Sharpe ratio each asset, is made by two UBS Global Research strategists, Stephane Deo and Ramin Nakisa, who argue that the basic tenets of risk parity are misguided. “Managing risk is not managing return,” they say in a July 2014 paper entitled ‘Weight Watcher – Much Better Than Risk Parity: Sharpe Parity’. “Fundamentally, risk parity completely ignores return and focuses only on risk. If we have a universe of two assets, one of which is trending downwards with low volatility and the other trending upwards with high volatility, risk parity would allocate the majority of capital to the falling asset.”
At issue, they say, is the fact that risk parity fails to accurately reflect the risks of the assets it utilises in its effort to optimise returns. “Another problem with risk parity is the use of volatility as the sole measure of risk,” they say. The volatility of credit, for example, is typically much lower than that of equity, but that does not capture the risk of the asset class. A bond with a high probability of default may have low price volatility, they contend, but a portfolio manager must also consider the illiquidity of corporate bonds to be a risk, and this risk is not captured by price volatility. “For this reason, incorporating credit, particularly high-yield credit, into a risk parity framework is very dangerous.”
At a glance
• Traditional risk parity approaches, which ignore return projections, could lead investors to weight their portfolios too highly towards bonds and too little towards equities.
• Some investment managers have sought to incorporate expected return assumptions in modified risk parity models to avoid such a performance mismatch.
• The modified approach is designed to preserve the long-term benefits of risk parity allocation.
• Practitioners say closer attention to business cycles and multi-asset risk factors promise a way forward.
However, while noting limitations to the strategy, Deo and Nakisa concede what the data shows: “Overall, we find that risk parity performs well, producing good risk-adjusted returns.” The key phrase is the notion of adjustment.
According to a paper by Clifford Asness and colleagues at AQR Capital Management, it is not possible to assert that equal risk is optimal be ca use it is better diversified. “To believe that, we must also believe we are not getting paid enough in equities to be so concentrated in them,” the paper says. Risk parity is thereby not merely a method of allocating risks: “It is inherently also a statement about our views on expected returns.”
At the same time, AQR cautions against viewing risk parity as a static or passive approach. “Instead of saying that equal risk is always the best policy regardless of expected returns, a risk parity investor should say ‘we do not believe expected returns on equities are high enough to give them a disproportionate part of our risk budget’.” It goes on to contend: “This important distinction is missing from the discussion of risk parity.”
While bond-laden risk parity portfolios are not an appealing prospect at a time of low and negative interest rates, seeking to advance the risk parity paradigm by adding return targets to the asset allocation process may fail to enhance the approach. “The solution that Sharpe-based investing suggests is that we should include a kind of expected return, in order to have a certain Sharpe ratio, and then to correct for valuation issues,” says Koen Van de Maele, global head of investment engineering at Candriam Investors Group in Brussels.
“That seems more a theoretical way to overcome a problem that certainly exists,” Van de Maele continues. “I’m not sure that it will add that much value because, in the end, the huge benefit of a risk parity approach was that you could have it systematically, and you did not have to put your expected view on the markets into the portfolio construction.”
Adopting Sharpe-based methodologies requires expected return forecasts for each asset class. “I’m not sure it will be quite successful to do it that way because I’m not sure there will be reliable ways or systematic ways to estimate the expected returns,” concludes Van de Maele. “And, quite frankly, if you would know the expected returns of the asset classes in advance, I guess I won’t need any risk parity anymore.”
Allocating risk in a portfolio should ultimately be done with an eye to the prevailing market environment. Edgar Peters, a partner at First Quadrant who leads its team which manages $4bn (€3.6bn) in risk parity assets, says his approach is “based on the idea that risk changes over the business cycle”. The firm’s guiding principle is that when allocating assets it is insufficient to use an average measure of market risk. “If your head’s in the oven and your feet are in the freezer, on average you feel fine,” as he puts it.
“Most people think there’s an average, and then there are the outliers at each end and the average is where you spend most of your time,” continues Peters. “But our research shows there are two different periods – one period where risk is chronically high, and another period when risk is chronically low.” Market reality is that the distribution of risk is bimodal – it exists in two different forms and the average is where you spend the least amount of time.
Financial markets are in a period of low volatility that began in late 2012. The cycles average about four years, according to First Quadrant’s research, but can range from as short as two years to as long as eight years. Risk allocation changes accordingly. “If we are in a high volatility state, we do risk parity then,” Peters says. “You’re not rewarded for taking risk in high volatility, so the whole idea is to diversity risk as much as possible in that environment.”
First Quadrant has shifted towards a diversified growth portfolio structure in the current state of relatively stability. It has increased its allocations to equities, credit and commodities relative to that in fixed income.
Credit allocation is adjusted to reflect how the asset class’s behaviour changes in the two risk regimes. In the high-volatility regime it acts like an equity, registering an 80% correlation to stocks. “We consider credit to be a bond at present, because in low volatility environments, credit has a low volatility, roughly the same as Treasuries, and a higher correlation with bonds than with equities,” Peters notes. While bonds are primarily used in risk parity to provide downside protection, such protection can now be achieved at lower cost with options, he adds. “That is why we’re not leveraged bonds at this stage of the cycle.”
The holy grail of asset allocation may yet be some way off but risk parity is alive and well in the institutional investment market.