Risk & Portfolio Construction: Villain with a role to play?
If you want to maintain decent returns with genuinely diversified risks, some practitioners insist that leverage is a necessity. Joseph Mariathasan finds out why, and looks at the counter-argument from multi-asset strategists who disagree
Practitioners of pure risk-parity strategies argue that true diversification of risk (as opposed to diversification of capital allocation to different asset classes) necessarily results in a portfolio whose expected long-term return is below that required by most investors – because the optimal portfolio without leverage must allocate a large amount of capital to assets with the lowest average volatility (and therefore the lowest expected returns).
Risk parity addresses this problem by applying leverage to the less volatile asset classes via derivatives in order to equalise all asset classes’ contribution of volatility to the volatility of the portfolio as a whole. Should we conclude that true diversification of a portfolio with reasonable expected returns requires investors to embrace the major ‘villains’ of the financial crisis – leverage and derivatives?
Proponents of those long-only and generally non-geared strategies labelled variously as ‘diversified growth’, ‘target return’ or sometimes ‘new balanced’ would strongly suggest not.
“We have achieved equity-like returns with much lower risk over the past decade,” declares Percival Stanion, head of asset allocation at Baring Asset Management, whose multi-asset strategies aim to have no more than 70% of equity market volatility while maintaining equity-type returns over the long term, with a maximum of 65% of equities in their portfolio.
Marcus Mollan, head of solutions research at Legal & General Investment Management, does not think leverage is required, either. Its typical multi-asset portfolio would currently have around 35-40% in equities, but would still be expected to generate returns akin to equities with lower volatility, he says.
So how are they able to achieve these equity-like returns while investing part of their portfolios in lower-return assets? There are two drivers of long-term performance, these fund managers argue.
Stanion offers the benefits of adopting a strong valuation discipline in asset allocation, which enables portfolios to avoid the worst of the downside losses in equities: long-term performance-matching equities is achieved by essentially cutting off the tail losses, which inevitably also leads to underperforming equities during bull markets. In Baring’s case, that means that allocation to equities has varied significantly. In 2005-06 it was 65%; at the beginning of 2007, it had dropped to 60% – and it was at a low of 15% by the summer of 2008.
“We then built up our positions in equities back to 40%, alongside a large credit exposure, and played those themes in 2010 and 2011 before moving our equity allocation back down to 20%,” says Stanion. “We have now built it back up to 50%.”
The other driver of higher returns is offered by Mollan – the ‘diversification bonus’. The expected rate of return of a combination of assets will be higher than just the weighted average of the returns from the individual assets, thanks to the effect of frequent rebalancing. By ‘banking’ a price rise in a certain asset class, rebalancing cuts some losses from mean reversion, resulting in a lower-volatility return series from each asset class and a geometric return that is far higher than the arithmetic return for the whole portfolio. Mollan sees this effect as adding 60-70 basis points each year to portfolio performance.
“The net result is a portfolio that is less volatile than equities but which will perform well – although it will underperform if equities produce stellar returns,” he says.
The important point about both sources of return maximisation is that they require some kind of forecast of future returns and risk. At bottom, this is about timing markets. In the case of Barings and its valuation-based allocation, this is obvious. The same holds true at LGIM, which varies its actual weightings from global market capitalisation weightings by taking into account forecasts of return (based on risk premia) together with an analysis of historic and forward-looking marginal contribution to downside risk.
By assuming a linear relationship between risk and return, returns from different asset classes can be calibrated according to their contribution to marginal risk. Underlying the portfolio construction is a quantitative analysis of how markets behave, based on the historical but de-trended behaviour of six asset classes – global equities, nominal government bonds, inflation linked government bonds, investment grade credit, property and a miscellaneous group incorporating commodities, infrastructure, emerging market debt and high yield bonds. A Monte Carlo type simulation helps form the best construction of portfolios according to the desired risk-return characteristics, with the typical portfolio holding around 35%-40% in equities.
LGIM’s approach is proving particularly popular as the default choice for UK DC pension schemes, where the absence of leverage and derivatives is a critical component of any marketing proposition for this essentially retail investor base. But for sophisticated institutional investors, is there a better alternative that might be considered, even though it requires the use of those ‘villains’ of the financial crisis?
Michael Mendelson, a principal and portfolio manager at AQR Capital Management, would say yes.
“The objective of diversification should be to generate a higher risk-adjusted return for your portfolio, as against just having the appearance of having lots of different things within it,” he says. “For that, you need leverage, or else you will be a low-risk investor. If you decide that you don’t want to have leverage but want to have enough risk, then you need to load the portfolio with high-risk assets, which have to be equities.”
As Mendelson points out, the more exposure a pension fund has to equities, the greater its asset mismatch against liabilities.
A key component of the risk-parity approaches such as those that AQR provide is a reduction in the equity weightings with an increase in bond weightings, together with a limited amount of leverage. They often use futures, which are the easiest way of obtaining leverage. AQR’s portfolios would generally have investment exposures of 200% of the asset values, adjusted as markets become more or less risky to keep the overall portfolio risk under control at all times. “If I have a 10% volatility level in a portfolio and the market risk doubles, then we will cut exposures in half,” says Mendelson. He notes that in 2008 AQR was applying very little leverage to its portfolios but still had a full risk exposure.
“The contention of risk parity strategies is that the risk contribution from fixed income is too low unless you leverage up the exposure to bonds,” as Ludger Hentschel at MSCI puts it.
“Risk parity approaches have been spectacularly successful in recent years. But bond yields are at 30 or 40-year lows, so there are big risks in leveraged bond portfolios should yields rise.”
This is a point made, implicitly, by Stanion at Barings, too. “We were able to add more value through tactical asset allocation than I would have expected,” he says. “But with bond yields now so low, the challenge is to find diversifying assets that can offset equity volatility.”
While he favours equities against “incredibly expensive” bonds at the moment – and advises pension funds that are selling equities to move into multi-asset strategies to hang on to them, instead – risk parity practitioners are, by definition, doubled down on that same bond risk.
Mendelson chooses to turn the observation around, arguing that, while low bond yields do imply that expected bond returns are low, it does not follow that bonds must inevitably go down in price.
“The low yields are priced into equity valuations as well,” he observes. “Rising rates may not be a great environment for stocks either. Lower bond yields imply that expected returns across a whole portfolio may be lower. You may have a higher natural exposure to bonds but you have no more risk than you have in equities. The idea is not to have lots of bond exposures, but to get away from equity concentration.”
Risk parity approaches are just an extension of traditional mean-variance optimisation with one constraint removed – namely, leverage.
“If you allow yourself to have some leverage in a mean-variance optimisation, you will immediately get some risk parity into your portfolio,” Mendelson argues. “Clearly there are risks associated with leverage, but cutting off all leverage is too extreme.”
Comparing the risk-parity approach to a traditional long-only multi-asset approach can be thought of as replacing an element of equity risk with the risks associated with leverage.
“There are lots of asset classes with leverage, such as private equity, real estate and so on,” says Mendelson. “The real job of a risk-parity manager is to be able to manage leverage.”
The idea of risk parity is not that it should necessarily replace a complete portfolio, Mendelson argues, but that the philosophy behind it should enable investors to adopt a more sensible approach to considering the trade-offs between different risks, whether those of equities, or of leverage.