Risk & Portfolio Construction: 21st-century portfolio construction
Dan Mikulskis makes the case for a risk-parity approach to constructing portfolios of liquid risk premiums
The events of 2008-09 prompted a long hard look at traditional investment portfolios; they were found to be flimsy in the face of the crisis, despite according with principles enshrined in a half-century or so of academic research. Since then, academics and others have suggested ways in which portfolios could have been constructed to fare better. Here, we examine one such approach – risk parity.
Portfolio construction has always been about putting an investment portfolio together in the simplest way that takes an agnostic approach to future asset returns and risks. The investor seeks to separate the strategic from the tactical, stepping back from forecasts and allocating for the long term, independent of market views.
Harry Markowitz in 1950 established the idea of the efficient frontier, laying the foundations of portfolio theory, the principles of which still stand. We suggest the following three steps can be used to address the portfolio construction challenge:
• Decide on the sources of return for which there is a fundamental reason. Not all risks are rewarded.
• Combine these, using available asset classes, in the way most neutral to a view on their future behaviour. Do not introduce spurious dependence on uncertain estimates.
• Scale the overall allocation to meet the return target and risk budget of the investor.
The obvious outcome of such principles is a portfolio which has at its core (but is not limited to) an allocation to liquid markets according with the risk parity principle.
Traditional portfolio construction
Although frequently and conveniently used as a benchmark, in reality a 60/40 stocks/bonds portfolio was never a standard asset allocation. Many pre-crisis portfolios, though, did feature fixed weight allocations to asset classes; allocations split between a ‘return-seeking’ portfolio, made up of asset classes assumed to provide risk premia (often equities) and a ‘low-risk’ portfolio (often domestic government bonds), split according to risk appetite; and allocations to credit and the dangerously-named ‘alternatives’ (comprising a variety of assets from hedge funds to private equity and currency).
Looking back, it is clear that a disproportionate amount of portfolio risk stemmed from the equity allocation. Credit tumbled with equity in 2008, and it transpired that many strategies in the alternatives bucket were, in fact, vulnerable to a broad liquidation of asset positions, meaning they plunged in value together. It also became clear that many active managers were doing nothing more than systematically harvesting existing risk premia. The realised equity risk premium over the last 10-15 years was in fact much lower than investors expected ex-ante, and the misguided dependence on this asset class was exposed.
Conversely, risk-adjusted returns of bonds were higher.
These reflections naturally led investors to reassess diversification. The focus shifted towards finding genuine risk premia in asset markets, regardless of market views. The result was a dynamic allocation to each risk premium, keeping the risk from each source constant and equal – risk parity began to attract interest. One 2012 survey covering 24 investment-consulting firms found that 78% of them undertook coverage of risk-parity managers, with 91% of those recommending them to clients.
Less is more
The bedrock of risk parity is the idea that allocation should be to sources of risk premia, not to asset classes, and that those sources of risk premia should be as diverse as possible. Academic research has suggested that the number of genuinely diversified sources of risk premium could be less than conventionally thought –maybe no more than three or four.
Focusing on such a small number of sources of return might seem overly simplistic, especially considering portfolio construction prior to 2008 emphasised ‘diversifying’ across many, many assets. The risk-parity portfolio can cover an investor’s exposures to liquid risk premia, but some assets fall outside of this and should still form a part of an investor’s portfolio – for example contractual (credit) cashflows, where these can be invested in at a rate of return that meets the investor’s return objectives after allowing for default. Writing in IPE’s 2011 supplement on risk parity, investment manager AQR commented that “Risk parity should be a complement to, not a replacement for, a traditional portfolio”.
In line with that view, an ally to risk parity is an allocation to trend-following or momentum-investing styles. Several reasons explored in academic literature, mainly behavioural, account for why trends in investment markets persist. Steve Thomas and Andrew Clare of Cass Business School, in particular, showed in an August 2012 paper that a multi-asset risk-parity allocation achieves a better Sharpe ratio than equities alone, or a fixed equally-weighted allocation among asset classes, and that this can be further enhanced by applying trend and momentum rules.
Well researched, documented and subjected to detailed scrutiny, risk parity faces valid objections that rightly identify situations in which it is not the optimal portfolio – after all, there is no free lunch – but many of its mainstream criticisms are based on misunderstandings.
Two common complaints have emerged: first, that leverage is bad; and second, that risk parity will not work if interest rates are low.
Excess leverage was certainly at the heart of the events of 2008-09 and naturally many feel it is best avoided in future. But in most implementations, risk parity does involve financial leverage. However, it is important to understand the economic equivalence of this leveraged position. A ‘conventional’ allocation to equities actually contains implicit leverage, considering a stock is usually a leveraged exposure to the underlying assets of a company. Indeed, it can be shown that, on average, a risk-parity portfolio contains less leverage than a conventional portfolio once the implicit leverage of equities is considered.
When it does use more leverage, it does so in a dynamic way, adjusting according to market conditions.
The Japanese experience disputes the claim about risk parity and low interest rates. Over the past 15 years the yield on 10-year JGBs has remained roughly constant, but they have delivered a substantial risk-adjusted return by rolling down an upward sloping yield curve.
Further, interest rates must rise by more than suggested by the yield curve for the fixed income component to deliver a negative return.
In fact, risk-parity portfolios are at their most vulnerable to negative returns when sudden unexpected moves in underlying asset classes occur, such as the surprise Fed tightening in 1994, because there is no chance to reduce exposures.
Long-held views about portfolio construction have certainly been challenged, in the last few years. Portfolios previously thought robust have been tested and found wanting. Now, in the period of reflection, our industry is faced with a chance to reassess our engrained assumptions. As academic research has a chance to catch up with recent events, new ideas such as those discussed here, come to the fore. And, as always, detailed scrutiny of new ideas will be key, along with the ability to adapt to proven new practices rather than remaining anchored to historical ones.
Dan Mikulskis is a director in the ALM & investment strategy team at Redington