Risk Parity

Risk Parity

Strategic Asset Allocation (SAA) starts from an analysis of historical returns, risk and covariances. It analyses numerous factors including changes in the global economy and demographic trends to come up with long term estimates of expected future returns, variances and covariances. “Efficient frontier” type of analysis searches for the optimal asset mix which will deliver the highest expected return for the target risk level acceptable to the plan sponsor. The resulting optimal strategic asset allocation mix is normally adopted for a period of 5 to 10 years. The management of the actual active portfolio mix may be slightly different and will be evaluated (in terms active returns and risk) against this long-term strategic asset allocation benchmark.

In a traditional strategic asset allocation study, asset classes can’t be levered and the sum of all investment allocation weights adds to one with no shorting allowed. 

Risk Parity relaxes some of these constraints, particularly the possibility of leveraging low risk assets to bring them to a higher level of risk with higher returns. 

Traditional strategic asset allocation allocates a significant portion of assets to higher yielding equities but in doing so, it allocates most of its risk budget to one asset class: equity. This is far from the best we can do in terms of diversification. The following table from BARRA, for a typical portfolio, shows that while equities represent 62.6% of this particular portfolio, the risk of the equity allocation represents more than 95% of the portfolio risk. While fixed income assets represent 26.7% of the total assets, they represent only 0.24% of the total portfolio risk! 

Thus the volatility of the portfolio is dominated by its stock component.  This lack of diversification of the risk budget is due to the long only constraint used in the Strategic Asset Allocation process.  Risk Parity relax this constraint.

The efficient frontier goes up since we are relaxing constraints. We can either obtain the same expected return with significantly less risk (diamond point), or increased expected return for the same risk (round point), or any other attractive scenario in between.

Note that all implementations of Risk Parity are fifferent. In some implementations, other constraints may be added which have the effect of lowering the frontier. Some implementations include tacical views. The conclusion remains: In any Strategic Asset Allocation study, if we relax the long only/leverage constraint, we obtain a more optimal solution (i.e. we end up on a higher return-risk frontier.

One can obtain leverage by borrowing or by using derivatives which adds some counterparty risk.

While Strategic Asset Allocation (SAA) will generally embed long term views of returns, risk and covariances, short term views normally are not parts of this long term asset mix target. Short term views are normally part of the implemented tactical mix with proper active risk monitoring vis-à-vis the long-term strategic benchmark. Some investors and board members are unfortunately tempted to alter the long-term strategic mix to fit their short term views with consequences that can be dramatic for their portfolio. They should be reminded of their limited forecasting skill as measured by their Information Coefficient (IC). When properly determined, the long-term SAA mix remains a tough benchmark to beat.

Risk Parity is an alternative and more efficient way of arriving at a long-term Strategic Asset Allocation mix.

Risk Parity criticisms

Many of the arguments given against RP apply as well to SAA. Some arguments against using leverage would also be applicable to the inherent leverage included in other asset classes like equity, real estate and private equity.

Ben Inker, head of asset allocation at GMO is the most cited critic: “The Dangers of Risk Parities” (The Journal of Investing, Spring 2011, Vol 20, Nb 1, pp 90-98). While he agrees with the math showing significant risk reduction and/or increased in returns for the same risk budget, his criticisms include:

1. Some asset classes have low future expected returns.

   It is generally agreed that SAA should not contain short term tactical view which are better implemented in the active portfolio with risk and returns measured against some long-term passive strategic asset allocation mix. SAA should be based on long-term views of asset returns only. With current low level of long interest rates, lower long term views of asset returns may be justified though.

   Ben Inker argues that long-term expected returns on some asset classes like bonds and commodities may not be attractive and therefore it may not be wise to leverage them. Such long-term views can be embedded in traditional SAA and in Risk Parity implementations. One should be reminded that any investor’s IC is low and that most investors had this argument for the last few years, i.e. that interest rates are now so low…

2. Some asset classes have skewed standard deviation.

   With his example on Asset Back Securities, he argues that low risk asset classes may hide some low probability/high loss scenarios and when these scenarios occur, having a levered position in those can be highly detrimental. Even the sovereign default risk by the US cannot be ignored.

   While the US equity market outperformed bonds for long holding periods, there is nothing guaranteed about this. In fact, in many countries including Canada, equities have underperformed bonds for relatively long periods (10, 20 and 30 years).

Which one is superior?

Anyone understanding risk budgeting will agree that removing the long only/leverage constraint will improve the optimal solution by leading to a portfolio with the same expected return for a lower risk budget, or higher expected return for the same risk, or anything else in between.

Risk Parity in its purest form is optimal under certain assumptions regarding expected returns, variances and covariances. Some less constrained versions may make better use of the relaxation of the long only/leverage constraint. While such approaches may certainly lead to superior ex-ante expected performance, ex-post performance over short to mid-term horizons can diverge on any side and should not be used to judge the superiority of any approach.

Missed argument and conclusion

It seems that some critical arguments may be missed. Despite decades of academic literature on the need to do a better asset-liability matching, pension plan still very often fund long term liabilities with broad bond index investments. It is well recognized that long duration bonds are a much better match for long duration liabilities. Leveraging bond investments is another way of obtaining longer duration bond investments. By leveraging bond investments, risk parity strategies achieve a higher duration portfolio and therefore a better Asset-Liability matching. This means that the volatility of the surplus would be significantly reduced. This could in fact be a second strong argument FOR removing the long only/leverage constraint.

While one would achieve a better asset-liability matching through levered bonds, bonds offer no protection against change in inflation expectation (except for inflation protected bonds). Some risk parity implementations may therefore reduce the inflation protection of the SAA by lowering equity investments and leveraging bond investments. That may not have been a problem over the last 20-30 years, but could be a problem in the future. If the liability is indexed, a potential alternative to consider would be to increase the allocation to inflation-indexed bonds vs regular bonds.

While the application of Risk Parity in its purest form may be too over-simplified, some intermediate implementation between traditional SAA and pure RP, i.e. SAA with relaxed long only/leverage constraint, should lead to a superior use of the risk budget, and in the long run, superior Sharpe ratios. It is definitely something that all investors should look at as an alternative to their current Strategic Asset Allocation program.

For more information on Risk Parity strategies and how to build a better risk-diversified portfolio, please contact us.

Dominic Clermont, ASA, MBA, CFA