After a decade of extremely high macroeconomic uncertainties and increased volatility within the stock markets, investors have started to re-think their traditional asset allocation models that have often fallen short of expected returns/losses. For that reason, new risk-based portfolio construction techniques like "Risk Parity" have become extremely popular among researchers and investors. For example, the Invesco Balanced-Risk Allocation Fund (ABRIX) just recently passed USD 1.5 billion in size while the AQR Risk Parity (AQRIX) has swollen nearly to USD 1 billion. After such a success, even Global X has filed paperwork with the SEC for a "Global X Risk Parity ETF".
Although the mathematical calculation of a risk parity portfolio might be a bit challenging, the main idea of this approach is quite simple: each asset class within the portfolio should contribute the same amount of risk (volatility) and therefore assets classes with lower risk (volatility), such as bonds, will get a larger part within the portfolio than high risk ones such as stocks. If an investor would only invest in an equal risk stocks & bonds portfolio, the fixed income part would have at least a weighting of 90 percent and therefore it is quite common to use leverage to increase the overall portfolio volatility and thus the expected return of the portfolio.
Despite the fact, that most risk parity funds have racked up an impressive track record so far, we would like to highlight some important facts, investors should know about this new portfolio construction technique. In general, risk parity is just a broader definition of portfolios that are balancing the risk of each underlying security within the portfolio equally.
Since each underlying security within the portfolio should contribute the same amount of risk, the strategic asset allocation becomes more important than ever! Theoretically, if a risk parity portfolio consists of four equity ETFs and one bond ETF, each of them should contribute the same amount of risk. In that specific case, the equity ETFs would contribute 4/5 of the overall portfolio risk, no matter how high those ETFs are correlated to each other! For that reason, it is widely common to build clusters of different asset classes which should contribute the same amount of risk to the overall portfolio volatility, whereas within each bucket all single securities are also being weighted according to the risk parity approach. For example, a risk parity portfolio could consist of three asset buckets (equity, commodity and fixed income) which are each contributing one third of the overall portfolio risk. Furthermore, within each bucket, every single security contributes the same amount of risk to the specific bucket. Despite the fact that this is a good way to avoid the problem mentioned above, investors who are searching for an exposure to this strategy, should have a close look, which asset classes/clusters are being risk balanced and more importantly, how their correlation coefficient looks like. In contrast to the "Most Diversified Portfolio", the weightings of the risk parity approach are not determined by their correlation coefficient and therefore the correlations of the underlying asset clusters can be a huge stumbling block!
In our article, we would like to review a hypothetical risk parity portfolio that consists of three risk balanced asset clusters (equity, commodity and fixed income), whereas within each bucket all single securities are also being weighted according to the risk parity approach.
Each asset cluster is contributing 1/3 of the overall portfolio risk (volatility)
Fixed Income Cluster
S&P 500 (IVV)
Nasdaq 100 (QQQ)
Stoxx 50 (FEZ)
FTSE 100 (EWU)
Hang Seng (EWH)
U.S. Treasuries (IEF)
UK Gilts (IGLT)
German Bunds (BUND)
Australian Govies (AUD)
Canadian Govies (CAD)
Japanese Govies (JGBL)
In our example, there is no allowance for transaction costs or brokerage fees. Moreover, we rebalance the portfolio on a weekly basis, whereas 1 day slippage is included and we are using a leverage factor of 4, since it is quite common to do so. As we have mainly used futures, the tickers mentioned above are just proxies.
Chart 1 compares our theoretical risk parity portfolio with the ABRIX and there we can see that those two portfolios are performing quite similar. This example shows that clustering is a quite classical way how to construct a risk-parity portfolio.
Chart 1: Theoretical Risk Parity Portfolio
However, in Chart 2 we have highlighted the rolling 40 weeks correlation coefficient of our underlying asset clusters as well as the historical performance of our theoretical risk parity portfolio. In general, a risk parity portfolio is well diversified, since each underlying asset cluster is contributing the same amount of risk to the overall portfolio. So even if we see falling bond prices in the future, the decrease will most likely be offset by an increase in equities and/or commodities as long as their correlation coefficient to bonds remain low or negative.
Chart 2: Rolling Correlation Coefficient
Nevertheless, we can see that an increase in the correlation coefficients of the underlying asset clusters is only a problem, if the overall performances of those specific buckets are also negative during the same time period. In 2008, the correlation coefficient among stocks and commodities has increased significantly and at the same time both buckets have performed negative. The same was true in 2001, where the correlation between stocks and bonds increased nearly to 64%. In such a scenario, the risk parity portfolio will not be diversified at all and the events of tail risks are increasing (Chart 3)!
Chart 3: Largest Drawdowns
Chart 4: Ratios
The bottom line: despite the fact that the risk parity approach has some disadvantages over other risk based asset allocation strategies, if offers investors an attractive risk/return profile and has therefore clearly a competitive edge over traditional balanced portfolios. Nevertheless, it can be quite dangerous to think that this strategy is a free lunch, since most risk parity funds/ETFs have not experienced a significant drawdown so far!