Here’s a little-known fact: The traditional 60/40 portfolio, when using the aggregate-bond index for its fixed-income allocation, has a 99% correlation to the returns of the S&P 500. Rob Arnott pointed this out in 2004, and it remains true today.
One way to overcome the limited diversification value offered by the aggregate index is to use a risk-parity approach. In this article, I explore the concept of risk parity in asset allocation and how it provides value for portfolio management.
Risk parity is a very simple idea. Investors want to combine assets with low correlations in a portfolio, so that fluctuations in one asset class will offset those in another. The simplest example is stocks and bonds, which are historically uncorrelated. Combining them will reduce a portfolio’s exposure to either of these asset classes individually.
The correlation of the portfolio returns to an individual asset class characterizes the degree to which the volatility of the portfolio is driven by that asset class.
Risk parity is gaining popularity. The State of Wisconsin’s Investment Board recently invested $600 million of its pension assets in risk-parity strategies.
The ideal situation for investors is to combine asset classes with the same level of volatility, but fairly low correlation to one another. Many of the asset classes that have low correlation to stocks, however, have very different volatility from stocks.
For example, the S&P500 has historical volatility that is almost four times the volatility of the aggregate bond index, which is why the ability of the bonds to mute swings in the S&P500 is so low despite their low correlations to the stock market. We can increase the impact of the aggregate bond allocation by increasing the allocation to bonds, but this in turn lowers the portfolio’s expected return. A portfolio with balanced volatility from stocks and bonds, for example, would consist of 20% in the S&P500 and 80% in the bond index, but this portfolio has lower risk than many investors desire.
One solution is to use leverage to increase the effective volatility of the bond position. By levering the lower-volatility asset classes, we would create a portfolio with lower correlation to the S&P500 and solve the volatility mismatch. But risk parity can also be achieved far more simply without leverage, by moving from the aggregate bond index to bond indices with higher credit or duration risk. For example, we might simply increase the duration of the bonds in the portfolio to match the volatility of equities. As we increase the risk of the bond allocation (either directly or via leverage) to more closely match the volatility of stocks, however, note that we are also increasing the portfolio’s sensitivity to interest rates.
The table below shows a series of asset classes, represented by ETFs, along with their historical volatilities and correlations to both the S&P500 and the aggregate bond index.
From the table above, we can see that the intermediate-government bond index (IEF) and the long-government bond index (TLT) have markedly higher volatility than the aggregate bond index (AGG), thereby making it easier to match the risk in the bond portion to the risk in the stock portion of the portfolio. Indeed, intermediate- and long-government bonds provide the perfect way to illustrate risk parity. Both IEF and TLT have essentially identical correlations to the S&P500. TLT is, however, about twice as volatile as IEF, and its volatility is close to that of the S&P500.
Note that gold and commodities are also natural asset classes for achieving risk parity with equities, because they have low correlation and similar volatility to equities. In this article, however, I am limiting my focus to bonds.
I used Quantext Portfolio Planner (QPP) to create projections for the risk and return of portfolios that are simple mixes of the S&P500 fund (SPY) with either TLT or IEF.
Risk vs. Return for S&P500 mixed with Intermediate Bonds and S&P500 mixed with Long Bonds
This chart illustrates the motivation behind risk parity. Each point on the chart represents the expected risk and return for a portfolio. The blue line (triangles) shows the risk and return for portfolios that are comprised of various mixes of the S&P500 and intermediate-term government bonds (IEF). This curve shows the familiar tradeoff between risk and return as we increase the allocation to bonds. The orange line (circles) shows the same set of portfolio allocations, but I have substituted long-term government bonds (TLT) for IEF in the bond portion of the portfolio.
The orange line offers superior risk-adjusted returns compared to the blue line. For every long-bond (orange) portfolio with comparable risk to an intermediate-bond (blue) portfolio (with the lone exception of the 100% bond portfolio), the long-bond portfolio will generate higher returns for the same risk level. This effect is not small. There is quite a range of risk over which the long-bond allocations will generate more than 1% per year in additional return. This higher expected return is the direct consequence of the risk-parity effect. The fact that longer-term government bonds have volatility that is closer to that of the S&P500 is the reason the allocations with longer-term bonds have higher expected returns.
While we have been looking at examples using historical risk and correlation and projected risk levels from a portfolio simulation, it is straightforward to corroborate the basic features of these results with option prices. The table below compares the average duration of the bond index ETFs that we have examined and the implied volatilities of the options on these ETFs.
The intermediate and long government bond indices have higher duration and, therefore, more interest rate risk than the aggregate-bond index. As such, the implied volatilities from options prices are higher for these ETFs. IEF and TLT are demonstrably more volatile than AGG and, based on their implied volatilities, we can expect this to continue into the future.
Historical volatility, simulated volatility, and option-implied volatility all show that TLT is the best choice for a risk-parity strategy, maximizing expected return for portfolios with applicable target-risk levels.
My portfolio simulations confirm the risk-parity effect, based on historical data on current option prices. You can create a portfolio with more expected return for a given level of risk by selecting assets with similar volatilities. We could expand these results using commodities or gold instead of or in addition to IEF and TLT, because both of these assets have fairly low correlations but similar volatility to equities.
Discussion
The effectiveness of risk-parity strategies in increasing portfolio return for a given level of risk is a natural consequence of basic portfolio theory. Having determined the target risk level for a portfolio, it makes perfect sense to combine asset classes with low correlation and comparable volatility to create a portfolio with the highest possible expected return. Specifically combining alternative asset classes with equities is not necessary, however, if you have a robust set of expected returns and risks for a range of traditional asset classes.
A mean-variance optimizer that maximizes return for a given level of risk will naturally overweight the asset classes with better volatility matching. Put another way, the optimizer will capture much of the risk-parity effect as long as you include the appropriate assets classes in the opportunity set available to it.
The relative outperformance of bonds in recent decades will favor the higher allocation to bonds that inevitably results from a risk-parity approach. Don’t accept risk parity because it has outperformed in this period of disproportionately high bond returns, however. While we have seen a multi-decade period of declining interest rates, these trends cannot continue forever.
Cliff Asness of AQR argues that risk parity is a generator of alpha – that this approach is not simply the result of portfolio effects, as in my examples. His position is that low-beta assets are consistently under-valued, and using leverage to increase the expected risk and return of these assets generates meaningful alpha. This argument is novel, and there is considerable support for the idea that low-beta equities are consistently underpriced. Fama and French noted this in 2003, and this effect has stood the test of time.
My own research into low-beta strategies is consistent with the academic research, but one may invest in low-beta asset classes independent of an explicit risk-parity strategy. The evidence that low-beta assets tend to be undervalued supports the idea that leveraging these assets, an alternate approach to risk-parity noted above, will lead to a superior portfolio.
In May of 2010, James Montier of GMO challenged the merits of strategic asset allocation in an article titled I want to Break Free. He argued that risk parity strategies have a number of flaws.
In the first of his arguments, he wrote the following:
…if you are targeting constant portfolio risk and one of your assets increases in volatility, then you are likely to want to reduce it (i.e., sell as volatility rises, and buy as volatility falls)…You would be selling throughout 2008 as the market fell and valuations improved, and you would be buying in 2009 as the market rose and valuations deteriorated. That sounds awfully like a momentum strategy of buying high and selling low, rather than a value strategy of buying low and selling high.
If you use a risk-parity approach in a tactical framework (re-allocating the portfolio in response to asset class volatility), Montier’s words may apply to you, but that in no way diminishes the basic value of volatility-matching demonstrated in my earlier examples. To make this point clear, I ran a risk-parity analysis for portfolios made up of TLT and SPY as of the end of 2007 and the end of 2008, and here is how these portfolios performed in each of the subsequent years:
The row labeled Risk Parity shows results for a strategy using SPY and TLT that was formed at the end of 2007 and rebalanced at the end of 2008.
The risk-parity allocations slightly underperform a 50/50 split between SPY and TLT, but both outperform a 60/40 portfolio. The risk-parity portfolios lose some ground due to the effect Montier points out: there is a higher equity allocation at the end of 2007 (52%) than at the end of 2008 (47%). Both the 50/50 and risk-parity portfolio, however, dramatically reduce the portfolio swings from year to year when compared to the 60/40 portfolio. The underlying principle is sound.
Montier’s second critique was that leverage brings its own set of risks by increasing portfolio exposure to certain factors. He correctly notes, for example, that levering the bond portion of a portfolio will increase the portfolio’s sensitivity to interest rates. We see this effect quite clearly when we increase the volatility and duration of the bond ETFs in our allocation. Adding leverage to a low-volatility, short-duration bond fund is quite similar: both effective duration and volatility will increase.
Montier’s point is that we are in a period of historically low interest rates, and rate increases in coming years are inevitable. Do investors really want to increase their portfolio exposure when rates have nowhere to go but up? This point relates to tactical considerations, but it deserves attention nonetheless.
Montier’s main thesis is that risk-parity asset allocation is likely to be inferior to a thoughtful combination of strategic and tactical factors. I agree. His point, however, does not negate the value of risk-parity considerations in a strategic framework, and we have seen that the benefits of risk parity can be exploited without leverage.
In my opinion, there are two considerations for investors thinking about risk-parity. First, remember that a good asset-allocation model already captures much of the benefit that risk parity offers. If your asset allocation recognizes that long-term bonds have volatility comparable to equities, you already capture a large part of the alpha from risk parity. Further, additional factor exposures bear consideration. Most notably, investors must be aware that there is little question that a risk parity portfolio that holds bonds in the current environment is quite likely to end up with substantially greater exposure to rising rates.
Those caveats aside, the expected alpha from risk parity comes from two factors: volatility-matching between assets in a portfolio and the alpha produced by undervalued low-beta assets. The first of these two factors is well understood and already implicit in standard portfolio theory. The second factor is the subject of ongoing research and deserves consideration in this light. Given the known and potential benefits, some attention to risk-parity effects needs to be on investors’ radar screens.
Geoff Considine is the author of a new book, Survival Guide for a Post-Pension World, as well as a book on the use of options strategies in wealth management. More information is available at www.quantext.com.
Geoff’s firm, Quantext is a strategic adviser to FOLIOfn,Inc. (www.foliofn.com), an innovative brokerage firm specializing in offering and trading portfolios for advisors and individual investors.
Read more articles by Geoff Considine, Ph.D.
