Rick and Larry continue to battle it out with their opposing views on using commodities in a well-diversified portfolio:
- Is there a beta for commodities? An alpha?
- How correlated are commodities benchmarks to one another?
HAI: Returns and volatilities for domestic equity benchmarks are highly correlated to one another. Are commodities benchmarks correlated to a lesser degree? What implications does this have for investment planning?
Ferri: Commodity indexes cannot be compared to stock and bond benchmarks. The risk and returns of highly diversified stock and bond indexes fall in very tight groups. The Russell 3000, the Dow Jones-Wilshire 5000, and the MSCI Total Market Index, for example, are almost identical. That shows the beta of the equity market.
Such is not the case for the commodity market. The long-term hypothetical risks and returns of commodity indexes are all over the map. Returns are based on index strategies devised to create alpha. As, such, investment planning using commodity indexes involves a leap a faith that a chosen index strategy will create alpha in the future.
Swedroe: Rick, commodity benchmarks are highly correlated. The Dow Jones-AIG Commodity Index and GSCI exhibited a 90% correlation from 1991 through September 2007. They're relatively good substitutes for one another. The correlation of GSCI to the AIG index is, in fact, tighter than that found between most equity asset classes like that of small-cap value to large-cap growth, or that of the S&P 500 to MSCI EAFE.
There's a major difference in index construction which favors the more equally weighted AIG index, though. The AIG index benefits greatly from the low correlation of the component commodities and their high volatility. It's quite logical that commodities have low correlations to one other. After all, why should gold and wheat be highly correlated? The low internal correlation and the high volatility lead to a large diversification return.
Since 1991, GSCI and the AIG index have earned similar annual returns but the AIG index has produced annualized returns that are about 2% a year higher.
Allow me to explain. Annual returns are based upon a simple arithmetic average, but annualized returns reflect the growth of an invested dollar. If, for example, Index A returns 20% in one year and 0% the next, its annual average return is 10%, but its annualized return is just 9.5%. It ends up 20% after two years. If Index B returns 10% and 10%, the average annual return is also 10%, but its total compound return is higher. It's annualized return is 10%, and after twoyears, it ends up 21%.
The greater the standard deviation, or volatility, of returns, the more negative the impact on annualized returns. AIG produced higher compound returns because rebalancing lowers standard deviation, thereby minimizing the risk impact on annual returns.
GSCI is more volatile than AIG and thus the volatility has a more negative impact on the compound returns. That's why it's important to choose the most appropriate index for a fund to replicate.
The AIG index also uses three-month Treasuries as collateral. TIPS should outperform three-month bills significantly over the long term, so investors in the PIMCO fund would benefit by perhaps another 1% or so.
Keep these points in mind when looking at backtested data using the older GSCI series. One should never look at the Commodity Research Bureau (CRB) index series - now renamed the Reuters-CRB Continuous Commodity Index - when gauging the historical returns of CCFs [collateralized commodities futures], though. For most of its history, CRB was dominated by agricultural commodities; that has little to do with economic activity. Thus, those that use that index to draw a conclusion are dealing with a "garbage in, garbage out" situation.
Ferri: AIG is a different strategy than GSCI, thus the risk and returns are expected to differ. As I said before, using TIPS is an interest rate arbitrage enhancement strategy in the PIMCO Commodities fund. That has nothing to do with an investment in commodities.
Ironically, CRB was the only live commodity index around in the 1970s, but there was little alpha generation in it. Alpha didn't show up because CRB equal-weighted all the component commodities futures. It didn't overweight energy.
So, because it doesn't prove CCF indexes work, you're going to dismiss CRB, Larry? Perhaps, going forward, the energy-heavy GSCI will be your next "bad index."
HAI: So, in the end, is there a beta for commodities? And, by extension, an alpha?
Ferri: The Russell 3000, the S&P 1500 and the Dow Jones-Wilshire 5000 all have risks and returns that are very close. The same is true for the Lehman Aggregate Bond Index and the Merrill Lynch Bond Composite. Unlike stock and bond indexes, there is no consistency of return or risk among commodity indexes. They're all over the place. That's because the investment strategies of all the indexes are radically different. An investor in commodities futures indexes is relying 100% on the active strategy of the index provider. Some commodity index providers claim their strategies add up to 4% alpha over the average return of commodity futures.
In the long term, commodities can be expected to produce the inflation rate. Thus, commodities have no beta, unless you want to call inflation beta. All the expected returns of commodity indexes are derived from investment strategy. Backtesting those strategies has shown hypothetical alpha, but there's no guarantee that it will persist in the future.
Swedroe: Since they have negative correlation to stocks, commodities' beta is negative with respect to equities. As for alpha, CCFs probably do exhibit alpha, but that simply shows that the Capital Asset Pricing Model (CAPM) or even the Fama-French three-factor model does not explain CCFs returns.
By definition, alpha is the expected value of that portion of CCFs return that is not explained by CAPM or the Fama-French model. Since there's no reason to expect CAPM or Fama-French to work well for CCFs, there's no reason to expect alpha to be zero. One might call the diversification return "alpha" for the portfolio.
HAI: If one were considering an investment in commodities, from what existing portfolio allocation should it come? Equities? Fixed income?
Ferri: From both. Since a commodity index strategy is neither a stock strategy nor a bond strategy, the best way to allocate is to take some from stocks and some from bonds. For example, an investor that is invested 50% in stocks and 50% in bonds should go with a 45% stock, 45% bond and 10% commodity allocation.
Swedroe: If you take the allocation from bonds, you'll raise the expected return and risk of the portfolio. That's not the reason to consider owning CCFs in my view. I believe that ownership ought to reduce risk. So I suggest taking the commodity allocation from the equity side, thereby reducing the risk of the overall portfolio.
I'd suggest taking about 5% to 10% of the equity exposure, depending on the risk aversion proclivity of the investor. The more risk averse, the higher the allocation. The size of the allocation should also be based upon the level of exposure to the risks of unexpected inflation. Retirees generally have higher risk exposure to unexpected inflation.
Ferri: I don't agree. Taking only from the stock side reduces the portfolio's expected return considerably because stocks have an expected real return of 4% to 6% while CCF index products have no expected real return except an unknown alpha. As such, risk is best reduced by taking half from the stock side and half from the bond side. If investors want to call commodities a separate asset class, then they should treat it as one with a separate allocation.
HAI: Are there any other issues investors considering commodities should address?
Swedroe: There are a few other issues. Since commodities provide a hedge against unexpected inflation and thus the risk of longer-term bonds, adding them to your portfolio allows you to take more risk on the bond side by extending maturities. This is almost always helpful on the municipal bond side where the yield curve is typically steeper than that of taxables. If we get long bond risk showing up, the CCF will act as a hedge. If we get the risks to CCF showing up - that is, deflation - then the longer-term bonds act as a hedge. We might think of adding CCFs and extending maturities is a bit of a free lunch. Another benefit besides the higher yields is that you reduce reinvestment risk when extending maturities.
Ferri: TIPS are a government-guaranteed hedge against unanticipated inflation. That's the investment to be used if one wants to extend bond maturity.
There's a cost to this "free lunch," too. The Vanguard Municipal bond fund's Admiral shares are cheap, costing about .09% per annum. The annual cost of a commodity fund is at least .75% before taxes. I haven't done the math, but I'm skeptical about this idea of using CCF index funds in a taxable portfolio.
Swedroe: Another consideration is that CCFs tend to produce very poor returns for a very long time and then have very short bursts of spectacular returns. It takes a very patient and disciplined investor, willing to rebalance during those long periods of underperformance and having the discipline to sell into the rallies to rebalance, to benefit from owning CCFs.
Ferri: That is true. CCFs exhibit many years of low returns with spectacular bursts every 25 years or so. However, energy and natural resource stocks produce a much smoother ride and provide positive returns even in down commodities markets. Stock index fund investors benefit more often from the current 15% current position of the U.S. equity market invested in natural resource stocks. How much more exposure to one industry do investors need?
Swedroe: Finally, I would add that since we own CCFs as insurance, one should root for CCFs to have poor returns as long as the poor returns are not a result of persistent contango. We never like to collect on our insurance. The reason for this is simple: With a negative correlation, when our small allocation to CCFs is producing below-average returns, the vast majority of our portfolio should be producing above-average returns.
Ferri: That's very expensive insurance. Not only is the cost for commodity exposure higher, the expected portfolio return is lowered when it's added, especially if the allocation comes from stocks.
Investors can purchase stock and bond beta very cheaply through market index funds. Investing in commodity index alpha is much more costly.
Investors who put commodity index products in a portfolio are paying a lot of money to give up proven stock and bond return beta for the hope of commodity index alpha, the hope that, after fees, alpha stays noncorrelated with stocks and bond beta. That's a lot of hope.
HAI: Thank you, gentlemen.
This debate started on Tuesday, February 12. Read the first part here!