Advisor Perspectives welcomes guest contributions. The views presented here do not necessarily represent those of Advisor Perspectives
The following is in response to a letter to the Editor from Larry Katz which appeared last week. That letter was in response to Geoff Considine’s article, What the New Normal Means for Asset Allocation, which appeared two weeks ago.
I received a number of positive responses to my article, The New Normal and Asset Allocation. A less-than-enthusiastic comment (see here) was sent in by Larry Katz, the Director of Research at Merriman, Inc. Mr. Katz took the time to draft his critique because I used one of Merriman’s published model portfolios as an example of the current standards in strategic asset allocation.
Mr. Katz’s responses are representative of the views of the ‘old normal’ in portfolio construction, and its tenets still form the bedrock of the investing world view of many advisors and investors. In the interest of maintaining a productive dialog, I have addressed Mr. Katz’s responses in the comments below.
Mr. Katz’s concerns with my proposed ‘New Normal’ world view are many, but the main issues can be summarized below:
- Mr. Katz believes that adjusting a portfolio to reflect the ‘New Normal’ themes described by PIMCO amounts to a ‘forecast’ – and that forecasts of future economic conditions are very uncertain.
- The inclusion of individual stocks is too risky and that the potential default risk of holding individual stocks is too high. It is far better to own some of everything in the market via market-cap weighted indexes
- Mr. Katz believes strongly that a portfolio tilt towards value and small cap stocks is a key source of value—and my portfolio is tilted towards large-cap stocks.
- Mr. Katz believes that I made a mistake by having a lower overall exposure to developed international equity indexes, as a whole, if the dollar does become weaker.
- Mr. Katz apparently does not believe that statistical measures of risk (such as portfolio Beta, volatility, and R-squared) are meaningful and prefers to rely on more qualitative risk measures such as the percentage of the portfolio in bonds and the total number of stocks in the portfolio.
There are other minor issues, but these are the big ones.
The idea that using a ‘forecast’ is too risky is easily cleared up. If you don’t believe that PIMCO’s New Normal is a compelling world view, don’t adjust your portfolio to reflect it. Mr. Katz finds fault with the heavy tilt of the New Normal portfolio towards the emerging markets vs. the developed markets, but this is one of the key tenets of the New Normal forecast.
My article was not a justification of the New Normal, but rather a discussion of how one might adjust a portfolio if they believed in the New Normal. There is, however, a further wrinkle. Any portfolio allocation reflects a ‘forecast.’ In Merriman’s case, the forecast is that the historical value added by a small-cap and value tilt to a portfolio will persist. There is nothing wrong with this world view - but it is a forecast. PIMCO’s New Normal is based on the idea that the financial world is changing in some fundamental ways.
Quantifying risk
Mr. Katz repeatedly refers to my model portfolio as ‘riskier’ than the original Merriman portfolio and criticizes my use of trailing three-year statistics in the article. Fair enough – but looking at longer time periods merely reinforces by analysis. I used the trailing three-year period to illustrate certain examples, but the model portfolio that I proposed is less risky than the Merriman portfolio on a wide variety of time horizons:
The first row of this table shows the annualized volatility and Beta (with respect to the S&P500) for my New Normal portfolio and for the Merriman portfolio for the three years through July 09. The next three rows show the same statistics for a variety of time horizons, including one (row 4) that excludes the most recent two years that Mr. Katz believes are a statistical anomaly. In every period, the Merriman portfolio has higher Beta and higher volatility than my proposed New Normal. Mr. Katz, however, repeatedly says that my model portfolio is riskier than the basic Merriman portfolio—without ever providing any quantitative measure of risk to back this up.
Mr. Katz’s assertions that my New Normal portfolio is riskier than the Merriman portfolio is based on the following:
- The Merriman portfolio contains allocations to 11,000 stocks via market-cap weighted funds
- The Merriman portfolio has less default risk from individual stocks because it holds so many stocks
- The Merriman portfolio is better diversified because it holds so many stocks
- The Merriman portfolio is less risky because it holds more bonds
These arguments may hold sway with a lay investor, but each of these has problems from the perspective of academic finance.
I am not aware of any analysis that supports the idea that you need thousands of stocks to be well-diversified. There are routinely articles that attempt to estimate how many randomly chosen stocks are required to diversify away as much non-systematic risk as possible. Recent estimates suggest that holdings on the order of 100-200 randomly selected stocks are sufficient to minimize the impacts of individual stock risks, even if the stocks are chosen randomly. Holding 11,000 securities is simply redundant.
Further, while Mr. Katz asserts that non-systematic risk is to be avoided because of the classical CAPM notion that this risk is not rewarded, more recent research suggests that the company specific risks are, in fact, rewarded with additional return. In the old CAPM world view, it was desirable to minimize stock-specific exposure because the risk associated with individual stocks was simply additional risk for which an investor could expect no reward. The recent research suggests that this is not the case, so the old argument that it was wise to simply buy many stocks is not justifiable.
Finally, Mr. Katz asserts that the high historical Beta of the Merriman portfolio (which substantially increases the portfolio’s exposure to market risk) can be ignored because simply looking at the percentage of the portfolio allocated to bonds as a proxy for risk is “more concrete and easier for clients to understand than speaking about Beta.” People may have a hard time grasping Beta, but it is a crucial measure of market exposure.
Quantifying diversification
Mr. Katz asserts that the Merriman portfolio is “massively diversified” because it holds so many stocks (11,000). The problem with simply counting holdings is that this is a poor measure of the diversification within a portfolio. As Professor William Goetzmann of Yale succinctly notes:
While the number of stocks in a portfolio is a useful heuristic for identifying the degree of diversification, this measure is insufficient to accurately characterize the diversification characteristics of a portfolio. To measure diversification more accurately, we exploit the covariance structure of investors’ portfolios…
This means that we can only properly account for portfolio diversification if we account for the correlations between positions—which is one of my key points. Holding thousands of stocks that are highly correlated (as a group) to the broader market means that you may, in fact, be relatively undiversified. I asserted that this is the case for the original Merriman portfolio. While Mr. Katz takes issue with this concept, he provides no substantive arguments to the contrary. Instead, he argues (with no quantitative support) that we should simply ignore the high correlations between the components of the Merriman portfolio because they might go away in the future.
For those readers who do not have the time to follow the references provided, I will simply note that all of these studies are authored by leading academics in finance—these are not fringe ideas.
Default risk and individual stocks
One of the standard arguments in favor of having very low exposure to any single stock is that this minimizes exposure to default risk if one company fails. This is clearly true. On the other hand, the argument for including individual stocks is that we can choose high quality stocks that have low Beta, low volatility, and a long history of stable performance—and this is a specific strategy that Bill Gross recommends for the New Normal. Choosing stocks based on low Beta and volatility exposes the investor to lower levels of default risk than choosing stocks at random. I explored this issue in depth in an article in March of 2008. While I did not include Bank of America (BAC) in my New Normal portfolio, Mr. Katz notes that if I had, a position in this one stock could have taken a big bite out of the portfolio in 2008.
Let’s look at this example more closely. For the 12-month period through July of 2009, BAC is down 53%. IYF, an ETF that tracks the U.S. financial sector is down 33%. By Mr. Katz’s logic, this shows that an individual stock is too risky relative to a broader aggregation of stocks. This does not tell the whole story.
When I calculate the Beta of the Merriman portfolio with respect to IYF (and using data through July 08), I get a value of 23.1%. When I calculate Beta with respect to IYF for my New Normal portfolio (using the same date range), I get a value of 14.8%. These Betas tell us how much either portfolio will tend to respond to a move in IYF. For the next twelve months, IYF dropped 33%, and this move tended to generate a loss of 7.8% in the Merriman portfolio (23.1% x -33%) vs. a loss of 5.0% in my New Normal portfolio. Now, let’s assume that we had invested 3% of the portfolio in BAC (just based on Mr. Katz’s assertion) and the remaining 97% in the original New Normal portfolio. The BAC position would contribute a 1.6% loss to the portfolio, so the total portfolio return for the 12-month period would be -6.4% (-1.6% - 0.97*5% ). In other words, based upon the net exposure to financials that we would have estimated back in July 2008, the Merriman portfolio still fared worse than my New Normal in the last 12-months, even with a 3% allocation to BAC.
The Merriman portfolio had a considerably higher exposure to financials than my proposed New Normal portfolio because of its higher Beta with respect to the S&P500 and the high weight of financials in the S&P500 prior to the crash. Mr. Katz suggests that individual stocks add too much risk. My example above suggests that being as highly correlated to the S&P500 as the Merriman portfolio adds more risk than even a substantial exposure to a big loser in 2008 like BAC. Oddly enough, Mr. Katz’s example serves to reinforce my point: a high Beta / high R-squared portfolio (like the original Merriman portfolio) was a riskier proposition in 2008-2009 than adding significant allocations to even an individual stock that sustained a major loss in 2008-2009.
The question of how much exposure to any single stock makes sense is very much a function of the characteristics of the stocks. The Vanguard Total World Stock Index has a 2% allocation to Exxon. The Vanguard Global Equity Fund has an allocation to 2.5% to Royal Dutch Shell, and almost 2% each to Pfizer and Altria. It is not meaningful to assert that a 2-3% allocation to any individual stocks is ‘too risky’ without some meaningful estimation of risk.
Exposure to Fama-French factors
An additional criticism leveled by Mr. Katz is that the original Merriman portfolio has a tilt towards smaller cap stocks and towards value stocks—which Fama and French suggested both add value—while my portfolio has a tilt towards large-cap stocks. What he ignores is that my portfolio also has a tilt towards lower Beta, which Fama and French have also noted as a significant source of performance: lower Beta portfolios out-perform. The Merriman portfolio has higher Beta, thereby sacrificing this benefit.
The tilt towards large-cap stocks in my portfolio is due to the emphasis on firms which are stable and pay dividends—one of the suggestions from Mr. Gross for coping with the New Normal.
Exposure to international equities
In the New Normal world view, the dollar will be weaker and growth will be concentrated in the emerging world. As such, exposure to developed international markets will be of limited value. This is already clear from the very high correlations between the EAFE index and the S&P500—and has been for years. Mr. Katz asserts that the strong tilt of my portfolio towards emerging markets “adds unnecessary risk” although he provides no quantitative support for this assertion—and the historical volatility measures do not support this.
Conclusions
Mr. Katz concludes that “Mr. Considine’s suggested portfolio makes several questionable bets and in so doing he arrives at a portfolio that is riskier than it needs to be.” Mr. Katz’s conclusions about the relative risk of my proposed New Normal portfolio vs. the Merriman portfolio are not, however, supported by any quantitative estimates of risk. To the contrary, he bases his arguments on his qualitative judgments using heuristics that are not justified based on academic research. I am surprised that Mr. Katz entirely discounts quantitative measures of risk in his analysis.
The PIMCO New Normal may or may not play out. Regardless, there is substantial evidence that attempting to diversify simply by combining market-cap weighted indexes and judging diversification based on ‘how many’ stocks are represented in the portfolio has not been an effective strategy in recent years, nor is there any evidence to suggest that it will be in the future. Mr. Katz asserts that correlations will come down and that this will increase the diversification benefits among market-cap weighted indexes. This may be true, but there is no support given for such a bet—and Mr. Katz even notes that these correlations have increased considerably through time. He may be correct, but this is just as much a forecast (perhaps a “questionable bet”) of specific future outcomes as those in the New Normal scenario.
If the elements of the New Normal come to pass—low growth, higher inflation, and increased importance of emerging markets—my analysis suggests that a portfolio structured like the one described in my original article will out-perform.
Quantext Portfolio Planner is a portfolio management tool. Extensive case studies, as well as access to a free extended trial, are available at http://www.quantext.com
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.
