Portfolio with Substantial Single-Stock Concentration

I was recently sent an interesting portfolio by an investor. This portfolio has a substantial concentration in a single stock, Procter and Gamble (PG), which the family wants to maintain. The investor had run his portfolio through our Monte Carlo planning tool and felt pretty good about the portfolio as a whole. I found the portfolio to be interesting on a number of levels, not least because many investors hold a substantial concentration in a single stock.

I asked the investor in question whether I could do a little article on his family’s portfolio and he kindly agreed to let me do it (anonymously, of course). I get a lot of positive feedback on this type of article when it shows a process for taking the portfolio apart. I get the feeling that many investors want to understand the process by which they can think through the investing process rather than being given a generic template. In that light, I will not be analyzing each individual holding or why this investor and his family choose to invest in them. I will look at the portfolio from a macro view—how all the pieces work together.

The portfolio, as sent to me, is shown above. I will note several things about the allocations. First, there is the 14% allocation to PG, which we are not going to touch. There are a couple of mining / basic materials firms represented here (AAUK, AEM) that provide light exposure to gold / precious metals. There are also a number of energy and utilities firms (EVN, ETP, SO, EXC) and real estate interests (JOE, TRC), both foreign and domestic. We have 10% in real estate (JOE, TRC). We have 10% in energy (VGENX, EVN, ETP) and 10% in energy utilities (EXC, SO). There is also a solid allocation to consumer goods (PG, JNJ, HRL) at around 20%. On top of the individual stocks, we also have some other funds which capture healthcare (VGHCX) and an emerging market ETF (EEM). There is a solid allocation to a broad large-cap growth fund (HSGFX) and to a fixed-income fund (HSTRX). Finally, we also have the holdings of conglomerates such as Berkshire Hathaway and Leucadia.

In general, this portfolio seems to have a lot of “moving parts,” i.e. a number of individual allocations at very small percentages. I do not see a lot of value in a 1%-2% allocation to something. This has to be considered in light of the body of research on the relative value of choosing individual assets vs. investing in asset classes. For an overview of the history of these studies, and their conclusions, I like this article by Vanguard’s institutional arm (pdf file).

The body of research suggests that while active management—such as choosing individual securities within a sector and timing the market—can add value, it is difficult to identify the managers who add value ahead of time. For individuals who want to pick stocks, the results suggest that it is far more important to choose the asset classes than the individual firms within an asset class. That said, investors are an optimistic bunch and many of us like to look at individual companies and believe that we are smarter than the average. It is a good idea to decide where you sit in this debate, but this portfolio sort of straddles the fence. On the one hand, there are individual stocks here, but the allocations are so small that they suggest some lack of commitment to these firms / asset classes. This is not necessarily a problem, but it may be a bit of a distraction.

When I ran this portfolio through Quantext Portfolio Planner, using three years of trailing data through 2/28/2007, I got the following results:

Over the past three years, this portfolio has generated 13.3% per year in average return, with a standard deviation of 6.2% (above). Nobody will complain about performance like that. Over this same period, an index fund like SPY or VFINX has delivered around 9.6% in average return, with a standard deviation of about 7%. This portfolio has clearly out-performed the S&P500, but this is not the most important thing to look at. Any portfolio that adds real estate, energy, and foreign assets to the S&P500 will have out-performed in the last few years. When we look at the Monte Carlo projections for future performance (Portfolio Projections above), the portfolio is projected to return 13.1% per year with a standard deviation of 14.2%. This is good overall performance for the risk level. I constantly hark back to the general rule that a 1-to-1 ratio between projected average return and projected standard deviation in return is a good benchmark to shoot for in a portfolio. To understand why, see this article.

So, our investor’s portfolio has a projected ratio of 0.92, which is pretty darn good. Quantext Portfolio Planner generates what we call the Diversification Metric [DM]. In portfolios with about this risk level (i.e. a lot of equity focus), a DM of 70% is the highest that portfolios ever get, and this portfolio sports a DM of 61%. This is a good sign. Further, I will note that this portfolio has a trailing historical R-squared (also written as R^2) of 65% and a Beta of 71%. These statistics tell us that this portfolio is not overly driven by the S&P500—which means that a broad U.S. market decline won’t tank the portfolio. From my emails with the investor who sent this portfolio to me, I know that this is not an accident. He is concerned with slowing domestic growth, so getting to a low-Beta / low R-squared portfolio is a goal.

There is another very interesting and important feature of this portfolio that is worth noting. I tend to get concerned when a portfolio’s trailing returns are notably higher than its projected future returns. In the case of this portfolio, the trailing performance is right in line with the projected future annual returns (13.1% per year vs. 13.4% per year). I have noted that this is a distinguishing feature of Berkshire-Hathaway’s top equity holdings, too. I consider this to be a desirable quality for a portfolio. This is especially positive because the portfolio is projected to hold up well in a higher volatility environment. QPP projects higher volatility for major asset classes in future years and I like to see a portfolio that does not become ridiculously volatile in that environment.

It is often valuable to stress test a portfolio by looking at how it has performed in major market declines. Unfortunately, there are several holdings in this portfolio that do not go back very far. EEM, ETP, GS, HSGFX, and HSTRX all have less than ten years of history, and some of these have substantially less. These assets make up 40% of this portfolio. I, personally, do not want 40% of my portfolio to have less than ten years of history. This is a matter of risk tolerance. Less history means it’s harder to get a sense of how a company holds up in diverse circumstances. We can estimate, but I prefer a little more history. One thing we can do is to zero out these holdings and uniformly scale up the relative proportion of the other holdings so that we have a 100% allocation. I have zeroed out these short-lived holdings and get the following:

This is not a portfolio that I am endorsing—it is just a way to see how the pieces have worked together in the portfolio for periods of time outside of the life-spans of the shorter-lived components. When I run this portfolio through Quantext Portfolio Planner for just the most recent three years (to match the analysis of the total portfolio), I obtain the results below:

The modified portfolio has many features in common with the original, albeit with higher Beta and higher total volatility (risk). This modified portfolio is a higher risk / higher return proposition not least because we no longer have our fixed income fund. The modified portfolio still has a low R-squared (R^2). The low R-squared suggests that this portfolio has not historically tracked the S&P500, which means that we can reasonably infer that this portfolio would provide some fairly good shelter in the event of a bear market in the U.S. When we look at this portfolio for the period 3/1/2000-2/28/2003, when the S&P500 generated an average annual return of -16.5% (with standard deviation of 17.8%), this modified portfolio generated an average annual return of 16.56% with a standard deviation of 13.7% (see table below).

There is another interesting feature of the bear market stress test (below). The Portfolio Projections show the outlook that Quantext Portfolio Planner [QPP] generated for the modified portfolio from 3/1/2003 forward (the historical data used to run the model ends on 2/28/2003). QPP projected that this modified portfolio would generate an average annual return of 11.79% per year, with a standard deviation of 11.75%. QPP’s projection makes this portfolio look pretty good, standing on 3/1/2004. As we noted earlier, this portfolio has generated very solid returns since that time, with reasonable risk levels, just as QPP projected.

In addition to the stress test (above), this (modified) portfolio has also generated very solid returns over the trailing ten-year period, with average annual return of 18.4% with a standard deviation of 12.9%, far higher return than the S&P500 over this same period, with less risk. That said, it is noted that this portfolio has many assets in its holdings that have had a great run in the last decade, including real estate, energy, gold, and utilities.

When we return to considering the original (total) portfolio, I conclude that this portfolio looks pretty strong. I do not believe that this is due to the detailed allocation, but rather because of the holdings that really take advantage of low correlation asset classes. All of this is managed even with a fairly large holding in a single stock, PG. This would not be the case for just any stock, but PG has some very nice portfolio qualities. Berkshire Hathaway has a major concentration in PG, and this is related to its portfolio impacts.

The previous analysis notwithstanding, this portfolio may not be the best choice for anyone, including the investor who sent it in—and this is where it is important for each investor to have a personalized plan. QPP projects the probability of being able to fund a desired income draw in retirement. There will be investors who find that they get the results that they want with a less volatile portfolio. There will also be investors who find that they will maximize their chances of funding their desired income stream if they are more aggressive.

Disclosure: the author has positions in JNJ and EXC.

Comments

[Paul Meisel:]

I like the article, but am not sure that I agree that picking asset classes is more important than picking stocks. I think that is an anti-Buffett conclusion, and if I have to choose between Buffett and Bogle, I take Buffett, no disrespect to either... and Cramer is a better read than Vanguard.

2007 Mar 07 08:32 PM

[Geoff Consi...:]

Yeah, there is something a bit paradoxical in all this. I pick individual stocks that I feel are good firms and that have good portfolio effects (Buffett). What I was trying to say is that you need to commit to one route or the other. If you are going to go with small positions in various firms, you will end up with less than ideal results. If you don't want to analyze stocks, go with asset class allocations (Bogle). If you do analyze stocks, take a smaller number of decisive equity positions (Buffett). If you read my articles you will see that I am a fan of buying individual equities in real concentrations--not trying to replicate an index. If you are nervous enough to go with 1-2% allocations to things, you may as well go with asset class investing. Thats what I was trying to say.

But Cramer? Haven't you seen the analyses showing the performance of what he pitches? He is the antithesis of rational portfolio construction :)

2007 Mar 07 11:37 PM

[Paul Meisel:]

"I pick individual stocks that I feel are good firms and that have good portfolio effects (Buffett)."

Exactly my approach, usually with stakes of about 4-5% but sometimes up to 15-18%. I have one 2% speculative position in a "local favorite" stock that my heirs will sell.

As for Cramer, he's like the Scandinavian buffer, if you eat every thing it offers you will ruin your health, so I look at the smorgasbord and pick and choose. Its a lot more entertaining that Louis Rukeyser ever was. I like NASCAR racing too, and both kinds of music, country and western.....

2007 Mar 10 09:31 AM