The Fairholme Fund (ticker symbol FAIRX) is a mutual fund with approx. $7.8 billion in assets managed by Bruce Berkowitz and associates. Morningstar awarded Mr. Berkowitz designations of the Fund Manager of the Decade and Fund Manager of the Year 2009 for Domestic Equity. Currently, Morningstar rates the fund Two Stars / Silver in the Large Value category, and says that “It’s more top-heavy than ever, but this fund is in some ways more stable.” The fund has recently been closed to new investors. Using Alpholio™ methodology, let’s see how Fairholme performed in the past eight years.
First, the total return chart, which assumes reinvestment of all distributions into the fund and each member of the reference portfolio, respectively:
From early 2005 through the trough of the recent market downturn in early 2009, the total return of Fairholme could generally be matched by the reference portfolio of ETFs. Subsequently, the performance of the fund and the reference portfolio significantly diverged, with the maximum difference at the beginning of 2011.
This is further illustrated by the cumulative RealAlpha™ chart:
An investor who bought the fund at the beginning of 2005 realized no cumulative RealAlpha™ in the four-year period to the beginning of 2009. While the cumulative RealAlpha™ peaked at about 17% in mid-2007, it was subsequently lost.
From early 2009 to early 2011, Fairholme generated a significant amount of RealAlpha™ by effectively riding the market rebound. But after that, not only did the fund lose that entire accumulated alpha, but it also subtracted an additional 15% of it. Only in 2012 did the fund restore its cumulative RealAlpha™ to a positive level (see a recent article from Bloomberg Businessweek).
It is therefore likely that Fairholme made big directional bets, which sometimes worked well, but often backfired. The overall numerical results demonstrate this very clearly:
In the entire analysis period, the annualized regular RealAlpha™ was just slightly positive. Even in the lag RealAlpha™ case, which tends to reward successful active management bets, the annualized alpha figure was not substantially higher.
Instead of investing in Fairholme in 2005 and staying put through late 2012, the investor would have been almost as well off by buying and maintaining a reference portfolio of ETFs determined by Alpholio™. This is further supported by volatility figures:
At almost 22%, the annualized volatility of returns of the fund in the entire analysis period was high, and could not be matched by that of the reference portfolio of ETFs. This implies that the fund was likely concentrated in a small number of positions. Indeed, the fund typically held less than 25 securities, with a large portion of its assets in just a few of them. For example, at the end of November 2012, more than 86% of the fund’s net assets were held in just eight stocks (see the Top Ten Holdings by Issuer table on p. 5 of the FY2012 shareholder report). Again, sometimes such a concentration was beneficial, but frequently it was not.
So, what types of securities did Fairholme hold? This chart shows the major investment “themes” of the fund over time:
The biggest equity theme by far is the equivalent position in VFH, a Vanguard ETF that represents the financial sector of the U.S. economy. Mr. Berkowitz started building up this position right after the market bottom in March 2009, and it persists to this day. Once it became clear that the U.S. government provided an effective bailout to large financial institutions, the sector rebounded. This is what generated close to 25% of cumulative RealAlpha™ of the fund by the end of 2010. However, as this large position continued, it caused the fund to underperform by losing all of that alpha and more within a year. Only in 2012 did the position rebound again and helped the fund recover some alpha, although with a significant fluctuation. Even with that RealAlpha™ gain, the investor should have carefully considered if he/she wanted to invest in, practically, a financial-sector fund.
The historical cash theme is represented by an equivalent position in SHY (short-term Treasury bond ETF). This position was at times very high as a percentage of the reference portfolio, indicating major market timing attempts in 2005 and 2007 in the fund that was expected to invest predominantly in equities. Arguably, this hurt the fund’s performance in that period. A related equivalent position in LQD (iShares Investment Grade Corporate Bond ETF) also illustrates cash-similar instruments held in this equity fund.
The next three largest equity themes are represented by JKJ (iShares Morningstar Small Core Index ETF), EWC (iShares Canadian market ETF), and VDE (Vanguard Energy ETF). An equivalent position in small-capitalization stocks is quite surprising for a “large-value” fund and signals a potential style drift. A large actual position in Canadian Natural Resources (CNQ) in 2006 explains the second of these themes.
How could an investor capture the relatively short period of time when Fairholme outperformed its reference portfolio? A simple algorithmic buy-sell signal is shown in the following chart:
The analysis starts with an assumption that the investor initially bought the fund in early 2005 and intended to hold this investment indefinitely, i.e. at least through early 2013. The blue curve depicts the cumulative RealAlpha™ in that entire period. Since there is a degree of high-frequency oscillation in that curve, its longer-term trend can be elicited from the exponential moving average (EMA) of RealAlpha™, depicted by the smoother green curve. (Note that the byproduct of smoothing is a slight lag of the EMA vs. its source.) Subsequently, a simple decision criterion is applied to determine whether the investment in the fund should be retained. So long as the fund generates positive monthly increments to cumulative RealAlpha™, the investment in the fund is considered beneficial. Conversely, if the fund’s cumulative RealAlpha™ begins to consistently decrease, the investment is no longer considered attractive.
This simple decision criterion indicates that the fund should have been divested in early 2008, reacquired in the second quarter of 2010, and divested again from the second half of 2011 until present. This led to the capture of most of the positive cumulative RealAlpha™ run during the market rebound, while avoiding the most recent fluctuation due to a dominant position in financials.