Deconstructing Peter Lynch

Actively managed mutual funds have earned their place among the most unloved of paper assets, and rightly so. The high fees combined with persistent under-performance are a serious drag on growing one’s capital. Peter Lynch, famed manager of Fidelity’s Magellan Fund briefly touched on the failure of active managers in his book Beating the Street and even went so far as to suggest allocations to index funds as part of one’s portfolio. This was somewhat prescient as the book was written in the early 1990s, well before the passive investment fad was in high gear. But remember, Lynch himself was an active fund manager. Based on return alone he’s considered among the greatest ever. From May of 1977 through May of 1990 Lynch captained Magellan to an annualized return of 29.06% compared to just 15.52% for the S&P 500.

Fidelity Magellan Fund Growth of $1 — *Source: Morningstar*

He also offered up a number of suggestions that should benefit investors of all types. Most of these have been repeated numerous times before, so I’ll quickly summarize:

Keep it simple: “Never invest in any idea you can’t illustrate with a crayon.”
Hold on during rough markets: “The key to making money in stocks is not to get scared out of them.”
Don’t try to time your investments, buy them on a regular schedule month after month (dollar cost average)
Have conviction in the long-run and ignore doomsday predictions about the world coming to an end
Using past performance as an indicator of future performance is futile
Most active managers underperform the broader average return
Index funds have a major advantage over active funds when it comes to cost

The book’s meaty center was a detailed memoir of his time spent managing Magellan. Over the course of three chapters Lynch highlighted a number of the stocks he picked and the hundreds of companies that he researched along the way. Reading these chapters left no doubt–Lynch was a dyed in the wool stock-picker and he moved at a frenetic pace. During his first four years as manager the fund saw turnover exceed 300%! I ultimately found myself struggling to understand what sounded like madness, and was confused as to how Lynch was actually managing the fund. Here’s some sample quotes to illustrate:

My top 10 stocks in 1978 had p/e ratios of between 4 and 6, and in 1979, of between 3 and 5. When stocks in good companies are selling at 3-6 times earnings, the stockpicker can hardly lose. [1a]

Pep Boys, Seven Oaks, Chart House, Telecredit, Cooper Tire–now I was beginning to see that some of my favorite stocks did have something in common. These were companies with strong balance sheets and favorable prospects but most portfolio managers wouldn’t dare buy them. As I’ve mentioned before, a portfolio manager who cares about job security tends to gravitate toward acceptable holdings such as IBM, and to avoid offbeat enterprises like Seven Oaks, the aforementioned servicer with a plant in Mexico. [1b]

Sounds like a value investor? But wait. During his appearance on Lou Rukeyser’s Wall Street Week he explains

…I said that I divided the Magellan portfolio into two parts: the small-growth and cyclical stocks, and the conservative stocks. “When the market heads lower, I sell the conservative stocks and add to the others. When the market picks up, I sell some of the winners from the growth stocks and cyclical stocks and add to the conservative stocks.” Any resemblance between my actual strategy and this attempt to explain it to 8 million viewers on the spur of the moment is purely coincidental. [1c]

By his own admission he had difficulty articulating how he was running Magellan. It’s also very possible that he was changing strategies during his tenure

One of the many ironies of my career is that when Magellan was a small fund I concentrated on bigger stocks, and when it became a bigger fund I found myself concentrating on the smaller stocks. [1d]

These are only a few, and I think they illustrate my point. Keep in mind that during Lynch’s tenure Magellan was a capital appreciation fund–a license to own just about any sort of asset that the manager expects to earn a profit. In the case of Magellan this meant domestic stocks, foreign stocks and even Treasury Bonds (with their absurd double-digit yields during the early 1980s). At one point Lynch even admitted: “…if it’s sold on a stock exchange, we’ll buy it.”

The one thing I’m left wondering about is the inherit bull market that set the backdrop for Lynch’s incredible success during the 1980s. A fact that I find incredibly difficult to ignore. Did Lynch actually create Alpha, or was his success a byproduct of market exposure and/or some additional factors that are known to enhance returns (size, value, momentum, etc.)?

Fortunately we live in a world where that question can be answered. Ken French has provided us with loads of data on these various factors, and Excel is equipped with statistical tools that enable multiple regression. Combining Magellan’s total returns with data from the original Fama-French three factor model produced the following results

Fidelity Magellan Fund 3 Factor Regression (May 1977 – May 1990)
	Coefficient	t-Stat	P-Value
Intercept (Alpha)	0.7779	6.1315	7.09e-09
Market (Beta)	1.1294	36.6352	1.25e-77
SMB (Small-Cap Minus Big-Cap)	0.5149	9.9292	3.05e-18
HML (High BE/ME Minus Low BE/ME)	-0.0155	-0.2815	0.7787

The above regression returned an R² of 0.938–a pretty good fit by any measure. The results tell us that Magellan had a Beta of slightly more than 1, but nothing excessive, indicating that the fund was largely exposed to stocks.

For the size and value factors a little decoding is necessary to understand the remaining coefficients. To get a grip on these I ran the Fama-French size and value portfolios through the same regression analysis. Generally speaking a SMB coefficient greater than 0.5 indicates a portfolio that is biased towards small-cap stocks, while a coefficient less than 0.5 would signal a large-cap orientation. Since Magellan came in at approximately 0.5 it was somewhat neutral on size, which aligns with Lynch’s comments (i.e. the portfolio contained both small-cap and large-cap stocks).

Similarly, a HML coefficient greater than 0.4 indicates a portfolio that is tilted towards value while a coefficient less than 0.4 means the portfolio is focused on growth stocks. Magellan fit into the latter; however, the t-Stat and P-Value suggest that this result was not statistically significant*–so we should be careful about labeling Magellan as a growth fund.

Finally, the big number in the above table is the Alpha, a fat 0.78% per month, or 9.74% annualized. Bonkers!

One word of caution: this sort of analysis only accounts for the factors that were included. In other words, if Lynch were exposed to factors other than size and value the above model would fail to indicate this fact. The return attributed to any additional “factor” would be lumped in with Alpha.

That being said, there are additional factors that can be included to enhance the above model. In 1997 Mark Carhart expanded the original three factor model to include a fourth factor: momentum. [2] The idea with momentum is that stocks with recent upward positive returns will continue to move up and those that have recently been declining will continue to move down. Carhart captured the difference between positive and negative momentum through a momentum factor (MOM), a version of which is also available at Ken French’s site. Additionally, a quality factor–Quality-Minus-Junk or QMJ–was developed by Asness, Frazzini and Pedersen of AQR. QMJ captures the excess return generated by high quality stocks–those that are profitable, safe, reward investors through payouts, and continue to grow profits (I’m simplifying for brevity). [3]

Adding these additional factors, momentum and quality, to the three factor model and turning the regression crank produced the following results

Fidelity Magellan Fund 5 Factor Regression (May 1977 – May 1990)
	Coefficient	t-Stat	P-value
Intercept (Alpha)	0.6385	4.2766	3.35e-05
Market (Beta)	1.1234	32.4279	6.38e-70
SMB (Small-Cap Minus Big-Cap)	0.5015	8.8466	2.23e-15
HML (High BE/ME Minus Low BE/ME)	0.0431	0.6684	0.5049
MOM (Momentum)	0.1432	3.815	0.0002
QMJ (Quality Minus Junk)	0.0435	0.4226	0.6732

The R² of this larger model was similar to the three factor at 0.943. While the numbers changed slightly there isn’t anything new to report regarding the SMB and HML coefficients. Both came in close to the previous three factor model and the conclusions reached previously remain in place.

A momentum coefficient greater than zero indicates a portfolio with exposure to positive momentum (less than zero corresponds to negative momentum). At 0.14 Magellan appeared to have at least some exposure to positive momentum, and in a statistically significant way. Similarly, a QMJ coefficient greater than zero indicates a portfolio holding high quality stocks while a coefficient less than zero implies “junky” stocks. Magellan’s coefficient was close to zero, but similar to the HML coefficient it was not statistically significant.

The exposure to positive momentum caused Alpha to drop from 0.78%/month to 0.64%/month (or 7.94% annualized). That’s still a pretty substantial amount of excess return that can’t be accounted for. Overall, it would appear that Lynch used a wide range of stocks: big and small, value and growth, high quality and low quality, with a little bit of momentum. More importantly, based on the above analysis he achieved a prodigious Alpha.

At the end of all this I’m left shrugging my shoulders. Maybe Lynch did generate Alpha, but luck or skill? Keep in mind that he managed Magellan for only thirteen years. Recent history reminds us that a manager can have great success for a decade or longer before imploding. Bill Miller’s Legg Mason Value Fund (LMVTX) [4] and Bruce Berkowitz’s Fairholme Fund (FAIRX) [5] are two such examples. Beating the Street is a fantastic story about success in the stock market, but I don’t regard it with any level of practical significance when it comes to my own investment decisions. As Lynch suggested, allocating towards passively managed funds probably isn’t a bad start.

*Using the more traditional values, a t-Stat greater than 2.0 or a P-Value less than 0.05 indicate that a regression is statistically significant–meaning that we’re confident that the results are not due to pure chance. In plain English: this is a way of quantifying that the results are meaningful.

More on factor analysis
AQR: Our Model Goes to Six and Saves Value From Redundancy Along the Way
Alpha Architect: How to use the Fama French Model
Alpha Architect: Basic Factor Analysis: Simple Tools to Understand What Drives Performance

References
1. Lynch, Peter. Beating the Street. Simon & Schuster. New York, NY. 1993.
(a) p. 100
(b) p. 109
(c) p. 114
(d) p. 115
2. Carhart, Mark M. On Persistence in Mutual Fund Performance. The Journal of Finance. Vol. 52. No. 1. March 1997. http://onlinelibrary.wiley.com/doi/10.1111/j.1540-6261.1997.tb03808.x/abstract.
3. Asness, Clifford S. Andrea Frazzini and Lasse H. Pedersen. Quality Minus Junk. AQR Working Paper. October 9, 2013. https://www.aqr.com/library/working-papers/quality-minus-junk.
4. Light, Joe and Tom Lauricella. A Star Exits After Value Falls. The Wall Street Journal. November 8, 2011. http://www.wsj.com/articles/SB10001424052970203611404577043910758867408
5. Cendrowski, Scott. Fairholme Fund loses half its value and its co-manager. Fortune. October 20, 2011. http://fortune.com/2011/10/20/fairholme-fund-loses-half-its-value-and-its-co-manager/