When considering PEG, the first thing we need to do is come to grips with some popular folklore that has grown up around this metric. It is widely believed that value investors should seek stocks with PEG ratios at or below 1.00. Some express this another way, saying the P/E should be equal to or less than the growth rate.
Aside from the mathematical awkwardness (since P/Es are whole numbers and growth rates are really percentages, it would be more accurate to say, the P/E should be less than or equal to the growth rate times 100), there are two large problems with this idea: First, it doesn't work. Second, there's no logical reason why it should work.
Figure 1 shows the results of a series of backtests I ran on portfolio123.com for three kinds of screens:
- Select all stocks with PEG ratios less than or equal to 1.00
- Select all stocks with PEG ratios less than or equal to 2.00
- Select the 25 stocks having the lowest PEG ratios
In all cases, the universe was confined to the S&P 500. That helps us see the impact of PEG, per se, without having to explore potential small-stock effects. Also, I'm defining PEG as P/E calculated with reference to estimated current-year EPS (switching to estimated next-year EPS in the fourth quarter) divided by the consensus Wall Street expectation of the company's long-term EPS growth rate.
|Use of PEG Ratio - Backtest Results
$100 grew to . . .
3/31/01 - 7/3/08
|Latest 12 months
7/7/07 - 7/3/08
|All stocks, PEG <= 1.00
|All stocks, PEG <= 2.00
|25 lowest PEG ratios
The tests covering 3/31/01 through the present show that in all cases, the PEG strategy beat the market. We could have done a bit better by keeping PEG at or below 1.00, but the degree of superiority over the 2.00 PEG approach is trivial. Simply selecting the lowest available PEGs, regardless of level, seems promising (and we'll refine it below). Meanwhile, over the past 12 months, all the PEG strategies underperformed, although the 2.00 PEG approach was the best of the three.
In considering the failure of the 1.00 PEG strategy to shine, consider where PEG fits into the overall theory of stock valuation. We start with the Dividend Discount Model, which articulates the core relationship between stock prices and the wealth expected to result from stock ownership.
P = D / (R - G)
where P = Price, D = Dividend, R = Required rate of return, and G = expected rate of dividend growth
Dividend can also be expressed as E (EPS) times Py ((Payout Ratio). Therefore, we can re-write the equation as follows:
P = (E * Py) / (R - G)
Some algebraic reshuffling would produce the following:
P / E = Py / (R - G)
Voila! We have the theoretical formula for P/E. As expected, growth is important and as growth rises, P/E will likewise move up. But the relationship between P/E and growth is not fixed. Other things also enter into the picture: the payout ratio (as well as some alternate expressions of the theory for non-dividend-paying stocks), and R, which depends on a variety of things depending on who you ask (many would cite the capital asset pricing model or a variation thereof) with most agreeing that interest rates are a crucial ingredient. An increase interest rates would typically cause R to rise. The equation is such that if this happens, P/E would fall, something we see all the time.
Making it work
Having freed ourselves from the confines of the 1.00 PEG ceiling, and having demonstrated that when used without the artificial restriction, PEG is pretty much like other widely-used value metrics (it worked in the past but is now faltering), let's consider how we might breath some new life into it.
I'll start by screening the S&P 500 universe for stocks with PEG ratios at or below 2.00 (using PEGLT).
My June 30th blog What's Wrong With Today's Value Investing? suggested the problem is information: there's a lot of it out there nowadays so unlike the past, we cannot be so quick to say stocks have low metrics because of Wall Street neglect.
But is it possible that even now, recognizing a greater level of overall attentiveness than we've seen before, some stocks might get less attention than others? Nowadays, neglect might be too strong a word, especially since we're confining ourselves to an S&P 500 universe. Let's say, instead, lesser attention.
Let's try to model that by looking at the number of analysts who issue long-term growth rate projections for a company compared to the number of analysts who publish current-year EPS estimates for that same company.
#AnalystsLTGrthRt / #AnalystsCurFY
Using this criteria, I'll sort from the bottom and identify the 25 best stocks (i.e., those where analysts are paying least attention to the big-picture long-term growth story).
But before calling it a day, there's one more value issue we need to consider: the possibility that a low PEG, and possibly a lesser degree of analyst attention, reflects poor company prospects. In my June 30th blog, I addressed this by seeking companies with upward EPS estimate revisions. But that's not the only solution. Today, we'll try another approach. We'll do a bit of technical analysis and add to our screen requirements that the 50-day exponential moving average (EMA) be above the 200-day EMA, and that the 50-day EMA be above where it was 20 trading days ago.
EMA(50)/EMA(200) And EMA(50)/EMA(50,20)
This certainly an especially aggressive set of technical requirements, but it's likely to accomplish what we want: filter out stocks whose recent price action reflects near-term pessimism.
Figure 1 shows the result of a 7/7/07 - 7/7/08 backtest I conducted on portfolio123.com based on the 25 best stocks as defined above (lesser analyst attentiveness) and assuming the screen would be rebalanced every four weeks.
The accessories with which we surrounded PEG, low rank for analyst attentiveness and favorable EMA trends, made quite a difference! Now, the screen would turn a hypothetical starting amount of $100 into $109.70, versus $81.80 for the S&P 500.
It's nice to know the model works in a bear market. But it would better if it could also turn in a good, or at least a respectable, performance during better times too (thereby recuing the penalty for timing miscues). Figure 2 shows an excerpt of the result of a backtest that extends back to 3/31/01.
Mission accomplished. Not only did we avoid losing ground to what we achieved over the longer haul with our PEG-only tests (see Figure 1 above), we actually came out modestly ahead.
So despite the theoretical and empirical shortcomings of the 1.00 PEG folklore, it looks like this ratio can be pretty useful, even today. The key is to refrain from looking at it on its own but to think instead of PEG being part, the core if you prefer, of a package of factors with the other items, the accessories, being specifically chosen to offset the challenges of value investing. In this case, we used one accessory to recreate at least the flavor of Wall Street neglect and the other to suggest company prospects that are at least moderately favorable.
To see if this could work under real-world conditions, I tweaked a bit. I cut the number of eligible stocks from 25 to 15, set a limit of 30 percent of assets for any single sector, and then ran a portfolio123.comsimulation assuming four-week rebalancing, a $10.00 commission for each trade, and 0.25 percent price slippage for each trade.
Figures 3 and 4 show the results of a simulation that ran from 3/31/01 through the present.
Figures 5 and 6 show the results of a simulation that started 7/7/07.
Table 2 shows the stocks that would be in the above portfolio right now.
(sorted by market capitalization)|
||Oil & Gas - Integrated
|Nexen Inc. (USA)
||Oil & Gas - Integrated
|Southwestern Energy Company
||Oil & Gas - Integrated
|Coca-Cola FEMSA, S.A.B. de C.
|Joy Global Inc.
||Constr. & Agric. Machinery
|Ternium S.A. (ADR)
||Iron & Steel
|Helmerich & Payne, Inc.
||Oil Well Services & Equipment
|TransAlta Corporation (USA)
|Votorantim Celulose e Papel S
||Forestry & Wood Products
|Bucyrus International, Inc.
||Constr. & Agric. Machinery
|Dresser-Rand Group Inc.
||Misc. Capital Goods
|CAE, Inc. (USA)
||Electronic Instr. & Controls
|Overseas Shipholding Group In
||Software & Programming