Introduction
Back in December 2013, I wrote my 3rd article, which highlighted the importance of using the historic P/E to know when to purchase a stock. The basic idea was that by looking at the historic P/E, you might find that some companies are traditionally valued at a premium to the market. For example, if you look at a selection of the table I included in my original article, you can see that at the time (Column labeled 12/2013) that Aflac (NYSE:AFL) and Franklin Resources (NYSE:BEN) were below their historic P/E Ratios where Archer Daniels Midland (NYSE:ADM) is above its historic mean.
Ticker | 10 Yr | 8 Yr | 4 Yr | 2 Yr | 12/2013 | Mean |
18.4 | 13.4 | 10.1 | 9.6 | 18.9 | 14.1 | |
23.8 | 15.9 | 14.5 | 10.3 | 10.1 | 15.2 | |
20.0 | 21.8 | 16.9 | 14.0 | 16.3 | 17.7 |
The outcome of the original article was using this information an individual would be able to determine the normal range that a stock trades within to find good "reversion to mean" candidates. These are stocks that you buy with a low P/E with the expectation that as things turn around the P/E will return to its historic levels. Now this is not an immediate return. There is usually a good reason for a discount price point being paid in the market. It is always important to remember the context. If you can use this to buy low then you can turn the concept around and learn to use it to sell high.
How To Determine High and Low
By now, anyone who normally reads my work should know that I am a statistics nerd. One of the basic concepts in the statistics field is a distribution. Actually that is like 50% of statistics -- understanding the shape of distributions and the practical applications of predicting things within those distributions. A distribution is simply a list of all values that are possible within the data and the frequency of which each happens. This is generically shown in a bar chart or line chart. Most people are roughly familiar with the normal distribution (shown below).
When reading the chart, one traditionally looks at the midpoint (labeled 0 on chart above). Depending on your data, this will be close to the average of all instances. Then to the left would be instances below the average and to the right instances above. The farther you get from the center, the less likely the event has the potential to happen.
This can help us identify when to buy low and sell high. To create a distribution, you just need a long list of instances and then to create a summary of the count of each instance. How does this apply to us? Well, each day, we have a close price of a stock. On top of that, you can calculate a P/E ratio based on that days price. Then, it is a simple counting and charting.
Applying This Concept to Southern Company (SO)
Most of you do not need an introduction to Southern Company. But by way of a quick introduction, SO is a holding company that owns various named organizations that provide the power utility to 4 Southern States. A selection of names includes: Alabama Power, George Power, Gulf Power and Mississippi Power. As a traditional utility, there is a good opportunity for earning value through the dividend (currently 4%), which has been growing for 16 years, making it a dividend contender.
I bought SO back in 2015 as part of my IRA. I made buys in April and May with the last purchase being done at $43.28. At the time, earnings were $2.71 per share, which creates a P/E of just under 16. If I looked at various cuts of time, the historic P/E average was 16.3. Thus, at the time, I labeled the buy as slightly under fair value. If I had used my distribution of P.E methodology, would I still agree?
To identify the distribution of P/E, I took data going back to the beginning of 2004 (~10 years before my purchase). I then took the daily closing price from Yahoo and the earnings from the FastGraphs data for the year. I then calculated a daily P/E and charted that below.
The X-axis is a continuous data element that represents the daily P/E. The Y-axis is the count of the instances that fall into that rounded data element on the X-axis. Taking a look at the yellow bar it shows a P/E of 16.3 with 121 instances. This means over the past 10 years, SO has a closing price P/E of 16.3. This is roughly the average of the distribution. The green bar is roughly the P/E at my buy price (16). This means, at my buy price, I bought roughly 2% under the average historic P/E. That is not great, but it is reasonable.
One of the advantages of the distribution is you can see how often an event happens. At my buy price, there were 42% of instances where the P/E was lower over the history of the chart. That means the other side is true as well. I purchased at a better price than 68% of instances. So while the discount was not huge to the average P/E, it still is pretty well off compared to the history of purchasers.
So how does this apply to the header of the article -- knowing when to sell? I have debating recently with selling SO. This is mostly based on the fact that I think it has gotten ahead of itself in terms of price relative to earnings.
If we use the distribution of P/E that I charted above, we can determine when things become overpriced. Now everyone will have a different threshold, but I would think that when you approach the 85% level, then I would say that is getting to be overpriced relative to current earnings. For the SO distribution, this is around the P/E of 17.4. Where are we today? The red bar on the right is roughly where we are today (18.7 P/E). This is at the 99.2% level.
This means that over the past 10 years almost 100% of the time someone sold SO shares, the P/E is lower than where it is today. That is a pretty powerful statement for a value investor.
Limitations
It is important to remember that we do not invest in the past. The value of the organization is equal to its future potential earnings discounted back to today's dollars. We look to the future return of the organization in the forward P/E to determine if something big is about to happen, which might highlight an increase in future earnings. Using FAST Graphs again, the 2016 and 2017 earnings are 2.85 and 2.96 respectively. Using the 16.3 average P/E, that means the 2017 estimated price would be around $48.24. The current price is ~53.50, which is 10% above the forward earnings estimated price.
Conclusion
To me, this is pretty good evidence that SO is currently over priced. The only question I have now is, do I have a better alternative place to put the money to work for me today to get the same solid dividend but with a better potential for capital appreciation? That is still to be determined.
Disclosure: I am/we are long SO, AFL.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.