Newbie Thoughts: Buy cyclicals when P/E is high; sell when P/E is low
If you just started out investing, always learn to treat cyclical businesses separately from non-cyclicals. Cyclicals are industries whose profits and stock prices rise and fall with the economic cycle. Examples include automobiles, oil & gas, forestry, and (most) capital goods. I never knew this when I started out investing but semiconductor companies are also cyclical (Early on, I mistakenly thought companies like Intel and AMD were growth stocks.)
The thinking I follow is that cyclicals should be bought when P/Es are really high (or infinite or negative) and they should be sold when P/Es are low. In other words, you should buy cyclicals during recession when profits are really low and P/Es really high. This is also when stock prices tend to be low. I didn't come up with this or anything. It's a widely held view in some quarters and even Benjamin Graham talked about it in The Intelligent Investor (I believe he discussed the example of Ford.)
Very few rules in investing work all the time so there are some exceptions. If you are stockpicker, the rule may not matter if you have found a company trading way below its intrinsic value.
Or if you are a macro-type investor, the rule may not matter if you are riding a trend or theme. For example, some macro investors believe in the commodity supercycle thesis (roughly it is the belief that commodities have entered a very long bull market and will continuously rise for decades.) Obviously those investors buy commodity companies at any price because they are certain they will go up.
Anyway, although the 'buy high p/e cyclical and sell low p/e cyclical' is not perfect, it does work most of the time. I want to illustrate it with ExxonMobil, which, for those living in some foreign country, is an oil & gas supermajor. I did not cherry-pick this example and simply picked it because it is a bellwhether for the commodity complex.
The following data, extracted from Morningstar, shows ExxonMobil P/E ratios and net income. I also plotted the 10 year chart from Yahoo! Finance.
.png)
Before I say anything, I should note that my discussion is not to point out some timing system. You can't perfectly time anything off what I'm saying. Instead, it should be used as a rough guide. For instance, the financial statments don't mark the exact bottom or the exact top. The obviously can't because the yearly numbers are one data point while the chart consists of multiple points (days) within an year.
ExxonMobil had a high P/E in 1999 but, without doing research, I 'm not sure if that's because of the wild bubble in US stocks or not. A more classic scenario, representing a cyclical high P/E, occurred in 2002. As you can see XOM posted a P/E of 21.7 in 2002 which also happened to be the trough earnings. Although it's hard to tell from the chart, that was also when the stock was trading near its low. The stock started off higher in 2002 but it traded near a multi-year low by late 2002.
We are too close to the peak to definitely mark it but, conversely, it appears 2008 was peak earnings for XOM. It isn't a coincidence that this also marked the lowest P/E ratio in many years. Unless we get a major bull market in oil right away, chances are XOM's earnings will decline over the next few years, while its P/E ratio rises.
If you did not do fundamental analysis on ExxonMobil or if you did not have some macro thesis for investing in it, the ideal time to buy XOM would have been in 2002 when the P/E ratio was at its highest. If you bought late in 2002, you would have purchased it close to its decade low; if you purchased it in early 2002, it would have been above the low but still not a bad point to purchase.
Again, assuming you don't have some fundamental reason for holding onto XOM or a macro thesis that says it is going up for a long time, the ideal time to sell would have been in 2008. We are too close to 2008 to say for sure but it does appear as if the 2008 P/E ratio is a low for the time being, which would be a sell under the thinking I'm suggesting here.
Just to finish off, don't blindly follow this rule. Growth stocks within a cyclical industry will break this rule. But it's true most of the time and I follow it. So what's the best cyclicals to look at now? Autos, homebuilders, forestry and capital goods (wait a bit). All these have started posting very low earnings, or even losses, so their P/E ratios are really high. However, I'm more macro-oriented so I would factor in the macro outlook. For instance, I am bearish on world trade and think China has overcapacity in manufacturing so capital goods can probably still suffer for a while. Homebuilders are another interesting industry. Although there is huge housing inventory, some of them have held up really well. Pulte Homes (PHM), which I have mentioned in the past, is actually up around 10% for the year, and has been in a range in the last 2 years. I am not bullish on homebuilders but just pointing out how the market seems to be pricing in a bottom for them. Some would be shocked how well some of these homebuilders have done. I'm not shocked with the stock but am surprised to see the exchange-traded bond (PHA) almost back to pre-Lehman levels.
The thinking I follow is that cyclicals should be bought when P/Es are really high (or infinite or negative) and they should be sold when P/Es are low. In other words, you should buy cyclicals during recession when profits are really low and P/Es really high. This is also when stock prices tend to be low. I didn't come up with this or anything. It's a widely held view in some quarters and even Benjamin Graham talked about it in The Intelligent Investor (I believe he discussed the example of Ford.)
Very few rules in investing work all the time so there are some exceptions. If you are stockpicker, the rule may not matter if you have found a company trading way below its intrinsic value.
Or if you are a macro-type investor, the rule may not matter if you are riding a trend or theme. For example, some macro investors believe in the commodity supercycle thesis (roughly it is the belief that commodities have entered a very long bull market and will continuously rise for decades.) Obviously those investors buy commodity companies at any price because they are certain they will go up.
Anyway, although the 'buy high p/e cyclical and sell low p/e cyclical' is not perfect, it does work most of the time. I want to illustrate it with ExxonMobil, which, for those living in some foreign country, is an oil & gas supermajor. I did not cherry-pick this example and simply picked it because it is a bellwhether for the commodity complex.
The following data, extracted from Morningstar, shows ExxonMobil P/E ratios and net income. I also plotted the 10 year chart from Yahoo! Finance.
.png)
Before I say anything, I should note that my discussion is not to point out some timing system. You can't perfectly time anything off what I'm saying. Instead, it should be used as a rough guide. For instance, the financial statments don't mark the exact bottom or the exact top. The obviously can't because the yearly numbers are one data point while the chart consists of multiple points (days) within an year.
ExxonMobil had a high P/E in 1999 but, without doing research, I 'm not sure if that's because of the wild bubble in US stocks or not. A more classic scenario, representing a cyclical high P/E, occurred in 2002. As you can see XOM posted a P/E of 21.7 in 2002 which also happened to be the trough earnings. Although it's hard to tell from the chart, that was also when the stock was trading near its low. The stock started off higher in 2002 but it traded near a multi-year low by late 2002.
We are too close to the peak to definitely mark it but, conversely, it appears 2008 was peak earnings for XOM. It isn't a coincidence that this also marked the lowest P/E ratio in many years. Unless we get a major bull market in oil right away, chances are XOM's earnings will decline over the next few years, while its P/E ratio rises.
If you did not do fundamental analysis on ExxonMobil or if you did not have some macro thesis for investing in it, the ideal time to buy XOM would have been in 2002 when the P/E ratio was at its highest. If you bought late in 2002, you would have purchased it close to its decade low; if you purchased it in early 2002, it would have been above the low but still not a bad point to purchase.
Again, assuming you don't have some fundamental reason for holding onto XOM or a macro thesis that says it is going up for a long time, the ideal time to sell would have been in 2008. We are too close to 2008 to say for sure but it does appear as if the 2008 P/E ratio is a low for the time being, which would be a sell under the thinking I'm suggesting here.
Just to finish off, don't blindly follow this rule. Growth stocks within a cyclical industry will break this rule. But it's true most of the time and I follow it. So what's the best cyclicals to look at now? Autos, homebuilders, forestry and capital goods (wait a bit). All these have started posting very low earnings, or even losses, so their P/E ratios are really high. However, I'm more macro-oriented so I would factor in the macro outlook. For instance, I am bearish on world trade and think China has overcapacity in manufacturing so capital goods can probably still suffer for a while. Homebuilders are another interesting industry. Although there is huge housing inventory, some of them have held up really well. Pulte Homes (PHM), which I have mentioned in the past, is actually up around 10% for the year, and has been in a range in the last 2 years. I am not bullish on homebuilders but just pointing out how the market seems to be pricing in a bottom for them. Some would be shocked how well some of these homebuilders have done. I'm not shocked with the stock but am surprised to see the exchange-traded bond (PHA) almost back to pre-Lehman levels.
Just a thought, but perhaps the problem is using a PE that is too short.
ReplyDeleteIn a business that is highly seasonal, you wouldn't rely on one quarter of earnings.
Similarly, in a business that is highly cyclical, we shouldn't look at one year of earnings.
I wonder if we would get reasonable results by looking at longer periods with cyclicals? Say, Price to 5 or 10 years earnings?
The goal of using p/e with cyclicals is to buy when they appear cheap and sell when they appear expensive. This means you are not holding it for the full cycle.
ReplyDeleteIf you average it, you'll just get the average p/e over the full cycle. I doubt it will be that helpful unless you were evaluating the attractiveness of the business over the full cycle (i.e. you plan to buy and hold for a long time.)