Saturday, 10 August 2013

Are stocks cheap? Does it even matter? (11 Aug 2013)

Bank of America Merrill Lynch Strategist Savita Subramanian penned a research article earlier in the week that has received a lot of attention.

In the article, Ms Subramanian looks at 15 different valuation metrics (see table below) for US and comes to the conclusion that "the market is trading below or in line with historical norms".

It is perfectly understandable why a sell-side analyst working for a broker-dealer would be keen to justify why equities are cheap/fairly valued, particularly when equity markets (well, the S&P500) are at or near all time highs. And particularly when a significant portion of the investment community remains sceptical towards equities and believes Central Bank policy perhaps has more than a little to do with the post-crisis performance of equities.

I'm not saying I completely disagree with Ms Subramanian's analysis and conclusions. The bigger question in my mind is whether the conclusions even matter, given where valuations are presently.

Trailing PE

source: http://www.multpl.com/

Let's take a look at trailing PE. (I will leave forward PE alone as that is based on analysts' earnings estimates which are always wrong. We just don't know the direction in which they will be wrong - although given the consensus has not once in the past 30 years forecast an earnings recession for the subsequent year they err on the side of being too optimistic - or the order of magnitude of the error.

According to the table above, stocks are fair to undervalued using trailing PE. A key disadvantage of using a trailing PE as a method of market valuation is that it doesn't tell you where you are in the business cycle and therefore has very little predicative power.

Consider that corporate profit margins are currently at record highs (see chart below which shows corporate profits/GDP):




A large part of the reason why margins are so high is because wages as a percentage of the economy are at record lows:









If current profit margins are unsustainable (e.g. because efficiency gains have been fully realised this business cycle) and start to mean revert, one would expect corporate earnings could start to fall unless we see a strong pickup in final demand - something that is not yet in evidence. Therein lies the danger of using the common trailing PE to assess whether stocks are cheap: there is a risk that the current trailing PE is artificially low, due to unsustainably high profit margins.

Shiller PE/CAPE

source: http://www.multpl.com/

The one metric in the table above that does make stocks appear somewhat expensive is the Shiller PE (also known as Cyclically Adjusted PE, or CAPE). The current CAPE is approximately 24, on the high side of average to be sure, but then again well below its historical high.

Instead of using last years' earnings to calculate the PE, the CAPE uses the average of the last 10 years' annual earnings adjusted for inflation. The proponents of the CAPE argue that its main advantage is that it eliminates fluctuations in the PE ratio caused by variations in profit margins throughout the business cycle. Proponents of the CAPE also endorse its use because:
1) It's less volatile
2) It has a strong tendency to mean revert over time, unlike PE's calculated using one year of earnings
3) It tends to do a pretty good job of predicting future expected stock market returns (John Hussman illustrates this better than most. See also John Mauldin's recent letter Can It Get Any Better Than This?)

Today, the CAPE has a number of detractors. They argue that the severity of the earnings recession we saw in 2008/09 negatively skews the 10 year average earnings figure used in the calculation, thus artificially inflating the CAPE. It's true that the earnings recession was severe however I'm not so sure that means we can simply pretend it didn't happen. Furthermore, the severity of the earnings recession forced companies to slash staff/cut costs which resulted in a subsequent boom in profit margins. Which brings us back to my previous point, regarding a key shortcoming of using discreet years as the basis of PE calculation.

Does it even matter?

"Not so much" is my simple answer to this question.

I acknowledge that the current level of the CAPE indicates we should expect rather meagre long-term returns from the US equity market. However this does not preclude the market from rising from here - possibly substantially. I am disinclined to get bearish an up-trending market simply because it looks expensive on one valuation method. That said, I do not have a huge amount of implicit trust in buying-and-holding the equity market at these levels of CAPE and remain alert to the possibility in a change in trend.

The Trailing PE has very little in the way of demonstrated predictive value of future returns. And given it is sitting darn close to the long-term average anyway, all it really tells is is that stocks could rise an awful lot or fall an awful lot from here. I am disinclined to get bullish the market simply because a valuation metric with little or no predictive capability in evidence says stocks are approximately fair value.

Valuation rarely has any impact on the tape until and unless it has reached an extreme level. Valuations do not seem to be extreme here which indicates to me markets could move a long way in either direction. To put it another way, valuations can currently be used to justify whatever narrative the author/analyst wishes to fit them to, whether bullish or bearish. Caveat emptor.

Trade well and follow the trend.