How to value a cyclical sector like Industrials?

Hi ar169 -

You should never anchor yourself to any value estimate or price target you derive from a single valuation approach. Instead you should be using several valuation approaches, such as DCF in combination with peer multiples (using several different multiples ranging from EV/EBITDA (TTM), EV-to-forward EBITDA (+1), EV/EBIT (historical and forward EBIT), to Market Cap/NI (historical and forward multiples), and with each valuation model/approach you should look at and apply a range of the peer multiples to the subject company versus a single point estimate. When dealing with industrial conglomerates like GE, sum-of-the-parts should also be considered. As to the issue of valuing cyclical names, select your multiples based on the ranges of trading multiples assigned to the subject company and its peers during historical periods when the economic outlook was similar to your perception of the economic outlook today.

Since we're talking about cyclical names here, if you think the economic outlook is strong and consumers will be buying more (i.e. economic growth will be solid) you should use multiples consistent with the company and its competitors' trading during the good times or, said more accurately, the multiples assigned to the stock and its peers during historical periods PERCEIVED TO BE the "good" times by the then-relevant market participants (look at what the sentiment was like historically and how that manifested itself in the relevant trading multiples); and vice versa if you have a more negative outlook for economic growth or industry-specific prospects.

You should always be thinking about how YOUR PERCEPTION OF FUTURE ECONOMIC STRENGTH may differ from what's embedded into the current price - consensus expectations are, at least in theory, what is embedded in price. If you can pinpoint a contrarian view and believe you have good reason for this, your entire edge might come from understanding that the stock's multiple will expand or contract over the coming year or two.

I would also caution against using EPS, period. **Several reasons for this:

1) street analysts base consensus EPS estimates off of management's "adjusted EPS" figures, which are often atrociously misleading as management teams tend to massage Adjusted or "non-GAAP" EPS to put the company in the greatest light possible or to allow for consistency of reported earnings (as the market assigns greater multiples to companies with lower levels of earnings variability, thus if management can artificially "smooth" EPS figures they think they can get a greater multiple). As a result, if you value a stock based on its peer multiples on price-to-earnings (using EPS as "earnings"), both your comparable company multiples will be screwy (due to the inconsistency in their denominators across peers; for instance, one peer may have an honest and straightforward management team that does little manipulation to non-GAAP EPS while another may have a management team that's clever and tricky, switching up its adjustments on EPS each quarter, etc) and so will your PE-based valuation for the subject company, as you likely haven't followed the stock long enough or studied the footnotes and the management team well-enough to fully understand how they're manipulating their own "adjusted EPS" figures. Long story short, unless you have consensus figures and historical trading multiples for your subject company and its peers in which EPS is stated according to GAAP, you are going to have a lot of manipulation going into your peer-trading estimates that ultimately may not hold up in future trading. And even if you had this data, EPS still isn't the greatest due to the issues described next in #2.

2) If you're going to use any earnings-based metric using peer multiples to value a stock, you should use Market Cap/Net Income (GAAP) to evaluate peer multiples and assign a range of multiples (a tight one but still not a point estimate) to the subject company's projected GAAP Net Income RATHER THAN USING EPS. Once you apply the multiple to your proje cted GAAP income for the company in question, look at the # of shares outstanding (diluted) and develop your price target this way. Why? EPS doesn't always reflect true operating performance or earnings or repeatable earnings levels - management teams can and often do use cheap debt and share buybacks to screw around with the denominator (i.e. # of diluted shares outstanding) and make EPS performance look stronger, even when the incremental debt takes the company past its optimal level (thereby increasing the company's true cost of capital in the long-run). Some companies will engage in this behavior, and some will not. This makes it hard to compare properly. Also, companies will use equity (common stock) to fund acquisitions that are highly dilutive to shareholders (and often highly value destructive) and use this to make it hard to decipher a "fair" multiple the company deserves on its reported (GAAP) earnings per share. Non-GAAP EPS is even worse, for the reasons discussed previously.

The way to do it right is to run several different forms of valuation analysis, each of which should in theory lead to a similar enterprise value estimate and ultimately a similar "price target" or PRICE TARGET RANGE (I suggest using a range rather than a point estimate).

If for some reason your valuations result in different price targets (or equity valuation ranges) under some or all of the various valuation approaches and these differences are meaningful, you should investigate: something is wrong with one or all of your models.

Also, with big conglomerate businesses like GE, you probably want to perform a Sum-of-the-Parts (SoTP) valuation as well, given the disparate multiples that are deserved across some of the different GE business lines. But ignoring SoTP for now for simplicity, suppose you value GE using DCF and peer multiples' for EV/EBITDA (use a RANGE OF MULTIPLES) and Forward P/E (always use a RANGE of MULTIPLES and not a single one), you should expect your valuation ranges to be roughly similar across each of your valuation models.

Hope this helps.