Waiting For Fair Value

The idea behind value investing is that you buy a stock well below its fair value, wait for it to appreciate to something close to its fair value, and then sell it.

The natural questions, then, are:

• How long does this process take?
• How often does it actually happen?
• Why does it happen?

Is it possible that this whole idea doesn’t actually work? Is it possible that if you buy a stock well under its fair value you have no better chance of it reaching fair value than if you assigned a more or less random price target to a random stock and held it until it reached that target?

I decided to study stocks listed on the major US exchanges over the last 22 years to try to get an answer to these questions. To be honest, at the outset of my study I had no idea what the answer would be.

Assigning a Fair Value to a Stock

As I’ve written, calculating an intrinsic value for a stock is a process fraught with difficulties, and I believe it cannot be done using an algorithm. It is a necessarily discretionary process. The most recent attempt I made at an algorithmic method relied on backtests I’ve done on the last twenty years of data, so that algorithm cannot be properly tested on that same data.

Instead, I’ve come up with a formula that assigns a fair value to a company based on comparing seven measures to those of the average comparable company: its sales, income, shareholder payout, tangible book value, research expenditures, EBITDA, and unlevered free cash flow.

Methodology

Here is my methodology. It gets a bit technical, so you can skip it if you’d like.

I start by using Portfolio123 to come up with aggregate ratios for each item. I’ll give an example with net income. I look at the current fiscal year’s mean EPS estimate for each company in a GICS subsector and multiply it by the number of fully diluted shares; if a stock has no EPS estimate, I’ll use its trailing-twelve-month net income instead. I then divide it by its market cap. I then exclude the top and bottom 16.5% of the results and all zero or negative results. I take a cap-weighted average of the rest. I’ll call this the aggregate subsector earnings yield.

Then I take the company’s estimated income (or actual income if there’s no estimate) and divide it by the aggregate subsector earnings yield for its subsector to arrive at what I’ll call an income-based fair value.

For example, let's take a large cap stock which I've selected randomly, Newmont (NEM). It has an estimated net income of \$3.05 billion this fiscal year, according to analysts. The earnings yield of its subsector (which in this case is the same as its sector, materials) is 5.6%. Divide \$3.05 billion by 5.6% and you get an income-based fair value of \$54.5 billion, which isn’t that far from its market cap (\$45.4 billion).

I do the same thing for the company’s sales. For the shareholder yield, I follow a similar course, using total equity purchased minus total equity issued plus total dividends paid, multiplying the latter by the indicated annual dividend divided by the dividend per share of the most recent fiscal year (all numbers here are for the most recent fiscal year). For the tangible book value, I simply use that item based on the most recent quarter’s balance sheet, without relying on estimates. For the research-based fair value, I simply make that zero if the aggregate for the subsector is zero.

For EBITDA and unlevered cash flow, I also privilege estimates, but divide by enterprise value rather than market cap and use the entire universe of non-OTC stocks rather than the subsector to calculate the aggregate; I then subtract from the result the company’s total debt, noncontrolling interest, and preferred equity, and add its cash and equivalents, to get back to its market-cap based fair value.