## Waiting For Fair Value

### The Mean Regression Puzzle

Now it’s entirely possible that what we saw with fair value is simple mean regression. By this, I mean that as long as one can come up with a “fair value” for a stock that is based on an average of some kind, then, by the strict and immutable laws of probability, stocks that are very cheap or very expensive compared to the average will be more likely to have average prices later. This has nothing to do with how much a stock is actually worth.

If, on the other hand, our measure of “fair value” is a good indication of how much a stock is worth, then a measure of “fair value” that is simply based on mean regression would produce a distinctly worse outcome.

### The Fundamental Difference

So here's how I came up with a mean-regression-based model.

Most fundamentals of a company are positively correlated with its market value. The reason the seven components of my fair value have been chosen by value investors over the years is not just that they’re correlated with market value but that they also make a positive contribution to it. If, for example, one had chosen total liabilities instead of book value or cost of goods sold instead of R&D, one would have favored companies that are likely to do poorly rather than companies that are likely to do well.

One can classify fundamentals into roughly four groups:

1. Positive fundamentals. These are the seven fundamentals I chose to base fair value on (income, sales, EBITDA, free cash flow, R&D, shareholder payout, and tangible book value): when a company’s positive fundamentals go up, it’s taken as a good sign.
2. Negative fundamentals. These are things like cost of goods, liabilities, debt, and inventory: when a company reduces its negative fundamentals, that’s taken as a good sign, and if you compare two companies with the same market cap, you might favor the company that reports lower figures for these items.
3. Neutral fundamentals. These are things like total assets, number of employees, SG&A expenditures, PP&E, depreciation and amortization, and so on. All of these correlate roughly with a company’s market cap but they are not normally used for valuation purposes; decreases or increases in these fundamentals are not necessarily good or bad, and if two companies with the same market cap had different levels of one or more of these items, it would not be considered a bullish or bearish sign.
4. Uncorrelated fundamentals. In this category would be all ratios, since ratios have no correlation with market cap; also included would be items that only appear sporadically on a company’s statements and items that vary wildly from company to company or from statement to statement.

In order to determine if value really is a significant improvement on mean regression, I have come up with a measure that we can use as a base rate: a fair value measure that is based on neutral fundamentals rather than positive ones. We’ll call it, for the sake of jest, the company’s “unfair value.” Like the fair value measure, it has seven components: total assets, number of employees, SG&A expenditures, gross PP&E, depreciation and amortization, working capital, and intangibles. I’ll compare total assets and working capital to enterprise value and the rest to market cap, using exactly the same sorts of calculations as I made for the seven components of the fair value estimate (for firms in the financial sector, I’ll use only assets to market cap, employees, and SG&A). Please remember that these are ratios that, as far as I know, nobody actually uses because they make little financial sense. Once again, I’ll buy companies at one-third of their unfair value and sell them when they reach their unfair value.