Financial Mathematics Text

Sunday, September 1, 2013

Useless Stock Metrics

So today I'm going to discuss metrics used in stock analysis that I think are useless and largely uninformative. In spite of this, many of these are popular and I would like to suggest they shouldn't be popular. (Granted, I think "pop" music shouldn't be popular so what do I know?)

Some of this will be an extension of a conversation at OSV forum (Firm Versus Equity (apples with apples)) as well as a great blog post by Prof Damodaran on the same subject (A tangled web of values: Enterprise value, Firm Value and Market Cap ). 

So without further ado . . .

Price to Sales (P/S)

Price to sales has got to be one of the most useless metrics that is quoted for stocks. It's both looked at on the individual stock level but also looked at, occasionally, for the entire stock market (say the S&P 500).

There are at least two things wrong with this metric.

Problem 1

The first has to do with leverage. I'm going to use an example (which we'll come back to when looking at the other metrics) of a rental property.

Imagine you have a house that's selling for $ \$100 $. It has the capacity to earn $ \$12 $ in rent and there are annual expenses (taxes, maintenance, etc) of about $ \$6 $ for Net Operating Income (NOI) of $ \$12 - \$6 = \$6 $.

Now let's consider two scenarios. First let's suppose you pay cash for the house. What will the price to sales (P/S) ratio be?

The "price" in this case is the value of the equity which happens to be $ \$100 $. The "sales" is the rent that's earned from the property which is $ \$12 $. This gives a P/S:
$$P/S = \frac{\$100}{\$12} = 8.33$$
So far so good. But now let's consider another alternative which I think is more common. Let's suppose that we use some leverage by putting 20% down and taking out a mortgage for the other 80%.

Revenue doesn't change at all; it's still the same $ \$12 $. What's different is the equity which is now only $ \$20 $. This gives us:
$$P/S = \frac{\$20}{\$12} = 1.67$$
The levered house actually looks "cheaper" on the basis of price to sales. This doesn't indicate that it's really cheaper of course. It just an indication that it's a lousy metric.

The problem with sales or rent or revenue is that it's not only produced by the equity assets. All of the assets (in our example, the whole house) are required to produce that revenue (rent). So comparing the revenue to the equity which is just one slice of the assets is a bad comparison.

A better comparison (and this can be seen from Prof Damodaran's table) is Enterprise Value to Revenue.

Problem 2

But the problems with this ratio do not end there. The second problem P/S is well stated by James Montier from his book Value Investing: Tools and Techniques for Intelligent Investment. (You can find the relevant chapter online as a standalone article here.) Here's one relevant issue:
Price-to-sales has always been one of my least favourite valuation measures as it ignores profitability. Reductio ad absurdum demonstrates this clearly. Imagine I set up a business selling £20 notes for £19, strangely enough I will never make any money, my volume may well be enormous, but it will always be profitless. But I won't care as long as the market values me on price-to-sales.
So I think that's another reason to consider this metric to be problematic. (He does suggest it can be used to identify overvalued companies since a company's profits will ultimately be constrained by the amount it sells.)

Verdict on P/S

This one should largely be avoided altogether (with perhaps the exception of identifying overvalued firms). But I think if you're going to use a metric of this sort, it should be Enterprise Value to Sales as this makes a much better comparison. All of the assets (whether financed solely by equity or some combination of equity and debt) contribute to the sales of the firm.

Return on Assets (ROA)

This is a fairly popular metric and I think it should be. After all, return on assets is an attempt to measure the productivity of the firm's assets. What's not to like?

The problem is the way that this is typically defined. Return on assets (ROA) is typically defined as follows:
$$ROA = \frac{\text{Net Income}}{\text{Total Assets}}$$
The problem is that numerator. Net Income (also known as "earnings") tells you how much profit the equity owners earned. We get a similar problem with leverage like we did with P/S. Let's go back to our house example.

In the case of the house, the total assets will be the value of the house: $ \$100 $. But what's "net income"? Net income is actually not a good comparison with Net Operating Income (NOI) since NOI does not subtract out interest or taxes where as net income does. NOI is more akin to Earnings Before Income and Taxes (EBIT).

For simplicity, let's ignore taxes (assume taxes are 0%). What's the net income going to be? Well, it's going to depend on whether or not we took out a mortgage to finance the house. Let's consider the two scenarios we did above.

(1) In the first case we bought the entire house with cash. Since there will be no interest (and we're ignoring taxes), earnings will be NOI. In this case ROA will be:
$$ROA = \frac{\$6}{\$100} = 6\%$$
(2) In the second scenario we assumed 20% down on the house and 80% for the mortgage. Let's assume the mortgage interest comes out to 4%. So total interest expense will be $ 4\% \times \$80 = \$3.20 $. As a result, this is what ROA will be in the levered case:
$$ROA = \frac{\$6 - \$3.20}{\$100} = 2.8\%$$
So from this we can conclude that the unlevered house is more productive than the levered house in spite of the fact that it's the same house!

Why would financing the house with some debt result in an asset becoming less productive?

Verdict on ROA

So I think ROA as it's commonly defined needs to go in the waste bin. But I think the idea of seeing how productive a company's assets are is a very reasonable piece of information to know. As a result, I'd like to suggest a few alternative ways to define it:
$$ROA = \frac{\text{Net Income }+\text{ Interest}}{\text{Total Assets}}$$
$$ROA = \frac{EBIT}{\text{Total Assets}}$$
$$ROA = \frac{EBIT(1-\text{ Tax Rate})}{\text{Total Assets}}$$
I'm not sure which one is better. But I do know that each of these would do a better job of showing how productive the assets are regardless of how they were financed (solely equity or some combination of debt and equity).

Enterprise Value to Free Cash Flow (EV/FCF)

Occasionally I'll see people use one of two enterprise value metrics that I think are pretty useless. One is, as the heading suggests, the ratio of Enterprise Value to Free Cash Flow (EV/FCF). The other ratio I see occasionally used is Enterprise Value to Cash Flow from Operations (EV/CFO).

Enterprise Value represents the total (unlevered) value of all the operating assets. Nonoperating assets are any assets that aren't needed for the basic operations of the business. For example, stocks and bonds owned by a manufacturer are not needed for the operations of a business but they are still part of the total value. (On the other hand, stocks and bonds owned by an insurance company are very much part of their operating assets.)

Enterprise value is a very useful metric to look at. The problem is that, like the two ratios we already looked at, it's a bad comparison. Let's take a closer look.

The way Cash Flow from Operations is calculated is by starting with net income (equity earnings) which doesn't include interest paid to debt holders and then various non-cash related expenses and revenues are backed out of that calculation. As a result, CFO (as it's frequently calculated) is a reflection of cash flows available to equity holders provided we ignore capital expenditures.

There is no clear standard on how Free Cash Flow is to be measured. In fact, depending upon which definition you use, my criticism may not be appropriate. I personally like to consider Free Cash Flow to be an equity figure. So it's similar to CFO except we deduct for various capital expenditures from it.  I then like to use the term Free Cash Flow to the Firm (FCFF) to refer to the cash flows generated by the firm so there's a distinction.

Looking at the Wikipedia article, you can see that some of the calculations include interest while others do not. The fact that the term is not well agreed upon leads to people using it incorrectly.

One could very well fix all of this if US accounting standards treated interest as a financing activity (in the same way that dividends, stock issuance and buyback, and debt issuance and buyback are all financing activities.) Interest is not technically an expense but a payment to one of the owners of the business. But I'm not going to be changing accounting standards any time soon.

Verdict on EV/FCF and EV/CFO

This really all depends on how the terms are defined. If they are defined as equity figures, where the interest is subtracted out, then these metrics have the same problems that P/S and ROA have in that they are comparing equity cash flows to the total value of the operating assets (which includes debt, equity and preferred equity). The result will be that firms that have debt will appear more expensive than they actually are.

On the other hand, if CFO and FCF (or as I like to call it, FCFF) include interest paid to bondholders then these metrics make for useful comparisons.

So I think it's important that the terms get clearly defined and then get matched with the appropriate valuation metric. EV/CFO(equity) and EV/FCF(equity) are useless metrics. EV/CFO(firm) and EV/FCFF can be useful metrics for valuation.


All of the above metrics involve making inappropriate comparisons. While it's OK to compare, say, debt to assets if you want to measure the amount of leverage used, it doesn't make sense to look at one income or cash flow metric and then compare that to an unrelated asset or value metric.

So let's get rid of some of these useless metrics and replace them with more appropriate ones.

No comments:

Post a Comment

Some common OpenID URLs (no change to URL required):