Session 35: Relative Valuation

In session 35 of this 36 session corporate finance class, I'd like to talk about a different way of approaching valuation. Where you try to value an asset based not on its fundamentals, but based on what other people are paying for assets just like it. It's called pricing, or relative valuation. And it turns out to be the way that 90% of what passes for valuation out there is really done. So in these last few sessions, we've laid out and looked at the details of intrinsic valuation. Or in particular, we use discounted cash flow valuation to value the company. But in this session, I want to turn to a different way of thinking about valuation. I'm going to call it relative valuation, but I think a better term for it is pricing. So let me back up and talk a little bit about where I'm going with this concept. In relative valuation, rather than value a company based on its fundamentals, its cash flows, growth, and risk. You value a company by looking at how the market is pricing other companies just like it. So if you break down relative valuation, it has three parts to it. The first part is to find other companies that you claim are just like your company. Tough to do, but that's the first step. The second is you got to standardize prices. What I mean by that is you cannot compare the price per share across companies because in a sense, it's an arbitrary number. Arbitrary, why? Because if I did a stock split, my price per share would halve. So you can't look at the price per share. You have to divide it by some number. And that's what you do when you use a multiple. So you divide price by earnings or price to book, you have a standardized price. So you find other companies like yours, you standardize the price. And then if you're careful, you probably will notice that your company's still different from those other companies in terms of growth and risk and cash flows. You'll have to control for those differences. So find comparables, standardize the price, control for differences. That's the essence of relative valuation. So the key step in this process, of course, is standardizing the price. Let me set up that process. Anytime you look at a multiple, you're going to see a numerator and a denominator, right? And in your numerator, you will see one of three numbers. You will see the market value of equity or market cap. You will see the market value of the firm, which is market value of equity plus market value of debt. Or you will see enterprise value, which is market value of equity plus market value of debt minus cash. So in the numerator, you should always see some measure of market value. In the denominator, here are your choices. You could divide that market value number by a revenue, which is the top line in your income statement. And you could divide market value of equity by revenues or the market value of the entire business by revenues. So that's your first choice. You could use revenues. The advantage of using revenues, of course, it's always a positive number, and you're always going to be able to compute the multiple. You could even use drivers of revenues. What do I mean by drivers of revenues? If you're a cable company, your revenues are driven by the number of cable subscribers you have. So I could divide your market value by the number of subscribers. So the first set of divisors are revenue-based divisors. You could divide by some measure of earnings, right? You could divide by net income if you're looking at equity income, operating income, or some measure of operating income if you're looking at the entire enterprise value. So you could divide by earnings. You could divide by some measure of cash flow. Again, that cash flow can either be to equity investors, where you add back depreciation and net income, or even use free cash flow to equity, or it can be to the entire business. We talked about the free cash flow to the firm, or it could be some variant thereof. It could even be a measure that's cruder. Earnings before interest taxes, depreciation, and amortization is a measure of overall operating cash flow. So you can divide by revenues, you can divide by earnings, you can divide by cash flow, or you can divide by book value. The book value is, after all, what the accountant tells you, what your equity is worth and what your business is worth. You could check to see what that number is relative to your market value. So the key part of getting a multiple is recognizing that you have multiple choices in both the numerator and the denominator. Now, I prefer to do intrinsic valuation, but I'm not dead set against relative valuation. In fact, I think it's a very useful tool. However, when you use multiples, there's a four-step process I would suggest you go through to make sure you don't get yourself into trouble. I'm going to start off by defining the multiple, passing it through a couple of tests before I use it, and I'll take you through those tests. Then I'm going to describe the multiple. Sounds fancy, right? But I'm going to go back to basic statistics. I'm going to look at the distribution, not just the average, but the entire distribution. What's the median, what's the high number, what's the low number? That's useful to know. Then I'm going to analyze the multiple. In other words, I'm going to take the multiple apart to see what variables drive that multiple. Sounds difficult, but it's actually relatively simple to do if you've understood discounted cash flow valuation. And only then am I going to apply the multiple. In my view, the problem with multiples is not that people use them, but that they're in a hurry to use them. They're in such a hurry to get to applying the multiple, they don't stop and define, describe, or analyze. So let's take the first of these tests. The definitional test. Whenever I look at a multiple, the first question I ask is, is this multiple consistently defined? Let me explain. As you just saw, every multiple is a numerator and a denominator, right? Here's the simple rule. If your numerator is an equity value, your denominator has to be an equity value as well. If your numerator is a firm or an enterprise value, your denominator should be a firm or an enterprise value as well. Let's go back and look at some of our choices. Remember, you could use price or market capitalization as your numerator. If your numerator is an equity value, your denominator has to be equity earnings, net income or earnings per share. If your numerator is enterprise value, which is market value of equity plus debt minus cash, or market value of the operating assets, your denominator has to be some measure of operating earnings, cash flow, or book value. So numerators and denominators have to be matched up. The second question I'm going to ask is, is this multiple uniformly estimated? You're going to be comparing your price earnings ratio or EBITDA across 15 companies, right? If you're going to be doing that, make sure you're estimating it exactly the same way for all 15 companies. That's a tougher test to pass than you might think. With price earnings ratios, you know what I'm asking you? Are you estimating earnings per share exactly the same way for all 15 companies? Now you might say the accounting standards are the same, but we know that the accounting standards can be the same and you can have different degrees of fidelity to those standards, right? If you're not careful, conservative companies are going to look expensive and aggressive companies are going to look cheap. But conservative and aggressive are being used in the context of how they measure earnings. So that's the first stop, is the definitional stop. Let's try this out on a couple of very widely used models. Take the P-E ratio. Numerator is usually the current stock price. In the denominator is where you see the action though. I can divide the price per share by earnings per share in the most recent fiscal year, that's called current P-E. I can divide earnings per share in the most recent 12 months, that's called the trailing P-E. I can divide price per earnings per share that I expect in the next 12 months, which is forward P-E. Or I can even divide price per share by expected earnings per share in the year 2016, 2018, or 2020, which would be really, really forward P-E. Now, why would I do this? Well, it depends on what I'm trying to show. I could divide price by earnings per share before extraordinary items or after extraordinary items. Primary earnings per share, partially diluted earnings per share, fully diluted earnings per share. In other words, I can take the same company and compute 15, 20, or even 30 different P-E ratios at any given point in time. You might wonder, so what? Well, I'm going to get numbers that are low from some measures of P-E and high from others, right? Which one will I use? It depends on what story I'm trying to tell you. In other words, if there is an agenda, the multiple that you're going to pick or the variant that you're going to pick of P-E ratio is going to fit that agenda. Remember that the next time somebody tries to use a P-E ratio on you. Now, let's look at the second multiple. It's enterprise value divided by EBITDA, or EB to EBITDA. Perhaps the multiple that's grown the most in usage over the last three decades. Now, this is a multiple that's widely used in infrastructure, cable. In fact, generically, about 30% of equity research is now based on EB to EBITDA. And one of the questions that's worth thinking about is why we net cash out of the numerator? The answer is actually surprisingly simple. Remember we talked about the consistency rule? We're netting cash out of the numerator because the interest income from cash is not part of EBITDA. And if you use that rule, think of what else you should be netting out of the numerator. Remember we talked about cross holdings in other companies, 5%, 10%, 15% of other companies? Those are minority holdings. The income from those cross holdings don't shop in EBITDA either. So if you're computing an EB to EBITDA for a company like Tata Motors, you have to net out the market value of those cross holdings in other companies because the income from those holdings are not part of EBITDA. If you have majority holdings in another company, you have a different problem. Your enterprise value, the market value of equity comes, the market knows you own only 60 or 65% of another company, so it reflects only that holding. But everything else in this computation comes from your financial statements, right? And if they're consolidated, you're going to be counting 100% of the subsidiaries EBITDA debt and cash in your analysis. That creates an inconsistency. That's why you'll often see analysts adding minority interest to the numerator. It's a half-baked way of bringing in that 40% that doesn't belong to you to the numerator, but if you don't do it, you've got an inconsistent multiple. So first stop with multiples, make sure they're consistently defined, and make sure you understand how they're being estimated. Second stop, describe the multiple. As I said, this is a statistical step. So start with the standard statistical questions. What's the average? What's the standard deviation? But don't stop there. Look at the entire distribution, and what you're probably going to see is an asymmetric distribution. Sounds fancy, right? But here's what I'm talking about. Usually when you look at a multiple, PE, EV to EBITDA, price to book, it cannot be less than zero. So your distribution is floored at zero, but it can be a high number. I've seen PE ratios of 15,000, 20,000, 25,000 pop up. And because your outliers are all big positive numbers, guess what? Your averages get pulled out by those large positive outliers. That's why the median might make more sense if you're talking about a multiple. Also check to see how many companies you're losing because you cannot compute the multiple. Again, think about PE ratios. When the earnings per share are negative, you cannot compute PE, you lose those companies in your sample. You might say, big deal. At the start of 2014, out of 7,800 US companies, 4,500 had negative earnings. That's a lot of companies you're losing in your sample. And indirectly, you're creating some bias in your sample, right? You're losing the smallest, riskiest, perhaps most troubled companies. That's no way to do a sampling analysis. So ask those questions and try to get them answered because it's easy to create bias in your assessments without even realizing you're doing it. Third stop, check to see what drives the multiple. I'm going to give you a very simple way of doing this in a minute, but here are the two questions you're trying to answer. What are the variables that determine this multiple, P, EV to EBITDA, price to book? And you're going to see that with discounted cash flow valuation, you're going to be able to answer that question relatively simply. You're going to be able to tell me the three or four variables that drive that multiple. Second, you have to tell me as those variables change, how will your multiple change? So let's take as an example, PE ratios. Intuitively, we know that higher growth companies have higher PE ratios than lower growth companies, right? But let's ask a more specific question. If your growth increases 5%, how much will your PE go up? If you're unable to answer that question, you're going to have trouble with multiples. So we have to be able to scale our multiples to the variables we're talking about. So let me give you the little trick that you can use to see what variables drive a multiple. Let's assume I came to you with an equity multiple. Let's take the PE ratio. Here's what I'm going to do. I'm going to go back to the simplest equity valuation model that I can think of, which is a dividend discount model and a stable growth one. In the stable growth dividend discount model, the value of a stock is the expected dividend next year divided by the cost of equity minus the growth rate, right? Now let me do some algebra. Let's assume that I divide both sides of that equation by the earnings per share. I'm allowed to do that. It's still an equation. I divide price by earnings per share and up with the PE ratio on the left-hand side. And when I divide dividends by earnings per share, I get the payout ratio. Ergo, I have the three variables that drive my PE ratio, the payout ratio, cost of equity, and expected growth rate. If you don't trust a company to pay out what it can in dividends, replace the payout ratio with a free cash flow to equity divided by earnings. For a potential payout ratio, you're home free. You can do this with any equity multiple. Go back to a stable growth dividend discount model and derive the variables that drive equity multiples. So price-earnings ratios are going to be driven by growth, payout, and risk, risk of equity. Price-to-book are going to be driven by growth, payout, risk, and return in equity. Price-to-sales ratios are going to be driven by growth, payout, risk, and net margin. So effectively, you're going to be looking at the variables that drive any equity multiple. You say, what if I have a multiple which is an enterprise value or a firm value multiple? Go back to a firm valuation model and keep it simple. Remember, stable growth, firm valuation model is expected free cash flow to the firm next year divided by cost of capital minus growth. You can use that to derive the variables that drive EV to operating income, EV to EBITDA, EV to invested capital. In other words, any enterprise value multiple, you can compute using the fundamentals. And here's a scaling trick you can use. When you're talking about enterprise value multiples, here's all you have to do. Wherever you're talking about payout ratio, retention ratio, talk about reinvestment rates or one minus the reinvestment rate. Wherever you're talking about growth, replace growth in earnings per share with growth in operating income. Wherever you're talking about cost of equity, talk about cost of capital. And wherever you're talking about return in equity, talk about return in capital. In fact, if you want to talk about margins in an enterprise value context, talk about operating margins rather than net margins. It's the same trick, just a different lens through which you look at the numbers. Now that we've defined, described, and analyzed, now we can apply. And here you have two questions you're going to have to deal with. First is, what are your comparable firms? Now I know the lazy way of answering that is to say, look, if I know what sector my company is in, I'm just going to look at other companies in that sector. But is that really a good measure of comparable? Microsoft is a software company, but is there any other software company in the world that's even remotely similar to Microsoft? From a valuation perspective, a comparable firm is one with similar cash flows, similar growth, and similar risk. Nowhere in there is there a sector component. Now, once you've got those companies together, though, that you call comparable, remember, no matter how careful you've been, there'll be differences between your company and those companies. And you have to find creative ways or not so creative ways of controlling for those differences. So let's try a couple of examples. I first looked at Disney relative to other entertainment companies. Again, I can't compare prices, but I can compare multiples. So we look at what Disney is trading at in terms of PE ratio, current trailing or forward PE. It looks very close to what the sector is trading at. Now here's a little trick I tried. Because these companies have very different growth rates, and I know that higher growth companies have higher PE ratios, I divided the PE ratio by the growth rate. I get what's called a PEG ratio. What does a PEG ratio tell me? It gives me the payoff I'm getting in PE for every unit of growth. The lower this number, the cheaper a company. If you look at the PEG ratio for Disney, it is much higher than the PEG ratio for the sector. I'm not willing to argue that based on this, Disney's overvalued, but on a relative valuation basis, I'm getting the conclusion that Disney is definitely not cheap and perhaps overpriced, given how other entertainment companies are being priced. Now let's try a second relative valuation. This is a relative valuation of Deutsche Bank. And when I looked at Deutsche Bank in November of 2013, it was trading at 67% of book value. There are rules of thumb out there. For instance, a widely used rule of thumb is if a company trades at less than book value, it's cheap. So using that rule of thumb, you'd conclude that Deutsche Bank is cheap, right? But before you do that, take a look at Deutsche Bank's return on equity. While Deutsche Bank looks cheap relative to the sector or even compared to the rule of thumb, it has a terrible return on equity. And remember what we said? Lower return on equity companies will tend to have lower price to book ratios. It's one of those variables that drives price to book. I need to control for differences in return on equity, right? Now here's a trick you can pull out of your statistics book. I went in and opened up to my multiple regression chapter. And remember, in multiple regression, you're trying to explain a dependent variable and independent variables. The dependent variable I have here is price to book. And I chose to use two independent variables relating to banks to try to explain the price to book. The first was the return equity, my hypothesis being the higher the return equity, the higher the price to book. And second, to measure the risk in a bank rather than look at the beta or the standard deviation, I looked at the bank's tier one capital, capital ratio. See, what does that measure? The higher this ratio, the more regulatory capital a bank has, the safer it is. So here again, I'd expect the higher this ratio, the higher the price to book. So I ran the regression across banks. The R squared is not bad. It's not great, but it's not bad, 24%. But the coefficients all have the right signs. Higher return equity companies have higher price to book. Higher tier one capital ratio companies have higher price to book. You notice the number in brackets below the coefficients? Those are my T statistics. And they're telling me that this is a statistically significant relationship. And I can use it in a prediction. I plug in the numbers for Deutsche into that regression. The tier one capital ratio for Deutsche as well, the 15.13% tier one capital ratio and the existing return in equity. I get a predicted price to book ratio for Deutsche 1.03. You're saying, what does that tell me? Given Deutsche's return in equity and tier one capital ratio now, and given how other banks are being priced, I'd expect it to trade at 1.03 times book value. It's actually trading at 0.67 times book value. That would make it significantly undervalued in my book. So if you're going to do relative valuation, be creative about it. Don't just compare your PE ratio to the average for the sector. Look at the data. Look across the sector. Don't throw away information. Draw on your statistical tools to see if you can make better judgments about why differences in companies translate into differences in multiples. And if you do that, multiples are a very good tool to have in your valuation arsenal. Thank you very much for listening.