Session 9: Estimating Hurdle Rates - Betas and Fundamentals

In this session nine of a 36 session corporate finance class, we're going to talk about better ways of estimating betas. In particular, we're going to look at the fundamental decisions that a company might make that can affect its beta. What type of business to be in, how to run that business, and how much it borrows to fund that business. In the last session, we talked about a regression beta. In particular, we took Disney apart and we looked at what a regression tells us about Disney. But I think I made clear that I was a little skeptical about using a regression to get a beta. And in this session, I want to build on that concept. If you look at where we are in the picture, we can use a regression beta to get an expected return. But that regression beta is a noisy estimate. Noisy in what sense? Remember that standard error in the beta gave you a range on the beta? In fact, I'll give you a very quick review of regression beta mechanics. Let me show you the regression beta that I estimated for Tata Motors. Just like Disney, I used Bloomberg beta page to come up with a regression beta for Tata Motors. If you trust this regression, the key word is trust, the beta for Tata Motors is 1.83. But there's a pretty hefty range around that beta, because there's a standard error on that beta estimate, which makes the true beta be any number from 1.5 almost up to 2.2. That's a pretty big range for a beta. And if you look at the R squared and the other statistics, they're all very dependent on the fact that I've run one slice of history, one regression, and I'm building off that regression. I'd like to show you how much of a difference it can make as to what you choose your index to be. I took Vale, which is a Brazilian company, and I ran two different regressions. One was Vale against the Bovespa. This is the local listing against the Brazilian equity index. The other is Vale's ADR, which is a US listing against the S&P 500. I get very different estimates of beta intercept and R squared. And you know what? I don't trust either of these numbers. In fact, building on this concept, let me take Deutsche Bank and Baidu and show you how different the betas for those companies can be, depending on the choices I make. With Deutsche, if I run the regression against the DAX, which is just the German equity index, I get a much higher R squared and a different beta than if I run it against the FTSE, the European index. Lower R squared, different beta. With Baidu, I get very different betas if I use the S&P 500 as opposed to the NASDAQ. What I'm trying to argue for here is when you see a regression beta page, that might look like a fact, but there are lots of choices that went into that page. In fact, I tell people that if they give me access to a Bloomberg terminal, about 30 minutes, and tell me what beta they want for their company, I will find a way to deliver a printed regression beta page that backs up that number. I just don't trust regression betas. They're statistical numbers, and they're noisy ones at that. So here's what I'd like to do. I'd like to go back to basics. I'd like to think about the economic fundamentals that drive betas. What are the choices that you make as a company that determine your beta? To understand this, I'm going to show you a few betas. And with each one, I'm going to give you a story. So these are companies, I looked up their betas, and I'm going to try to give you an economic story for a beta rather than tell you they came from a regression. Look towards the top of the page, you see Bulgari, right? Very high beta, why? I think it's because it produces a very discretionary product, a luxury product. What's wrong with it? Nothing wrong with it. In good times, you sell a lot. In bad times, you don't sell very much. Your regression beta reflects the fact that because your product or service is discretionary, your profits, your revenues, will also be much more volatile, and your beta reflects it. Keep going down the line. See Quest Communications, used to be a telecom company. Why do telecom companies have high betas? Part of it might be their fixed cost structure. They have lots of fixed costs. And part of it is self-imposed. They borrow a lot of money, which creates a fixed cost they have to make in good times and bad times. It makes their equity earnings much more volatile. It pushes up their beta. Keep going down the list. You see Microsoft, right? Beta 1.25. When I first started tracking Microsoft in 1986, right after their IPO, Microsoft used to have a beta in excess of two. As it's become larger, more diversified, and accumulated cash, its betas tended to drift down. Keep going. You see beta for GE, 1.10. GE is perhaps the most diversified large US company you will find, a true conglomerate, probably in 25 to 30 different businesses. You think, so what? As companies get more diversified, their betas should tend to move towards one. In fact, the only reason GE's beta stays above one is the largest single piece of GE is GE Capital. And GE Capital has a lot of debt. Keep going. You're going to see Exxon Mobil. Pretty low beta, right? You're surprised. You're saying oil prices, they're pretty volatile. Exxon Mobil is exposed to oil price risk. If oil prices go up, Exxon Mobil's stock price tends to go up. If oil prices go down, Exxon Mobil's stock price tends to go down. But if your biggest macro risk, in this case oil price risk, moves in the other direction for the rest of the market, you can see why Exxon Mobil will end up with a lower beta. In fact, if your biggest source of risk actually cuts in the opposite direction, and other companies are hurt by it, you could actually end up with a negative beta. That's actually the lowest company on that list, Harmony Gold Mines, actually has a negative beta. That's mind boggling, right? Because think about it, if you have a negative beta, your expected return is actually going to be less than your risk free rate. You say, why would I do that? Why would I invest in something that earns less than the risk free rate? You're effectively buying insurance. That's what a negative beta tells me. A negative beta means that there is some macro risk that you act as a hedge against. Gold prices historically have been a good hedge against inflation. Maybe if the rest of your assets are financial assets, you will invest in a gold mining company and settle for less than the risk free rate. The final example there is Altria, or Philip Morris as you used to know it. Why does it have a low beta? Because it's product of services and addiction. If your product or service is an addiction, your revenues are going to be much more stable. If revenues are more stable, your beta is going to reflect it. So if you look at this graph, you can see the fundamentals that drive betas. Let me summarize them because ultimately, there are only three choices you make as a company that determine your beta. Here's the first one. Tell me what you do. As I said, the more cyclical your product or service, the higher your beta should be. Because betas measure how you move with the market. Automobile companies and housing stocks should have much higher betas than food processing companies. I also argue that the more discretionary your product or service, the higher your beta. Because in good times you sell more, in bad times you sell a lot less. So if you're a luxury retailer, Gucci, Tiffany's, your beta should be much higher than if you're a discount retailer. So that's the first choice you make that drives your beta. What kind of business are you in? Here's the second choice you make that drives your beta. Tell me something about your cost structure. The greater the proportion of your costs that are fixed costs, the higher your beta will be as a company. And here again, the intuition is simple. If you have a lot of fixed costs, good times become great times, bad times become awful times, everything gets magnified. Sectors which have high fixed costs, like the airlines, should have high betas. Sectors which have low fixed costs should have much lower betas. And within the same sector, companies that have lower fixed costs, Southwest Airlines among the airlines for instance, should have lower betas than companies with higher fixed costs. So what kind of business are you in? What type of cost structure do you have? Now staying on the cost structure, one of the interesting questions you often face when you look at a company is, how do I measure how much of the costs in my company are fixed costs? Cuz after all, accountants don't break costs down in income statements into fixed and variable costs. They break them down into cost of goods sold and SG&A and depreciation and other expenses. So here's a shortcut that might or might not work for you. If you have lots of fixed costs, small changes in revenues will translate into big changes in income. That's because everything flows through to your bottom line. So if you take your percentage change in operating income, or earnings before interest and taxes, and divide by your percentage change in revenues, over a long enough time period, you should get a measure of operating leverage, and you can compare that to what the sector average looks like. In fact, I tried this for Disney, going back to 1987. And looking at the percentage change in revenues and the percentage change in operating income, going back to 1987, the number I get is 1.01. Basically, operating income is 1.01 times more volatile than revenues. Now, what did I compare it to? I wasn't sure. In fact, I actually computed a second number for Disney post-1996, when they acquired ABC, and the number I got was 1.25. Again, I had no comparison. To get a measure of comparison, I did this for every entertainment company. The average across entertainment companies is 1.35. Now, if Disney's numbers are lower than the sector average, think of what it tells you about Disney's fixed costs. I know what it tells me. Seems to suggest that Disney has less fixed costs than the rest of the sector, and should have a lower beta. So what kind of business are you in? What kind of cost structure do you have? Third and final issue that drives your beta is how much have you borrowed? When you borrow money, you create a fixed cost you did not have until you borrowed the money. In fact, you can measure the effect of debt on betas by defining two different betas. The first beta I'm going to call an unlevered beta. That's a beta of the business you're in. The second beta I'm going to call a levered beta. That's a beta of your equity. And what connects them is how much you've borrowed, which I measure with the debt to equity ratio. It is true that when you have that interest expense, the fixed cost, that you get a tax deduction. So I'm going to net out the tax benefit when I compute the effect on beta. But the levered beta is the beta you observe for your equity, and it can be high even for a safe business if you borrow enough money. Now think back to regression beta. Is regression beta a levered beta or an unlevered beta? Well, we directly did not impute a leverage into it, but it is in fact a levered beta. And the reason it's a levered beta is your debt to equity ratio affects your stock returns. But the debt to equity ratio that's built into your regression beta is the average debt to equity ratio you had during the period of the regression. In the case of Disney, the average debt to equity ratio in the five years of the regression was 19.44%. The regression beta was 1.25%. The marginal tax rate for Disney, and we'll come back and talk a little bit more about that in future sessions, is 36.1%. Here's what I'm going to do. I'm going to take the regression beta, which is a levered beta, and take out the effect of debt. It's called unlevering a beta. And I come up with 1.1119. You say, what does that tell me? That's the beta Disney would have had if they'd chosen not to use any debt in funding their business. It's also the beta of the businesses they're in. It's the unlevered beta for Disney as a company. You say, why do I need that? Well, if I have the unlevered beta, I can tell you what the beta for Disney will be at any debt to equity ratio. That's a pretty useful thing to be able to do. So in this table, for instance, I've computed what the beta for Disney will be at every debt to equity ratio, ranging from 0% to 900%. At 0%, my levered beta is equal to my unlevered beta, because I have no debt. As I go to higher and higher debt to equity ratios, look at how high my beta becomes. You can take any company and move its beta into levels you haven't seen before, 5, 6, 7, if you borrow enough money. That's something to think about the next time you see a leveraged buyer, where the acquirers might take on 80%, 90% of the overall capital in debt. The effect on beta is going to be high, and the cost of equity is going to go up. So in summary, your beta doesn't come from a regression. It comes from choices you make as a company about what business to be in, how much to borrow, and how much your fixed costs are. Once you make those choices, your beta is, in a sense, predestined. Thank you very much for listening.