Session 10: Estimating Hurdle Rates - Bottom up Betas

In this session 10 of a 36 session corporate finance class, I'd like to offer you a better way of estimating betas. I call this a bottom up beta, but I'm going to start with the business of businesses a company is in and build up to a beta that is more robust and better estimate of what's coming in the future. In the last session, we laid the foundations for thinking about betas. We argued that while you might get a beta from a regression, Your beta as a company is determined by choices you make about what business to be in, what your fixed cost structure looks like, and how much you borrow. In this session, I'd like to take a pragmatic step towards estimating betas, which in a sense will allow you to bypass the regression beta process. To understand this, let's go back again. We're trying to estimate the hurdle rate for a company, right? We talked about risk free rates, we talked about equity risk premiums. We're now in the process of talking about betas. We've laid the foundations for betas by both looking at regression betas and the determinants of betas. Now let's talk about a process for estimating betas that's better than a regression beta process. To understand what we're going to do, let me lay out a property that betas have that make them incredibly easy to work with. Betas are always weighted averages. You're saying, what are you talking about? What does that even mean? Let's take an example. Let's assume you have a mutual fund, Fidelity Magellan, huge mutual fund. Hundreds of stocks in your portfolio. And you ask me what the beta of your portfolio is. It's very simple. It's a weighted average of the betas of the individual stocks in your portfolio in proportion to how much money you've invested in each one. That's true whether you have five stocks, 50 stocks, or 500 stocks in your portfolio. The beta for a mutual fund is a weighted average of the beta of the investments in that fund. But here's the property we're going to use in corporate finance. Remember we talked about GE, 25 to 30 different businesses? The beta for GE as a company is a weighted average of the betas of each of those businesses. If we could somehow get betas by business and attach weights to the business, we can get a beta for a company by weighting the betas of the individual businesses. I call this a bottom-up beta. But don't get too caught up in terminology. You can call it whatever you want. But here's how the process works. Let's assume you have a company. Let's assume a company is in two businesses, steel and chemicals. Here's the way I'm going to estimate betas. I'm going to go find as many publicly traded steel companies as I can and as many publicly traded chemical companies as I can. Why publicly traded? Because I can get regression betas for each of them. Then I'm going to average the betas across all steel companies and average the betas across all chemical companies. So I'll get an average regression beta for steel companies, an average regression beta for chemical companies, right? That regression beta is a levered beta. So I'm going to take an average debt-to-equity ratio for steel companies and unlever the beta for steel companies. An average debt-to-equity ratio for chemical companies, unlever the beta for chemical companies. I'm going to get an unlevered beta for being in the steel business and unlevered beta for being in the chemical business. Some people like to call these pure play betas to reflect the betas for the businesses. Then I come back to you. So you told me you're in two businesses. Tell me how much value you get from each business. You might say 70% from steel, 30% from chemicals. In fact, you might say you don't know the values, that all you can give me are revenues. I'm okay with that. Let me start with that. So 70% steel, 30% chemicals, I get a weighted average. That's the unlevered beta for your company. Final step, ask you what your debt-to-equity ratio is. You can give me your actual debt-to-equity, a target debt-to-equity, or some other number in the middle. I'll tell you what the beta for your stock should be. Now, why am I doing this? Because I don't like regression betas, right? But where did I get the betas for the steel companies? From running regressions. So you're saying, what's the gain? What's the advantage of using an average of 100 regression betas, rather than using one regression beta? Remember the law of large numbers in statistics? What I read when I see the law of large numbers, is the average of 100 bad numbers can be a really good number. An average of 100 bad regression betas can actually give you a good beta for a business. So one reason I prefer bottom-up betas is they're more precise. The second reason is I get to set the weights. Put differently, if you entered the chemical business yesterday, there is zero chance that a regression beta for your company would reflect it. But I set the weights. In fact, I can be proactive. If you tell me you're gonna be in the technology business tomorrow, I can bring in the weight for the technology business and a beta for a technology business. With bottom-up betas, I regain control of the process. I'm no longer at the mercy of a single slice of history and a single regression. So let's try this for Disney. One of the advantages of applying this approach to Disney is Disney gives you a fair amount of detail of the businesses they're in. And Disney breaks themselves down into five businesses. And these are the five businesses they describe themselves as being in. The first is broadcasting. That includes Disney Cable, it includes ABC, and of course, it includes ESPN. Then the second business is the theme park business. Disney World, Disneyland, Euro Disney, Tokyo Disney, and the very beginnings of Hong Kong Disney in 2013. The third business is the movie business, which for Disney stretches across multiple entities. It includes Disney animated studios, the old Disney, the part of Disney. It also includes Pixar, Marvel, and LucasArts. Very different entities, but they're all under the movie business. The fourth business that Disney's in is the consumer product business. The licensing revenues they get when they license out characters for people to make toys and other characters out of, and maybe even games or software. So that's in the consumer product business. And finally, there's interactive gaming, which is Disney's newest business. It's an attempt to come up with apps and other games for computers, smartphones, and other devices. Five different businesses, and Disney actually gives me a fair amount of detail. It gives me revenues and income, depreciation, capital expenditures, even identifiable assets by business. And I'm gonna use this information to try to come up with a bottom up beta for Disney. So here's the first step. I need to come up with betas for each of the five businesses, right? So I'm gonna take an example. In fact, I'm gonna take the business where the easiest time coming up with the beta, and then talk about how I came up with the betas for the others later. Let me take the movie business, that's third on the list. I wanted a beta for Disney's movie business. So I went looking for publicly traded movie companies. So I'm gonna go to the next page, but I'll return back to this page. But here's the movie business in the US. I found ten publicly traded US companies, which are movie companies. You can see that they're very different market capitalizations. They range from very small. Odyssey, for instance, is a tiny company, to 21st Century Fox, which is huge. But in a sense, the market caps don't bother me. These are all publicly traded, so I have regression betas for each of the companies. I can also get their market value. In addition to market cap, I can get the debt and cash, which allows me to compute both the debt to equity ratio for each company. And I need market values for those. And how much cash each company has as a percentage of firm value. Firm value is market value of equity plus debt. The enterprise value column that you see there is market value of equity plus debt minus cash. I'm going to keep track of both these numbers. Here's how I'm going to come up with the beta for the movie business. I take the average regression beta across these movie companies. Number I get is about 1.24. Actually, let me take that back. The average numbers that you get across these companies is skewed by outliers. There are a few extreme numbers, not in the beta column, but in the debt to equity. So I'm going to use the median numbers. It allows me to kind of get rid of the outliers. The median regression beta across these companies is 1.24. First step is I unlever the beta. The median gross debt to equity, total debt over equity for these companies, about 27%, I unlever the beta and I come up with about 1.066 as my unlevered beta. I'm almost home. But here's what I notice. These companies have cash of about 3% of their overall value. So one way to think about this is the unlevered beta for the company is 1.066, but 3% of these companies is cash. And if I want an unlevered beta for just the movie business, I should be dividing that 1.066 by the 97% that's just movies. That gives me my unlevered beta for the movie business. It's about 1.099. That's the process by which I'm going to get unlevered betas. And while I will not go through the process in detail for the other businesses, I did exactly what I did for the movie business for the other businesses as well. But there were some issues I had to deal with in terms of sample size with the other businesses. And let me talk about some of the ways I got around the issues of small sample size. Take the broadcasting business. There aren't too many publicly traded big broadcasting companies left in the US. They've either been assimilated into other companies or they don't trade on their own. So here's how I expanded my sample. I look for other companies that generated their value from the broadcasting business. A couple of examples. I use Nielsen, a company that rates TV shows as part of the broadcasting business because it derives its revenues from the broadcasting business. I use Kingworth, a company that syndicates TV shows and sells them to broadcasting companies as being in the broadcasting business. I ended up with about 26 companies in my sample. And that's a pretty good sample. I'm okay with that. And I use that to come up with the unlevered beta for the broadcasting business. In the theme park business, initially, I stayed US focused and I found only two publicly traded theme park companies, Six Flags and a Midwestern company called Cedar Point. But basically, these are only two companies. I need a much larger sample. So to expand my sample, I went global. I looked at theme park companies in Europe and Asia. They said, you can't do that. Why not? Beta is standardized around one. There's no reason why I need to stay US centric. I ended up with about 20 companies and an unlevered beta for being in the theme park business, consumer products and interactive gaming at a much easier time. At the end of the process, though, if you look at the very last column of the unlevered betas for each of the five businesses, first part of the mission is complete. Let's move to the second step. Oh, before I go to the second step, one more pause. Remember the studio business and the way I came up with an unlevered beta. Here's another way you can think about what I'm doing. Think about a balance sheet, right? You've got cash and the movie business on one side. You've got debt and equity in the other. This is the different way I have of working through how I ended up with an unlevered beta for the movie business. I start with the equity beta, which is what a regression beta is. It's 1.24. I use the debt to equity ratio to come up with an unlevered beta for the entire company. That's a 1.066. Then I look at how much cash the company has as a percentage of value, 3%. And that's how I come up with the unlevered beta for the movie business. It's a useful exercise to do this at least a few times before you get comfortable with unlevered betas for companies and unlevered betas for businesses, because it's easy to get confused among the three concepts. Now that we have the unlevered betas for the five businesses, let's start weighting. To get the weights, I initially started with revenues, but I did not stop there. I could use revenue weights, but then I'm assuming a dollar in revenue in each business is roughly of, say, of similar value. Instead of doing that, I went back to those publicly traded companies from which I got the betas. So the movie companies, for instance, I had their enterprise value for each company, right? Remember the market value by equity plus debt minus cash? I divided that enterprise value by revenues for each movie company, and I came up with the median value across movie companies. I applied that multiple to the revenues that Disney got from the movie business to get an estimated value for Disney's movie business. See what I'm trying to do? I'm trying to convert revenues into value, and I'm using that EV to sales, enterprise value to sales ratio from comparable companies to estimate the value. Once I make those estimated values, I use those estimated values to get my weights, because ideally, I'd like these weights to be value weights. Based on my estimates, the biggest part of Disney right now is its broadcasting business. The next biggest is the theme park business. The movie business is only about 13%, and the consumer products is about 7%, and everything else is 1%. This company is made up of five businesses, but some businesses are bigger than others. I take a weighted average of those unlevered betas, and the weighted average I get of 0.9239 is my unlevered beta for Disney's operating assets. Now, Disney does have some cash, about 3.9 billion. That cash has a beta of zero. I could compute a beta for Disney as a company, an unlevered beta for Disney as a company, and that's what the beta of 0.8978 at the bottom tells you. That's the beta for Disney as a company, but Disney as a company also incorporates cash. I'm actually going to use the operating asset beta 0.2939 more than I use the unlevered beta for Disney as a company, but I thought it'd be useful to compute both numbers. Final step in the process is I need a levered beta, and here I ran into a bit of an issue. I need a debt-to-equity ratio by business, and Disney does not borrow by business. It borrows as a company. In fact, it has about $15.96 billion worth of debt. Now, I could assume that every business has the same debt-to-equity ratio, but rather than do that, I decided to try to allocate the debt that Disney has across the businesses based not on their market values, but based upon their identifiable assets. Why that? I thought that made more sense than using market cap or cash flows or earnings. I mean, you might choose a different proxy, but I allocated the debt across the divisions. That gives me different debt-to-equity ratios for each business, and those debt-to-equity ratios are what I use to come up with a levered beta for each of the five businesses, and you can see how different they are. The levered beta for Disney as a company is close to one, but they're much higher for some businesses and much lower for others. The theme park business has a much lower beta than the broadcasting business does. The cost of equity that come out of this are very different across the businesses as well. The cost of equity for Disney as a company is only about 8.5%. For the broadcasting business, it's much higher. For the theme park business, it's much lower. Now, here's why it matters. Disney's a multi-business company. When it looks at an investment, that investment doesn't cut across businesses. It's often in a single business. So let's say that you're looking at a movie project. I don't care, you can name the movie. Let's assume this movie project is expected to deliver a return on equity of 9.5%. See why I picked the 9.5%? It's higher than the cost of equity for Disney as a company, but lower than the cost of equity for the broadcasting business. The question I would like you to think about is would you accept this investment? Put differently, what should you compare this cost of equity or this return on equity to? Should you compare to the cost of equity for the company taking the project? Because after all, it comes up with the money. Or should you compare to the cost of equity the business the project is in? Logically, I think it should really be the business. And here's why. If you use the company's cost of equity across the board, here's what you're going to end up doing. You're going to end up subsidizing your riskier businesses with your safer businesses. And over time, your safer businesses are going to shrink and you're going to become a riskier company. The right thing to do at a company, if you're a multi-business company, is to come up with different hurdle rates for different businesses. I'll also have to tell you that most companies don't do this. Part of the reason for that is if your only way of getting betas is a regression beta, you get only one beta for a company, even one as large as GE. You have no way of estimating betas by business. This is one of the bonuses of using a bottom-up beta approach is you can estimate a cost of equity by business and use that cost of equity to make a judgment on whether that business is a good business or not. That's pretty much what I want to say about bottom-up betas, but we'll come back to this in the next session. Thank you very much for listening.