Okay, so is everybody in a group? Is there anybody who's not in a group yet? Okay.
Now the next question is going to be, right? What's the next question going to be? Have you picked your company? He can teach the class now for me, sir. How many of you have not picked a company yet?
You know the rest are lying, right? If you think you're the only person, but do pick a company soon, right? Because we're into risk-free rates and risk premiums.
So let's do some very quick review. Let's see how much the last two classes have stuck. So I'm going to pick on you. How do you get a risk-free rate to do a valuation?
A risk-free rate. No, no, before we do that, what's the first thing we need to do? Pick the currency to do the valuation, right?
Because you can pick any currency. The first choice is the currency. And once you pick the currency, you have to get a risk-free rate.
Less work if you have a government that is default-free, like the US, Switzerland. more work if it's Indonesia, India, Brazil. You've got to clean up.
So the risk-free rate is going to be driven by the currency choice you make. Okay. I won't pick on you because you've listened to so much of me.
He's been sitting through every class today, so I feel sorry for him. You help me on the equity risk premium. How do you get an equity risk premium for a company? Don't repeat the question back. That means you're thinking where you're using it to kind of hedge here.
Equity risk premium for a company. What do you do? You take the beta.
No, no, we haven't even opened the beta box yet, right? We take where the company does business. That's good.
The equity risk premiums of each part of the world, and we take a weighted average. So the risk-free rate is driven by currency choice. The equity risk premium is driven by where you do business. What's a missing piece? Whatever you've not talked about yet.
We haven't talked about beta, the measure of relative risk. Today's class is going to bring that final piece of the puzzle in. And before I show you the test that we're going to start the class off, I'm going to give you a preview.
You all take in your first finance class, foundations of finance, etc. Can somebody tell me how you were taught to estimate betas? You want to tell me how you were taught to estimate betas, what we asked to do?
You run a regression of what against what? Not expected. Actual returns on your stock versus returns on an index and the slope of the line you are told is the beta. What horrifically bad advice.
Because to begin with, what does it do? It makes beta into a statistical number, right? You run the regression and it comes out of there.
So today I'm going to dig a hole and I'm going to bury regression betas. And I'm going to replace them with what I call bottom-up beta. Sounds fancy, but remember how I got an equity risk premium for your company based on where you did business? I'm going to get a beta for your company based on what businesses you're in.
It's called a bottom-up beta. And in the process, one of the challenges I'm going to face in coming up with a bottom-up beta is coming up with other companies that do the same thing that you do. So kind of a preview, so I want you to think ahead. So when we get to that stage, you have two choices.
You can pick a company and try to find companies just like yours and end up with a small sample of companies. Get their betas and try to get a beta for your company. Or you can pick a much larger sample of companies which are a little different from yours. Do you see what I'm talking about? If you're doing Microsoft, you can say, look, I'm going to pick only big...
Applications software companies, in which case you might get two or three, or you can say I'm going to pick all software companies, accepting that some are going to be bigger, some are going to be smaller, some are going to be higher growth, some are lower growth. What do you think is a better strategy? Go for a small sample of companies very much like yours, and this is going to show up when you talk about geographic choices.
If you pick an Indian steel company, If you say, look, I want to look at only other Indian steel companies, your sample size is much smaller than if you say, I'm going to look at steel companies globally. So a small sample of companies very much like yours or a larger sample of companies that don't look like yours. You think a small sample?
That was an unfair question because... When I talk about bottom-up betas and why they're better, you know what I draw on to argue for replacing a single regression beta with an average beta across companies? It's called the law of large numbers. You know what the law of large numbers is, right?
You take 100 crappy numbers, you put them in a pot, you stir them up, the average is magically precise. I never quite got this in my statistics class, but I think I get it now. Your average is...
What does the law of large numbers require? Large numbers. So guess what?
I'd much rather get a large sample of emerging market steel companies than three Indian steel companies. When you're doing pricing, it's different. When you're trying to come up with a PE ratio for your stock, maybe you want to stay with the small sample. With betas, we're going to go big because we want large samples. And when we talk about comparable firms, at least in the context of measuring risk, what are we talking about?
Are we talking about other companies in the same industry group, the same market, the same market cap, none of the above, all of the above? How should I go looking for comparable? Incidentally before I do that, did you get that, what you needed to do to get S&P Capital IQ?
I sent you a attachment. You look puzzled. I did send you an attachment.
I know I did. records of it. If you get a chance, go open the attachment, because it tells you how you can access Capital IQ. You're saying, what's that got to do with what we're talking about here?
Capital IQ allows you to screen all 43,000 publicly traded companies in the world for whatever you want. you want. So you say, I want all steel companies with a market gap between 100 million and everything is done in US dollar terms, between 100 and 300 million in Asia.
And magically it'll list out every company that it found. So it's a great way. So you don't have to go hand collect this data from Bloomberg or physical. So you're going to be able to do this.
So when Capital IQ opens up and says, what screens do you want me to run? The question I'm asking is, what are you going to screen for? Do you want to look for the same country, same? sector.
You can try all of the above, but I'll tell you from a valuation perspective what a comparable company is. And this is something I'll come back and talk about more. In valuation, when I did the intrinsic valuation model, What did I say the value of an asset was? Present value of the expected cash flow is discounted back at a risk-adjusted discount rate. Did I say tech cash flows are different from steel company cash flows?
Cash flows are cash flows. So if I have two companies with the same growth, the same risk, I should have the same value for them even though one might be a utility and one might be a tech company. I know it sounds crazy. But I don't invest in companies because I like steel or I like software. I invest in companies to collect cash flows.
So we're going to talk about how we might be able to expand the notion of comparable, not just for betas, but for pricing, so that we have a much bigger sample. And finally, I don't know whether we will get to this, but at some point in time, we'll have to leave equity after we've got the risk-free rate and talk about the cost of debt. So I'm going to set up the problem.
Let's assume you're valuing a company which has a billion dollars in bank loans outstanding. And those bank loans carry an actual interest rate of 4%. The loans were taken a couple of years ago and rates were lower. The rate right now...
has gone up the risk-free rate itself is already up to five percent and this company's default risk gives it a two percent spread over the five percent so i've added a lot of stuff in this problem so they have billion dollars in bonds that they've already borrowed it's on the books it is a four percent rate but right now the risk-free rate is five percent and they have a two percent default spread on top so i'm going to throw I want to estimate the cost of debt for this company and I'll give you the choices. I can say look I'm not going to play games, their actual debt has an interest rate of 4%, it's called a book interest rate, I'm going to use that as my cost of debt. That's your first choice. The second is you say look I can't borrow less than the risk-free rate, I'll just use the risk-free rate instead 5%. The third is to add the 2% spread to the 5% and come up with 7%.
But since you hate to throw away any information, your fourth choice is to take the 7 and the 4, add them and divide by 2. Why would you do that? People hate to throw away numbers in a problem. So my question is, what should I use as my cost of debt? The actual cost of debt or what they will have to pay today if they went out and borrowed money?
It's going to make a big difference in my cost of capital, right? Today again, this is a preview. A cost of debt is the rate at which you can borrow money today. I don't care what's on your books. It's going to make your life actually much easier because I've told you that.
Because what I just said, don't go digging through the books to see what bank loans they have and what it is. Because it doesn't matter. It's the rate at which you can borrow money today, which means to estimate the cost of debt.
My entire focus is going to be to come up with the right default spread to attach to. For some companies, that's going to be easy. For other companies, it's going to be messier.
But if I can do that. I have a cost of debt. And finally, to bring these costs of equity and cost of debt together, I need weights.
Because ultimately, the cost of capital is a weighted average. Let's say the company has a market value of equity of a billion, and you're computing a debt to capital ratio. But the company that I just described, that company had a book value of a bank loan of a billion. So I'm just going to give you very rough choices. So basically, a billion in equity.
They have a billion in bank loans on their books. So is their debt ratio 50? Looks like it should be 50%, right? But I gave you a piece of information in the last problem, which might affect your answer.
What did I say they borrowed at? 4%, right? What did I say they would have to borrow it if they went and borrowed today?
7%. So if the rate today is 7% and they have debt on their books on which they're paying only 4%. You see what's going to happen to them? That bank loan is not traded, but if it were a bond, when you have a coupon rate way below the market interest rate, what happens to the price of the bond? It's going to drop.
So the market value of that bank loan, even though it's not traded, if I estimate market value, is going to be less than a billion. You say, but I worked in an investment bank last summer. We always use book value of debt as market value.
Investment banks do that all the time because it's how different can they be. Until you run into a distressed company and when they discover how different. This is something we're going to do with almost every company. We're going to take book debt and convert to market debt. It's not rocket science.
We're going to act like it's a gigantic bond and reprice it. Right? Because all I need is the remaining maturity in the interest rate and I can do it. Which means that the debt for this company is going to be less than a billion. Which means that the debt ratio is going to be less than 50%.
So it's a lot of stuff I've kind of given you a preview. So let's get. the process rolling. Let's go back to the...
And let's start with relative risk measures. So the way we're all taught how to estimate betas is to run a regression. So that's how your finance class will run a regression of returns in the stock against returns in a market index.
The slope of the line is the beta. I have three problems with regression beta, and I'll list them out, because, and I will back this up. First is, when you run a regression in statistics, forget about what you learned in finance. When you're in a statistics class, you run a regression, you report the coefficients, right?
The intercept and the slope. But right next to the intercept and the slope, you also report standard errors in t-statistics. Practice again, we conveniently forget after we leave the statistics class. You see where I'm going? When you run a regression to get a beta, That regression gives you a beta number, but it also has a standard error that we don't even look at in finance because we act like it doesn't matter.
I'm going to argue that when you bring that standard error in, you're going to discover that when you run regression betas, the range on a beta is going to be huge. That you don't have a fact. You don't have a number.
You have a range for the beta. That's the first problem. Second is, by definition, a regression beta is backward looking.
Why? I used returns in the last two years, the last five years, and if your company has changed, what company hasn't changed in business, makes change in leverage, your regression beta no longer fits. And third, if I gave you a private company, let's suppose I ask you to value Uber. You need a beta, right?
Try getting a beta for Uber on Bloomberg. They'll say, not available. Why not?
It's a private company. There are no stock prices. So if I gave you the division, I'd say, GE wants to sell off GE Capital. Can you estimate a cost of equity?
You can't get a beta for GE Capital. You're handicapped. The only way you can get a beta is through a regression. You're stuck. So I'm going to show you how Bloomberg terminals have been both a godsend.
and a curse. In most consulting firms and banks, if you're asked to get the beta for a company, you know what you do? You run as fast as you can towards a Bloomberg terminal, you find your stock, and you type in the word beta. And a page like this pops up. It's like magic.
So when you run this regression, for instance, with, this is GoPro about a couple of years ago. Know what GoPro does? It actually makes cameras for hyperactive, hyper-sharing.
Individuals basically they strap a GoPro to their head and take you on a four hour hike as if anybody wants to go with them. But for a while it was a star company right everybody was going to buy GoPro. So this was at the peak of their glory and their risk.
I ran a regression beta. I got a raw beta of 1.60. I'll talk about the adjusted beta in a minute.
But if I trust this regression the beta for the stock is 1.60. You see this is good. I can do my valuation now.
Before you get too excited though. about the beta, keep going down. What does it say the standard error of the beta is?
0.50. Remember your statistics. If I tell you the regression beta is 1.60 and the standard error is 0.50, I'm not telling the beta for GoPro is 1.60. I'm saying it's somewhere between 0.60 and 1.60. Do you see how I got the range?
Just add or subtract two standard errors. This is, of course, a particularly egregious case because it's a young growth company and regression betas are extraordinarily noisy. You think, but what if I have a regression beta that looks really, really good? This looks like a much better beta, doesn't it? Every point is on the line.
This is a beta page I printed off for Nokia in 2001. I've saved the page. The R squared is like 94%. If this were a statistics class, I'd get a gold medal. The standard error is close to zero.
You think, this is the way it should be. But why is Nokia's beta looks so good? What did he say the beta was? It's a regression of returns on the stock against returns on a market index.
Notice how in a finance class they leave the market index kind of fuzzy. You go to Bloomberg, Bloomberg picks your stock, then it goes looking for a market index. And it is very parochial about the way it does it.
In the case of Nokia, what is a market index? It's the hex. I didn't even know there was an index called, I thought it was a witch's curse until I ran this regression.
The hex is the Helsinki exchange. You're saying, what the heck is Bloomberg going to Helsinki for? Nokia is a Finnish stock.
So it went to Helsinki. And I make a confession. I didn't know what was in the hex. And I decided to take a look, and I wish I hadn't. In 2001, Nokia was 80% of the hex.
I think Nokia owned Finland. So what do you have here? A regression of Nokia against Nokia. And what are we finding?
The two move together an awful lot of the time. Congratulations. Let's move on. So you can't trust regressions when they're noisy and you can't trust them when they're looking good. You say, but what about the adjusted beta?
Because notice that both for Nokia and for GoPro, there's an adjusted beta. So you're probably thinking there's somebody at Bloomberg, a team sitting together adjusting these betas. Let go of that illusion.
The adjusted beta for every single Bloomberg beta page that you see. It's 2 thirds times the raw beta plus 1 third times 1. You're saying, what? Let me try some numbers.
If your raw beta is 1.8, your adjusted beta would be 2 thirds of 1.8, which is 1.2, plus 1 third of 1, which is 0.3. So 1.8 will become 1.53. 1.5 will become 1.33.
1.2 will become 1.13. 0.6 will become 0.7. Notice the pattern here. What are they doing? They're pushing betas towards 1. What's so magical about 1?
That's the average. They're trying to be helpful. They're saying, if your company got bigger and more diversified, its beta is going to move towards one.
So we're going to move it towards one for you. To which your response should be, please stop helping. Because these weights must be magic weights, right? 2 thirds and 1 third.
Where did that come from? Why not 0.7 and 0.3, 0.8? And how come it's the same for every company?
It's not a fair question, because I didn't know the answer. I decided to call Bloomberg. Not the mayor, but the company. For a long time, it was a little confusing who you were calling. Maybe I should have called the mayor.
And they put me in touch with a beta calculation guy at Bloomberg. It's actually a guy at Bloomberg whose life it is to maintain this page. Imagine how exciting his life must be. He goes to a cocktail party. He says, I'm the beta guy at Bloomberg.
Dozens of people gather around for anecdotes. Not. The guy was pathetically grateful to get a call from the outside world. I don't know what they do to the guy. Maybe they keep him locked up in a basement room, feed him through a hole in the wall.
He says, I'm so glad you called. I have all day to answer your questions. I have all day to ask him questions because I'm afraid if I hung up the phone too soon, he might do something rash.
So after about five or ten minutes of what I thought was polite conversation, I hit him with a question. I said, why 2 thirds and 1 third? He said, huh? And two minutes of silence.
And the paper's being rustled, terminals being turned off and on. Finally comes back and says, I don't know if it was here when I got here. I said, what?
He said, I was hired two years ago. It was already here. Don't blame me.
I said, I'm not blaming you. I don't know where the numbers come from. I don't know. Here when I got here.
I just finished my 11th book last year. I have the title for my 12th book picked out. You know what it's going to be called? Here When I Got Here. Because you know how often that is going to to be the answer about questions you're going to ask.
When you get on your desk, why do we use that number? I don't know. See how when I got here.
When did he get here? 1979. It's amazing how much evaluation we've outsourced. The numbers come from outside.
The bottom line is neither of these regressions are reliable. One is way too noisy. The other one has an index that it basically is the stock. And either way, it is not the beta that you're going to use.
And here's the final issue with regression. A couple of years ago, Valiant, the Canadian pharmaceutical company, was in trouble. Stock had dropped from 200 to 20. Let me ask you a question.
It's kind of a, don't even look at the page. If I ask you, as an investor, do you think Valiant is risky? that has lost 90% of its value over the last eight months? What's the answer gonna be? Yeah, it's incredibly volatile and risky.
You know the beta for value to look like? It's 0.36. How do you explain that? How can a stock that is doing so terribly badly end up with a beta close to zero?
Exactly. Beta doesn't measure how much you move. It measures how much you move relative to the market. So if you have a stock that drops every day, horrible investment, right? But if it drops every day, no matter what the market is doing, guess what its beta is going to look like?
It's going to look like this. zero. It doesn't mean Valiant is a safe stock. I mean, I would take that out of the regression. It just means that this slice of history was such a bad slice.
And the same thing can happen in the other direction. If your stock is the target of an acquisition bid and the price keeps going up and up, its beta is going to look low. So betas are noisy, and they look good, you should not trust them. And they look bad, you should not trust them. And they reflect a slice of history.
And finally, we talked about bias, how as an analyst you're going to go find numbers. That'll help you tell whatever story you want to tell. This is a stock called Bombardier.
Bombardier is a Canadian aerospace company that's a bit of a basket case. It's got incredible amounts of debt. Let's say you're the equity research analyst valuing Bombardier, and you're trying to come up with a value, but you have a bias. You want to come up with a low. So let's say you want to come up with a low value, you want to come up with a high value.
So you want to come up with a low value. What number do you want for a beta? High beta or low beta?
You want a high beta because a high beta will lower your discount rate. You, of course, want to come up with a low beta. And the advantage of having a Bloomberg terminal is when you get the beta page for a stock, you can actually change the index.
And if it has a dual listing in a different market, you can get a beta for Bombardier against a Canadian exchange where it's basically listed, or you can take a regression of Bombardier at the ADR, which is a U.S. listing against the S&P 500. And there are some stocks where you can try five or six or seven different indices and end up with five or six or seven different betas. You think so what? Do you see the potential of a game play?
You're going to try five different betas and what will he pick? He'll pick the lowest of the five numbers, he's going to pick the highest, he's going to... and if you didn't know this, they'll show you the Bloomberg... look it's a Bloomberg beta.
I've told people you give me 30 minutes, you give me access to a Bloomberg terminal, you tell me what beta you want for your company. I will deliver a printed page. I'm not kidding. You want a 0.6 beta?
Not a problem. You change your mind and make it 2.5? Not a problem.
If you get a chance and get on a Bloomberg terminal and you can take about 10 minutes, take a stock and try changing the starting point and the ending point. Daily to weekly to monthly, you can change that option. Change the index just to get a sense of how much game playing can affect betas.
So I don't like betas, at least as they come from aggressions. But before I talk about how to estimate beta rate, let me talk about the fundamental reason why some people don't like to use betas. What do we say the essence of intrinsic value is?
You don't trust markets, right? So you come up with the cash flows, the growth rates, etc. for your company. And if you're doing this right, your risk should also be an intrinsic value measure.
But when you use a regression to get a beta, what are we using to estimate risk? We're using stock prices. Now do you see why if you go to Omaha in the middle of Woodstock and you say the word beta, you're going to be shunned? Value investors think that beta does not go with intrinsic valuation, and they understand.
But here's what I don't get. You know that 95% of value investors don't do discounted cash flow evaluation. They pick stocks with low PEs, low price to books.
But if you ask them, why do you not do discounted cash flow evaluation, the most common reason they give is, I don't like betas. It's like throwing the baby out with the bathwater. So this page is my counter to those people who say, so you will run into people, your managing director, saying, what the hell is this beta that you learned in class, and I don't like betas?
Okay. At the risk of losing your job. Ask him a question or her a question.
Why do you not like betas? Because there are two fundamental reasons people have issues with betas. One is because it's a price-based measure and they don't like price-based measures.
And if that is your problem, I have alternatives. Because beta, after all, is a measure of relative risk. You know what I mean by relative risk? A beta 1.2 means you're 1.2 times more risky than the average stock. Maybe I can come up with a number that just looks like a beta without using stock prices.
You're saying using what instead? what intrinsic value people look at to measure it. They look at earnings and how volatile they are. Maybe I can compute the standard deviation in earnings over time, and let's say that number is 20%, and I can compute the standard deviation in earnings for the market, and it's 25%.
You know what I'm going to do? I'm going to divide the 20 by the 25. I'm going to come up with a 0.8. It looks like a beta, right?
But it's based entirely on earnings. If you don't like prices, I will use something else. But you need to tell me what you think of... as your measure of risk, and I can compute something that looks just like a beta. So if your critique is you're using price-based measures, not a problem.
I'll use something else. But here's the other reason people don't like to use betas. When you use beta, and we talked a little bit about this two classes ago, whose perspective are we taking?
An investor who is diversified, right? That's what allows us to ignore all the other risks. So the other reason investors don't like...
beta some investors is they say look I don't plan to be diversified I don't want a measure of risk that is just the diverse of the risk you cannot diversify away I want something more complete and if that's your reason I have an answer for you as well Instead, when I use beta, I'm looking at the portion of the variance that can be explained the market. That's why it matters when. But what if I said look, you know, I look at your total variance. If I look at valiant's total variance, it's huge. It's twice as high as the market's variance.
So you know what I'm going to do? I'm going to take the variance, the total variance for valiant, and divide by the variance for the market, and that number is going to give me a 2, which I'm going to use just like a beta. My point is... If you run into people who say, look, I don't do discounted cash flow valuation because I don't like betas.
Betas make no sense. Don't pull out modern portfolio theory. Don't put a cop an attitude and say, look, I went through a portfolio theory class. You didn't. Reason through why they don't like betas and try to come up with something that they can live with.
Because to me, valuation shouldn't ride on whether you use betas or not. It's a measure of relative risk. And if you can come up with something different, all the more power to you. So when you think about price-based measures, you can look at relative standard deviations. So this is a measure where you just look at the total standard deviations rather than the portion.
You also have proxy models, where you basically don't even try to measure risk using standard deviations at all. You look at sectors, and you say, this sector is risky, this is safe. Tech sector, historically, risky. Utilities are safe. It's kind of dicey, because sectors change over time.
Or you can use something like market caps. Small companies are riskier. I am willing to live with a lot of different variations of risk if that is what you think you want to bring in.
In fact, many people try to combine the CAP-M with proxy models and that's when you see when you see valuations you see people use risk free rate plus beta times risk premium and then they'll add a small cap premium or a low price. So basically there's that's their way of saying look I think the CAP-M is incomplete I'm going to bring something else in to make my discount rate complete. And if you don't like price-based measures, I guess you trust accountants more than markets, then build something off accounting numbers, revenues, earnings. So again, my point is don't get stuck on your dislike of betas. Because I'm not naive.
I know that some of you will hear what I say about betas. And then you will say, look, it doesn't make sense. It assumes modern portfolio theory.
Fine. Just throw betas out. Replace them.
The one thing I don't want you to replace with every company is equally risky. That's the one thing that is not an alternative. Think of something else than as your proxy for risk and keep moving.
So let's talk about where betas really come from. When I go to corporate finance departments at investment banks, I throw this question out, where do betas come from? I once had a guy who tried to explain to me where babies came from, and I said, look, that's not the question I asked.
I have four kids, I know where they come from. And if you, but most of the time they look at me like I have two heads. What do you mean where do betas come from? They come from Barra, they come from Bloomberg.
Because let's face it, for most of you, once you get on your desk, that's where betas come from. You go look them up. And if you dig a little deeper, you get this, oh, you get a regression. The truth is betas don't come from regression. They don't come from Barra.
They come from choices you make as a company. In fact, there are three choices you make as a company that will determine your beta. First, tell me what kind of business you're in, what kind of product or service you offer.
Remember, betas measure how you move with the market, right? So if you're a cyclical company, a cyclical business, I would expect you to have a higher beta than if you're non-cyclical. Housing and steel companies and automobile companies should have higher betas than consumer product companies. I think about 20 years ago, that's the only distinction I used to draw.
But increasingly, it's tough. to classify the world into cyclical and non-cyclical. Is Facebook cyclical? Is Facebook non-cyclical?
You can see that this challenge is big enough. So I'm going to give you the way I differentiate across companies. The more discretionary the product or service you offer as a company, the higher beta will be as a company.
You know what I mean by more discretionary? If your customers can live without your product, they can delay buying it, they can defer buying it, you will have a higher beta than if you have to have... if you're producing a product, that they have to have. I'll give you a sector where this plays out and you can see the differences.
Take retail. Huge sector, thousands of companies, right? But if you look at retail, you have luxury retail, you've got department stores, you've got discount retailers, and you've got grocery stores.
Just generically, think of those four classes. Given what I just said about discretionary, non-discretionary, which of those four groups should have the highest pay? Luxury retailers, because you can live without that Gucci.
You say, I can't. It's impossible. Go to Canal Street. You'll get something that looks like it.
It might be misspelled, G-U-C-H-I or something. But who cares? You save so much money. Luxury retailers will have the highest.
Department stores are going to be the next highest. Then discount retailers and grocery stores. But to show you, even within grocery stores, the shades of gray you can get.
You can have a ShopRite or a Pathmark. You can also have a Whole Foods. In California, now you have Sprouts. You think, what's different? They all sell groceries.
But if you're feeling flush with cash and you want to live longer, you're willing to pay three times what a regular eggplant costs at Whole Foods because it makes you feel much better. But if you lose your job, guess what? You know you're going to die anyway. You think, what have I? The pesticide, who cares?
I'll just go buy the cheap eggplant. You know what? Within the grocery store business, I would expect Whole Foods and Sprouts to have much higher betas than the traditional grocery stores. It's a very interesting, a different way. In fact, every class you ever take is in the service of this class.
You know in Econ 101, you talked about a concept that's very closely tied to what I'm just talking about. Elasticity of demand. That's basically it, right?
If you have a product with elastic demand, you should have a higher beta than if you have a product with inelastic demand. You know tobacco companies have really low betas? Why?
If you want a really low beta, make your product a service in addiction. You see those nasty people? If Morgan Stanley could addict people to structured products, would they?
In a second. Every company tries, it just doesn't work as well for them. So don't pick on a company because it worked for them. You tell me what you do, I can already start to lay the foundations for whether you should have a low beta, an average beta or a high beta.
So that's the first stop. Any questions on that? Here's my second stop. You had a question? Second stop.
I'm going to ask you, tell me something about your cost structure. Tell me how much of your costs are fixed and how much are variable. See, why does it matter?
If you have a lot of fixed costs in your business, what's going to happen? In good times, you make lots of money. In bad times, you lose lots of money. Every shock to revenues gets magnified, right?
So if your cost structure is heavily structured around fixed costs, I would expect to have a higher beta than if your cost structure is heavily variable. That's the reason airlines should have very high betas. The cost structure for an airline, pretty much every expense is fixed.
But to show you, even within the airline business, you can have outliers. What's the most written about company and case history? More cases have been Southwest Airlines. What makes Southwest different? They buy the aircraft like anybody else, but for a long time, at least the aircraft, which is a fixed cost, but for a long time every Southwest flight in the country was a Boeing 737. They always said what, because you know why?
It saved the money on maintenance crews, they kept the cost low. They have to spend on fuel like every other airline. I hope they don't economize because sometimes I do take Southwest.
But they're the only one of two airlines that consistently hedges. They hedge all the time. As opposed to what?
As opposed to most airlines that start hedging when oil prices hit 120 and stop hedging when they get to 30. They start and stop at exactly the wrong times. Third, they have a unionized workforce, but it's the most flexible workforce of any of the airlines. It's the only airline where I've been checked on by a stewardess.
Has that ever happened to you at United? They're checking the job description. No, I've been working here 35 years.
I've never checked somebody on. I can't start with you. But here's the biggest saving that they had for much of their history as they grew. I remember the first time I flew Southwest out of the New York area. I made my reservation.
I didn't even check the airport. I didn't know if I was going to get there. I'll get to the airport.
So the day of the flight, you know, luckily about three hours before the flight, I checked see which JFK, Newark or LaGuardia. And it said I slip. I don't mean to insult people who grew up in Islip.
I had no idea where Islip was. Can you imagine getting into a cab and saying, take me to Islip? I don't even know what will happen.
It turns out to be this small town in Long Island with an airport. And they flew out of Islip. And I was flying to LA. And I said, OK. I found my way to Islip.
I get on the flight. I expect to land in LAX. I land at the airport.
I come out. There's this huge statue of John Wayne. And I said, this is not LAX. I ended up in Santa Ana. And this was a pattern consistently across the country.
They flew out of Midway, not O'Hare. And why do you think they did that? Have any of you flown United out of Terminal C in Newark? How many gates does United have?
I think they do this just for laughs. They make you come into gate 147, give you a connecting flight at gate 3, and as you run, they're watching you up, saying, you're never going to make it, let's switch the signs on. It's like rats in a maze. But they pay for all 147. 50 gates.
The way airlines get gates at the major airports is you pay for those gates, whether one flight leaves out of them or 50 flights leave out of them. It's a fixed cost. And Southwest early on said, we don't want to be like other airlines. Why would you want to?
to be airlines it's been feast of famine bankrupt of billions year after year since 1977 it's a horrifically badly run business so guess what given what I just said I would expect the beta for Southwest to be lower than the betas for other airlines and finally I'm going to ask you whether you've inflicted some fixed cost on yourself what am I talking about how much have you borrowed because when you borrow you create a fixed cost you do not have until you borrow which is interest expenses you got to make the interest expenses in good times and you make them in bad times, it makes your equity earnings more volatile. Three basic questions. What is your product and service? How discretionary is it? What does your cost structure look like?
And how much have you borrowed? By the time you finish answering those three questions, I should be able to come up with the beta for your company. So I'm going to take you through the process by which I estimate betas, and it's actually a pretty straightforward process. So let's say you're a company, and you say I want a beta for my stock. I say okay.
I'm going to start with the business or businesses you're in and I'm going to take away the damage. Now if I had this were a perfect world I would take the beta of the business here and adjust for your fixed cost structure. The problem with that is the information and fixed variable cost is all over the place. You don't see it in income statement.
But if you did I adjust for our business. So basically I'd like a business beta then I'd I did it just for fixed costs, and then I did it just for financial leverage. That's the sequence I'd go through.
So I use an approach that I call a bottom-up beta. You can give it whatever name you want. And here's how it goes.
If you want to adjust for operating... leverage, I'll give you the mechanics for doing it, and then you're going to see very quickly why almost no one does it. To adjust for operating leverage, I need to know how much of your costs are fixed and variable.
So if I could go through income system, say 60% of your costs are fixed, 40 are variable, and I can do this for every company in your sector, adjusting for differences in operating leverage is trivial. I adjust for just the same, I adjust for debt to equity ratios. The reality is very, very, very difficult to get that information.
So I'm going to skip that step. Thank you. wave my hands and say, unless it's a company like Southwest or Whole Foods, where I know I have to consciously adjust, I'm going to assume that if you're a steel company, your car structure looks like that of a typical steel company.
If you're a chemical company, it looks like that of a typical chemical company. And I'm going to build on that basis. And adjusting for financial leverage is much easier because it's been around a long time. And that we should be able to get the information because all it requires is how much debt you have and how much equity in market value terms.
The conventional approach, and this is the one that I'm going to use, is called the Hamad. And here's how it goes. To get the beta for your equity, it's called a levered beta, I start with the beta of the business or businesses you're in. That's called the unlevered beta. So already you can see multiple betas coming at you.
It's also called an asset beta. But basically, it's the beta of the business. is your end. I estimate how much you borrow with the debt to equity ratio.
So as you borrow more, that debt to equity ratio is going to go up. And I do give you one consolation price when you borrow money, which is the tax code gives you a benefit. In what form?
Interest is tax deductible. So even though you borrowed a lot, the government acts like a co-guarantor by giving you this tax benefit. That's what the one minus is.
So I start with the unlevered beta of your business that you're in. I take the debt to equity ratio, reflecting how much you borrow. you're going to borrow.
And then I adjust for the tax benefit. I come up with a levered beta. When I invest in a stock, it's a levered beta that you would see in the stock, not the unlevered beta, because the unlevered beta is the beta of the whole business.
So let me ask you a question. Can you have risky equity in a very safe business? What does it mean when I say very safe business?
Unlevered beta is a low number, right? And I said the equity beta is a levered beta. So how do I make equity in a safe business into a very risky equity?
I borrow a ton of money. I do a leverage buy. out of a safe company, think of what my equity is going to look like right after the leverage buyout, after I've borrowed all that money, my equity got risky but it's self-inflicted. I did it because I borrowed money. There are two ways a company can end up with high betas.
One is by being in a really risky business, the other is by going out and borrowing money on a safe business. Now, one of the challenges that... with the traditional approach, and the reason I'm showing you this alternative is not because I plan to do it in my companies.
It's just not worth the effort. Is when you use the conventional approach to adjust to this Hamada equation, where you take unlevered and levered betas, you're effectively assuming that debt has no market risk. It has default risk, but the risk is not market risk, which means the beta of debt is zero. That's implicit in that first approach. And you get some pushback from people saying, well, that's not necessarily true.
There are times when debt is zero. debt has market risk. And that's especially the case when you get to be a triple C or a double C rated company. You know what?
I can adjust for that risk if you can give me some information. Here's what I need to know to adjust for the fact that debt sometimes has risk. I need a beta for the debt.
You're saying, how am I going to get that? Just like you get a beta for the equity. In fact, on my website, I have, I think, betas by ratings class. If you're a triple C rated company, it's like having a beta of 0.3.
You plug those numbers in, you'll get a new beta for your equity, which will be lower than what you got. with the conventional approach because now you've got somebody else bearing the market. So let's summarize.
When you think about what drives beta, you want to start with the business beta, you want to adjust for operating leverage if you can and as I said it's tough to do, then you want to adjust and when you're done you're gonna have a levered beta for your company that reflects the choices you made, not some regression you ran of your stock against the market over the last two or five years. So here's the sequence. Let's play again.
Let's assume that you have a company. I'm going to make you have the company. So what businesses do you want to be in? Pick any two businesses.
Make them fun businesses. Oh, you can't pick a company. So you want to be in social media. So online advertising.
Pick another business, too. Fintech, okay, you really want to go out to the edge there, right? So basically online advertising and let's say portfolio management, because fintech can be banking. So you want me to estimate your bank. So here's what I'm going to do.
going to say I'll be back tomorrow I'm going to go find as many publicly traded online advertising companies as I can and I think there about a hundred and twelve across the world and because they're publicly traded what can I look up for each of them I can look up a beta for each of them. I know the beta is a crappy estimate of risk, but I look up the beta for each of them. I take a simple average.
I now have an average for the online advertising companies that I list. Now, we said that beta that we see, which is a regression beta, is affected both by the business and the debt that they have as companies. So I'm going to do some cleaning up.
Remember how I levered betas? I'm going to take out the effect of debt from these average betas. I'm going to end up with an unlevered beta for online advertising companies, essentially cleaning up for the debt.
Then I'm going to do the same thing for the FinTech business. If I'm publicly traded, because FinTech companies are young and not traded, but the business they're in has traded business with other portfolio managers. I take their average beta, I clean up for the leverage, I get an unlevered beta of being in portfolio management.
So now I have two business. and the unlevered beta of being in the two businesses. I come back to you and say look you told me you were in two businesses can you tell me how much of your value you get from each business? Your reaction might be I don't know how much value I get but last year I got 80% of my revenues from online advertising. advertising 20% from FinTech.
I say, OK. That's good enough. I take a weighted average of my two unlevered betas. I now have an unlevered beta for his company.
What's the last step? I come and ask you whether you have any debt. If you say no, I'm done. That is now your levered beta because you have no debt.
But if you do have debt, I compute your debt to equity ratio. I plug that into the equation I showed in the previous page. I come up with a levered beta for your company.
Why am I doing all of this? Because I don't like regression betas, right? But where did I get the 112 online advertising company betas? From regressions. All I've done is replace one regression beta with the average of 112. I talked about the law of large numbers.
You're saying, what's the big deal? It's still crappy numbers. The average of 112 crappy numbers is magically precise. And it sounds completely senseless until you remember what I meant by crappy numbers.
Some of these betas are overestimated. Some are underestimated. By averaging out, what have I just done? I've averaged out my mistakes.
I'll give you a rough sense of how much benefit the law of large numbers gives you. Remember that GoPro beta, which had a standard error of 0.50? Let's say I find 100 companies just like GoPro.
Each has a standard error of 0.5. So they're all horrifically bad betas, right? I take an average of the 100 betas. Remember, each had a standard error of 0.5. You know what the standard error of the average is going to be for about 100 companies?
It's actually a very simple statistical... equation that you can use. It'll be 0.5, which is the average standard error, divided by the square root of the number of companies I have in my sample. So I have a hundred companies, square root of 100 is 10, my standard error, even though 100 beta is at point, my average is going to have a standard error of 0.5.
1 tenth the standard error of a single regression beta. Now do you see what the advantage of the law of large numbers is? When I want to get a beta for a chemical company, you know how many publicly traded chemical companies there are globally?
673 companies. Some are crappy beta, some are too high, some are too low, but by taking the average of 673, I get the weight of the law of large numbers behind me. And there's no more game playing, right?
I can't sit there trying different things. Once I decide I'm in the chemical business, the beta is right there. It saved me incredible amounts of time not going to the Bloomberg, checking out this regression beta, playing games.
And this says, look, you tell me what business is it. Let's move on. So if you take a look at both the Apple and the Amazon valuations, do you guys value Amazon yet?
Have you built in the National Inquirer effect? I did not. So somebody actually emailed me and said, where's the National Enquirer effect? I don't even know what that effect is. Is it good?
Is it bad? Will Amazon start selling pictures of naked Jeff Bezos? I don't know. I know whatever it is.
So I don't know what the story, how that story plays out. But if you take a look at the betas that I have for Apple and Amazon, I broke Apple down into three businesses. The biggest business by far is the Smartphone business. 75% of the revenues come from the iPhone.
The second business is computer services. That's all the stuff you buy in the App Store. And the third business is the computer hardware business. I took away the average of the three.
I have a MacBook Pro. I've had one since. It pisses me off that they don't pay any attention anymore to me from the MacBook. But I understand.
You know why I understand? Because the MacBook Pro now is a throwaway item for Apple. It is not at the center of the universe anymore, because so much of the value comes to the iPhone. Everything is about the iPhone.
When I valued Amazon, this was a tricky one. I did put a significant portion of the revenues into retail. I think it's $149 billion. But there are two other businesses. One is the cloud, which is AWS, which is really a computer services business.
And the other is Amazon Prime. It's a subscription business that I separate out because it's very different from the retail business. Three businesses.
It requires that you be creative. You don't look up just the SIC code. Be creative because you're building up to a risk for your company given what it does. That's why when you do Facebook, it is an online advertising company. You can dance around how it sells its stuff online.
It's an online advertising company valued as such. So this is the process I'm going to use. And I'll take you through a couple of companies.
So my arguments for bottom are beta. First. it's going to be far more precise than any individual regression better. So even if you're in a single business, you're better off using the average beta for the business than your own regression beta.
Second, if you've changed your business mix and your financial leverage, guess what? I can reflect it, right? How? I pick the weights. So if he tells me that he's going to be in the casino business tomorrow, not a problem.
I'll bring it in because I know the beta of the casino business, and I'll bring it in as a weighted average. If he says I plan to double my debt next year, not a big deal. I'll bring that debt in. It allows me to be proactive.
I'm no longer a victim of a regression beta that showed up on a Bloomberg beta page. I set the terms. And third, if I ask you to value Airbnb, You have a template now, right?
What business is Airbnb in? The hospitality slash hotel business. There are publicly traded hotels out there.
I can take their betas and I can get started and do my valuation for a private company. Same thing, if you want me to value GE Capital, I can look at publicly traded financial service companies, come up with a beta. I can value private businesses, divisions of companies, pieces of businesses.
I no longer have to get that Bloomberg beta page in my hand. So I'll take you through an example of how this plays out. Vale, mining company, and I want you to get a beta for Vale.
So I broke Vale down. Actually, most of the time when you think about where will I get the business breakdown, look in their annual report, 10K. Somewhere they're going to break it down, and you then have to decide how fine the breakdown has to be.
They break themselves down into four businesses, metals and mining, iron ore, fertilizers, and logistics. There are the number of firms that I've found as my comparable companies. 48 metals in mining. You think only 48?
Even with 48, how much lower is my standard error going to be? Square root of 49 is roughly 7. 1 7th, right? So I'm still getting a big...
See, I've only 25. It's still 1 5th. I've only 9. It's still 1 3rd. Anything beats 1. That's what the regression beta is. It's a sample size of 1. Even if I have only 4, I would like to have more, but I'm still beating your 1. So the largest sample, and you can see, fertilizers, 693 companies. There's the unlevered beta for each of these four businesses.
Now to decide what the weights were, remember I said I'll ask you how much revenue you got? I started with the revenues for the four businesses, which they gave me. And if I were in a hurry, I would use revenue weights.
But in my experience, revenues and value don't always go together. If you're in a low margin business, the same revenues will have a much lower value than in a high margin business. Remember those 48 companies that I got betas from?
I also collected one other statistic on each of the companies. It's called the enterprise value to sales. Sounds fancy, but I took the market value of each company and divided by the sales for each company. So the way to read this, a typical metals and mining company trades at about 1.97 times revenues.
A typical iron ore company trades at about 2 and 1 half times revenues. A fertilizer company trades at about 1 and 1 half times revenues. See what I'm going to do?
I'm going to take your revenues from each business and multiply by that number. So for fertilizers, multiplying 3,777 by 1.52 gives me an estimated value for the fertilizer business, and I get that for all four businesses. The weights that I get, at least based on my judgment, is Vale is about 76% iron ore, 16.7% metals and mining, about 5.39%.
Fertilizers and about 1.8% logistics. Incidentally, Vale sold their logistics business about a year after I did this. So if you'd asked me to value just the logistics business, I'd have focused just on the beta for the logistics business.
So I actually used those betas to come up with 11 beta for each business. I did cheat because Vale doesn't break its debt down by division. I assume they all had the same debt to equity.
Sometimes I adopt that rule, sometimes I kind of bend that rule, but essentially I have a levied bade and a cost of equity for every single business. If you're wondering where the equity risk premium for Vale comes from, where does it come from? It doesn't come from Brazil.
It's an iron ore, global iron ore mining company. It comes from where they sell their iron ore. And guess what their biggest country exposure is to? Good, you took my lesson. Whenever you don't know the answer questions, say China.
It is China. 37% of Vale's revenues come from China. When China sneezes, Vale catches a cold. But let's face it, you know why China is 37%, right?
What do you need iron ore for? You need it for steel and construction. And half the infrastructure built around the world in the last 15 years was built in China. So this is not a mystery. So when we talk about equity risk premiums, that's why we bring in country risk exposure.
So let's review. Risk-free rate reflects currency choice. Equity risk premium reflects where you do business.
Beta reflects what businesses you're in. Everything has a place in that equation. Don't mix and match. Don't let your beta carry country risk in it, because then you're going to end up double counting.
This keeps the process clean. Now let me do another example and then we're kind of ready to move on. I wanted a beta for Embraer. Embraer is a Brazilian aerospace company. With quotes around the Brazilian.
They get about 3% of their revenues in Brazil, 97% overseas. So I wanted a beta for Embraer. Remind me again what the bottom-up beta process is. You have to go find other companies like it.
So initially I looked for other Brazilian aerospace companies. Guess what I found? Just one called Embraer.
That's not much of a bottom up, if you replace Embraer with Embraer. So I said, let me look for Latin American aerospace companies. I came back with one, still called Embraer.
I'm kind of getting stuck. I look for emerging market aerospace companies. I got like two.
And then I said, why the hell am I doing this? Who does Embraer sell its aircraft to? United and Swiss Air.
Who does Boeing sell its aircraft to? United and Swiss Air. This is a global business.
I'm going to look at aerospace companies listed globally and that's where I got my unlevered beta. It was an unlevered beta for publicly traded aerospace companies. So you had Bombardier, you had Boeing, you had EADS, basically all the companies out there. Often when people, when I tell people to use bottom-up betas, you know, they listen and then about two weeks later they say, look it didn't work for me. I have a company that's, you know, I can't find any comparables.
And I tell people that if that's the case they haven't tried hard enough. Usually when you cannot find comparables, one reason is because you stay regionally locked in. So you want to value an Indian company, you look at only Indian companies, maybe you can't find comparables because there are some markets that are smaller. And especially if you're in the Portuguese market, God help you if you stay within the Portuguese market.
I think there are only 80 companies listed in the entire market. So first thing I suggest to them is go global. The second is move up and down the chain.
Let me explain what I mean by that. When you think about doing a company and you say, look, I can't find companies like, think of other companies that do well when you do well and do badly when you do badly. Using that definition, if you're manufacturers, you're suppliers, you're part of a chain, go up and down the chain.
But try not to define your business way too narrowly. A few months ago, I got an email from a Chilean analyst who's valuing his company. And he said, look, you know, I read your section on bottom-up betas, but it doesn't work for me.
I can't find any comparable companies. So I said, what kind of company do you have? He said, it's a shrimp fishing company in Chile.
So I said, what exactly were you looking for? He said, I was looking at your industries to see if you had shrimp fishing in there. I said, OK, let me ask you a question. I don't mean to be insulting, even though I was. I said, what do people use shrimp for?
She said, they eat them. I said, do they eat other stuff as well? He said, I guess so. I said, use food processing. and move on?
But shrimp is unique. Really, where does it go? Different stomach?
When you define businesses way too narrowly, you're always going to get in trouble. Kind of step back and say, okay, what does this company do? Where does it do business? How does it do business and move on? In fact, with Embraer, one of the things that I'm doing is I'm going global for a Brazilian company, and I'm having no qualms about doing it because they all sell into the same market.
I'll tell you when I might have a few qualms about taking global betas and using them for a company. If I have an emerging market telecom company, I might be in danger if I use developed market telecom company betas. And here's why.
What I say drives betas is how discretionary your product is. services, right? If you live in the US or Europe, at this point, your cell phone is not discretion. My guess is you probably are more attached to your cell phone than to your food. You say, take my food away, but my cell phone, no, I can't give that up.
It's basically part of your everyday. But there are parts of the world where a lot of people don't know. They might have a cell phone, but very limited service.
And how well you do as a company, as a telecom company in the emerging market, might very much be tied to how well the country does. You see what I'm saying? Telecom in an emerging market might be more discretionary as a business than in a developed market, which means you want to get a beta for emerging market telecom companies.
Incidentally, if you've explored my website and you clicked on data, I'd suggest you do it. I estimate betas by business for around 100 different businesses, and I do this every year. I have 43,000 companies in my samples. I have the law of large numbers working in many different places. mega power behind me.
But I report five different betas. I report a beta for US companies. Why? Because those betas are probably the cleanest betas because the index is broad. I also report a beta for European companies.
I report a beta for global companies. I report a beta for emerging market companies. I report a beta for Japanese companies. So essentially, because Japan has some very unique structures for its industries.
I even report betas for Indian and Chinese companies as standalone because there are thousands of companies. I don't report betas for Indonesian companies because there are only 600. 700 companies. But the reason I do that is for myself, because for the rest of this year when I want to value a company, let's say it's a steel company in Nigeria, you know what I'm going to do, right? I'm going to go look up the beta for emerging markets. steel companies if I feel this is an emerging or global steel companies and I'm going to move on it is the place I go to get my baits in fact I don't even go look it up I built it in my spreadsheet I had picked the business and go looks up the bay this way there's as little human intervention as they can be because I know what happens when I get to pick because my biases are going to find a way so and I'm sitting there playing with where should I put Apple I'm going to want to come up with a low bait and a high value God only knows what choices I will make so this part of the process I want to pick the businesses and let the cards fall where they might.
Any questions on bottom of betas? One final point and this is especially for those of you might end up working in Europe or Latin America. When I do debt to equity ratios I look at what's called gross debt, the total debt.
Much of Europe and Latin America the practice is to use what's called net debt. You think what's net debt? You take the gross debt and you net cash out.
And often I will get emails from people saying, you know, I see your entire book, you use gross debt ratios, can I use net debt ratios instead? And the answer is, absolutely, if you're willing to stay consistent with them all the way through. So if you look at Embraer, for instance, its gross debt to equity ratio is about 19%. And the Leavitt beta that I get using its gross debt to equity ratio is about 1.07. So I said, let me try the net debt approach.
And the minute I did this, I knew that most people, when they do this, would freak out. Because you get a negative number. Why?
Because the cash balance exceeds the debt. And if you're doing Apple, remember, net debt ratio for Apple is going to be a big negative number. Here's what you need to do.
The fact that it's negative doesn't mean you can throw the rules. You still stick with the debt-to-equity ratio, which means your levered beta is actually going to be lower than your unlevered beta. And intuitively, here's what's happening.
Cash is now becoming an asset with a beta of zero. it's dragging a beta down. You're saying, but this is going to give me very different costs of equity.
You're right. If I use net debt to equity ratios, I get a much lower cost of equity than with gross debt to equity ratios. And if you stop right there, you're going to say, this is going to make a big difference in my valuation. But remember, cost of equity is just an intermediate step to get to cost of capital. When I do gross debt ratios, here's what I will do.
I'll take that 11 beta 1.07, come up with a cost of equity based on that, and then... Computer cost of capital using the gross debt, which will be, you know, give me a debt ratio of 12% or 13% in the cost of debt. I'll end up with a cost of capital that is lower because my gross debt ratio will drag the time.
When I do net debt ratios and I come up with a cost of equity, my cost of equity is going to be much lower. But when I do my cost of capital, remember the debt now will have a negative weight attached to it. My cost of capital actually end up being higher than my cost of equity.
Work out the math and you'll see this. I'll end up with roughly the same cost of capital. with either approach. So the bottom line is don't lose sleep over this.
If you want to go with net debt ratios I have absolutely no trouble with you doing it as long as you stay consistent. The promise, many equity research analysts when they see that negative number they say that can't be, they'll replace with a zero. If you start playing games like that you're going to get very different answers. So if you want to do net debt ratios stay with it all the way through.
So let's recap the cost of equity. The risk-free rate has to be in the same currency as you do your cash flows in. And for some currencies that might require cleaning up the government bond rate.
The beta would reflect what business or businesses you're in. And your equity risk premium you have two choices. You want to stick with the tried and the true, the way people use them.
You can go with the historical premium. But the problem with the historical premium as I said is it's... both backward-looking and it might no longer hold, or you can do what I suggest, which is look at a forward-looking premium, an implied premium, and if you want to get a country risk premium, add on top of that country of the implied premium.
I have eight minutes. I can still get about six pages done. So we've got a cost of equity, risk-free rate, beta risk premium. Let's mop up.
The cost of debt is the rate at which I can borrow money long-term today. Two words I threw in there, long-term and today. You know why I threw the long-term? If I let you tell me what your cost of debt is based on the debt you actually took, given that the term structures are sloping, short-term debt is cheaper than long-term debt, guess what you're going to do as a company to make your cost of capital lower?
You'll keep borrowing short-term debt and say, look, my cost of capital is done. I'm not going to give you that. I'm going to assume that you're rolled over cost even if you decide to use short-term debt.
So I'm going to be look at a long-term cost of borrowing today which means I need a default spread. So I'm going to start easy. Companies where it's easiest to get the default spread are companies which have bonds outstanding. Why is it easy?
Because those bonds are traded and there's a yield to maturity out there. That is actually a market interest rate on the debt and if that bond is a nice clean bond that's my cost of debt. I almost never do this because I don't trust individual bonds.
And here's what. Can a risky company issue a safe bond? Absolutely. What does it have to do?
It has to carve out its safest assets and back up the bond, right? So when I look at an individual bond in Bloomberg and I look up the rate, who knows what I'm seeing? I might be seeing the safest part of the company, right? But if you issue bonds, what else goes with that issue? What can you usually find in the company?
It's not required, but almost every company that has bonds outstanding, you'll be able to find this as well. Remember how we had sovereign ratings? We have corporate ratings, right? S&P and Moody's rates your bonds.
It's not a prerequisite, but generally if you're going to issue bonds, you're going to get a rating. And if you get a rating, I'm home free, and here's why. If I trust the ratings engine, I have to put that precursor, and you have a triple B rating as a company.
Do you have a rough sense of how much a default spread should be? Yeah, there are other BBB rated bonds out there. I can look up the typical default spread and add that spread on. So if you have bonds outstanding or a bond rating, my life got simple.
85% of the companies in the world don't have bonds or ratings. They borrow from banks, and it's only bank loans. There's no rating. So I'm going to take you through a process of how I estimate ratings for companies where I cannot find an actual rating. I'm going to call it a synthetic rating to kind of illustrate the fact that this is a rating I'm imposing.
It's not coming from Moody's. But the way I'm going to do this is I'm going to act like a ratings agency. We think ratings agencies is mysterious, complex organizations, but what they do is actually very simple. They take your financials as a company. They compute ratios.
In fact, they're very open about which ratios they compute. You go to the S&P website, they list the eight ratios they look at. Interest coverage, EBITDA to fix charges, debt to, you know.
So they take the eight ratios, they compute them for your company. And based on how strong you look on those ratios, they give you a rating. But because they can't give away the source, they claim that there's this magical thing that happened.
So all you observe is the name of the company and the rating. They said, we do special things in Moody's to look at this. So about 20 years ago when I started thinking about ratings, here's what I did.
I said, okay, you won't tell me, I'll back it up. I'll do some reverse engineering. So you know what I did? I collected the ratings for every rated company.
So I had the spreadsheet, I had 1,500 companies and the rating for every company. Then I took the eight ratios they claimed they used, because they told me what the ratios were, I collected the eight ratios. So now I have an Excel spreadsheet with the ratings and the eight ratios. ratios, and I did a sort based upon the rating from triple.
You know what I'm trying to find out? Is there something I can get from the data that will help me decide? And it turns out that seven of these ratios are pure noise.
So why would you throw them in there? You have to throw them in there to make it more complex, because then people can't imitate you. Because if everybody knew you were using one ratio to do the bulk of your ratings, why would they pay you to rate them?
They would rate themselves. And as soon as As soon as you look at the numbers, it turns out 50% of the variation in ratings across companies comes from variation in one ratio, the interest coverage ratio. For those of you who don't remember what the interest coverage ratio is, it's EBIT, earnings before interest and taxes, divided by interest expense. Now, the higher that number is, the bigger the buffer you built as a company and the safer you are. So here's what it is.
For Embraer, that number is 3.56 if I take the EBIT and divide by the interest expense. You say, now what? You say, I don't know whether 3.56 is high or low.
Remember those 1500 companies I had? When I looked at those ratios, I created a lookup table based on the interest coverage ratio. You see lookup tables in Excel? Basically, you tell me what your interest coverage ratio is, I guess your rating.
And this table I updated at the start of every year on my website where you tell me, I compute your interest coverage ratio, let's say 3.56, and you're an emerging market company, the rating, so the second column is for riskier companies, the first is for larger, safer companies. is a riskier emerging market company with a 3.56 interest coverage ratio, the rating I would give them would be double B. And essentially, I would use this for any company.
So if you have a company on which you don't have a rating, don't give up yet. You can compute the interest coverage ratio. All you need is interest expense and EBIT from last year.
You have the lookup table. Using the lookup table, you can come up with a rating. And once you come up with a rating, you're home free because it's a default spread that goes with that rating. At the end of the game...
you have to end up with a cost of debt for the company. One final point and we'll end for the day. So I can do this for Embraer to come up with a default spread, right? But Embraer is a Brazilian company.
You think, so what? You know, for a long time, the ratings agencies had what was called a rating ceiling. Does anybody know what the rating ceiling was? No company in a country was allowed to have a rating higher than the country. Imagine how unfair this is going to seem to a...
Turkish company. But they actually had a hard ceiling. Now the hard ceiling is gone because it really never made sense. But they still factor in where you're from. So if you're a Turkish company as opposed to a German company with the same interest coverage ratio, guess what?
You'll have a higher cost of debt. So here's how I bring it in. If you ask me to estimate the cost of debt for Embraer, I'll start. And let's say I'm doing it in dollar terms.
I'll start with the risk-free rate in dollars. I'd add the default spread for Embraer based on the rating. And then I'm going to add a portion or all of Brazil.
Remember the country, the default spread for Brazil is the country. I'm going to say, look, you're going to carry two burdens on your shoulder. One is a company default risk. The other is a country. And I'm doing this because this is the way the real world operates.
So I'm going to stop there. When we come back, we're going to mop up the last of cost of debt and come up with the cost of capital. And then we're ready for cash flows.
You should, because none of that has any effect on beta. You should. It's just that with the regression beta, it should have a stronger... That's like the Nokia versus... So it's basically a market construct.
It means nothing. It doesn't mean the company is less risky or more risky. It's just cross-subsidizing risk because they cross-help. It's really...
Yeah, exactly. Okay. Thank you. It doesn't matter. It's got nothing to do with where it operates.
Remember, if I... No, wait, wait. It doesn't matter.
Where it sells will show up in the risk premium. It should have absolutely no impact on the beta. It doesn't matter, right?
It doesn't matter. The beta measures risk to a diversified investor. So if I'm a diversified investor and I'm buying the Nigerian company, it doesn't matter to me where it sells the stuff. It affects it through the risk premium.
The risk premium is the place for country, absolutely, right? So my point is if you try messing with the beta to make it higher for a Nigerian company, you have two problems. One is you're double counting and second is you're going to lose a fundamental equation that has to hold across beta.
which says they have to average to one. So you start nudging up betas through emerging market companies, you either have to nudge down betas for other companies, or you're gonna have an equation problem, right? So you can't mess with betas.
Betas are not meant to carry where you're exposed to risk. The risk premium will take care of it. So you have two Nigerian companies.
One gets all its revenues in Nigeria, the other gets half its revenues in Europe and half in Nigeria. The latter is gonna have a lower cost of equity because the equity risk premium is gonna be lower, not because its beta's gonna be lower. Thank you. You're welcome.
Yes? I had a question about the synthetic, I guess, rates that you calculated for this, for the beta, as well as for the risk-free rate. I'm curious, like, Now wait, for the base, for the synthetic, you started with the word synthetic, what's base, what is this?
Sorry, the rating. Okay, the synthetic rating. I guess I'm just curious if you would ever, like, back, like if you would check, like, the rating. That is currently available.
You could, you could, but why waste your time? What difference is it going to make to your cost of capital? If you've got the rating to be single A and it happens to be triple B, your cost of capital hardly moves.
The cost of debt is not a big... enough player in this game for you. So I could compete Ali Baba's cost of debt precisely, but it's going to make absolutely no difference to the end value because debt is only 5% of its overall capital. And who cares whether the cost of debt is 6% or 5.8% or 5.3%. So this is again with cost of that what you want is a rough approximation So you can move on when you're a fixed income person you can sit there and finesse it and get the third That's because you're buying bonds your party.
I can understand, but we're not buying bonds. We're valuing equities We're valuing businesses and a rough approximation Yeah, same reason why you applied a lambda to the equity risk premium is because Embraer gets so much of its revenues outside Brazil, people lending to Embraer don't feel as much risk as they would lending to a Brazilian company with all its revenue. Exactly.
So that's my...in fact, if you feel the two-thirds is kind of arbitrary, take it out. Just add the entire country to first. Once I got the double B rating, the default spread with that rating is what I've added on.
Thank you. Okay. Professor, where do you post the solutions for a weekly challenge? I'll have to add the link. It'll show up on the webpage.
Did I not add the link for the first week? I couldn't see it on your website. Maybe I forgot.