Beta Calculation
While you can check the beta calculation in my previous post,
I do want to point out some changes in the process here, and an IMPORTANT common sense
check you should conduct.
- The past decade
has been when people get used to the low interest rate and bond
yield. The US 10-year government bond yield has been between 0-2% during
this period. However, governments worldwide flushed generous
stimulus checks, and benefits have kept the inflation high, making the
global bond yield jump since a higher rate of return is necessary to
attract bondholders when the inflation is around 5-8%. Then, the future
bond yield will depend on how much the unexpected inflation will be.
Now, it becomes a personal
judgment question, just like most decisions to make in finance, and I believe that the inflation will be long-lasting for a while, and the
long-term US 10-year bond yield will be around 5%.
I do want to caution that the
government will never be able to determine the interest rate. The government
bond yield reflects the expected GDP growth rate in addition to inflation.
Academic professionals and financial institutions are obsessed with the idea that the central bank can decide the interest
rate so that the recession, economic booms, and unemployment. But when I added
the GDP growth rate to the inflation rate, I found that the Federal Reserve System is not
so much determining the rate as following that rate. Financial institutions boom the public with Fed interest rate decisions because they can have a scapegoat by doing so. Whenever the equity stock market
slumps, they blame the Federal Reserve Bank for
increasing too high. However, I have never heard of any institutions, regardless
of JP Morgan, Goldman Sachs, or Morgan Stanley, saying they did so much
merger and acquisition in the past decade thanks to the Fed keeping the interest
rate low.
2. Technology companies have more ways to record expenses, such as R&D expense one-time change, than food wholesale companies I valued last time. In other words, their reported earnings are more likely to be off the reality. So, I need to adjust operating income with R&D amortization for each company, just like I did to Oracle in my last post.
The
following graph shows the default spread based on interest coverage from New York
University Stern Business School.
Figure 1 shows the interest coverage ratio, default spread, cost of debt, and market value of debt for CommVault Systems, Inc., one of the data management companies I picked as the sample to estimate Oracle beta. Without R&D adjustment, the company has a default spread of 7.73% since the company has negative for most of the time. And you can tell that the negative earnings in 2023 primarily affect our estimation of the company’s risk.
Figure 1
The market value of debt and default risk before R&D
adjustment
(No debts except leases are found on the company
report.)
However, after we adjusted for R&D expenses, you
can tell that the company has a high interest coverage ratio except for 2018.
(There are no interest expenses from 2019-2021 and 2014), reducing the default
spread to 0.69% and largely increasing the Market Value of Debt. So, if we
don’t adjust income for CommVault Systems, we can underestimate its debt/equity
ratio. That sounds kind of counterintuitive since how could the company have
MORE market value of debt while it became LESS riskier. The reason is that if the
company has a 7.78% default spread, as calculated in Figure 1, few lenders
will lend them loans and few lessors will lease them operating leases. And the
interest expense will be much more than the $0.472 million in Figure 2. Just
like a person with a 300 FICO score will be unlikely to get loan approval and
will pay much more interest expense for a 10k loan than a person with a 500 FICO
score borrowing the same amount.
Figure 2
After-R&D adjustment
Companies with some one-time charge
will also distort our estimation.
Figure 3 shows
the differences for Alteryx, Inc., another one of 20 data management companies I
picked to estimate Oracle’s industry beta.
As you can see from the median and average, if we use
accounted reported numbers,
The most recent year's interest expense primarily affected
Alteryx’s interest coverage ratio, and using the average will give few meaningful
data for us.
Figure 3
Calculation based on accountants’ numbers.
But if you check Figures 4 and 5,
you will find that the main reason for the negative interest coverage ratio
for the year 2022 is a single one-line item called “impairment of long-lived
assets”, and you can check Figure 5 about what that means if you are
confused. The accountant reported the expense when the company stopped using some
of its corporate offices, but that is not going to happen every year and is not
even really an operating expense. And I
checked the company’s previous annual report, and it does not have such an item
until 2020. So, this is a one-time expense that the accountant used to reduce
the company’s tax liabilities. If we include it, we will overestimate the company risk to estimate the company default spread over the long term.
Figure 4
Figure 5
In Figure 6, I added back the one-time charge. You can
see that the average and median are much closer to each other, so I am
comfortable using either of them. You will also see that the default spread and
the market value of debt changed a lot after the adjustment.
Figure 6
Calculation after adjustment
As I become more and more proficient at valuing the
market value of debt, I don’t need a lot of time to value the market value of
debt for 30 companies. The cost of using the accountants’ numbers will be much
more detrimental to my valuation.
More explanation about debt calculation
In my previous post about Costco's stock valuation, I
included the following formula. But I do think I need to give you more explanation to better understand debt so that you can check if you are right when you do your calculation.
Figure 7
When people buy bonds, they receive coupon payment
each period if it is a coupon bond and then receive the principal at the end.
For example, if you buy a bond with a 10% annual coupon rate and $1000 face
value, you will receive $100 each year for 10 years and the $1000 at the end of
10 years. So the market value of the bond = Present value of all future coupon
payments + Present Value of the face value. I use this formula to calculate the
market value of total debt by treating it as a giant bond with annual interest
expense as coupon payment, book value of debt on balance as the face value, and
pre-tax cost of debt as the discount rate.
After you do your calculations, you can check if you
are right if you understand this idea.
The following Figure is the market value of debt of SAP, the
renowned German technology, I calculated. Notice that I used the German
10-year government bond rate rather than the US here since it is a Germany-based
company. After the calculation, I found that the present value of
future SAP debt is larger than the current book value. It feels
counterintuitive at first thought since we usually expect the PV to be
lower. Think about it. How much would you pay today to receive $10k after 5 years? You will expect a number lower than 10k, right?
So why do I get such a high present value of future debt?
Figure 8
My first reaction is that the interest expense I used
in Figure 8 is the company’s most recent annual interest expense since it is
easiest to get and a company’s interest expense changed few over time. But if
you see Figure 8, you will find that SAP’s 2022 interest expense is much higher
than the average.
So I decided to use the average interest expense for the
past 10 year, and I got a much lower present value of debt, but is still
higher than the book value (face value) of debt. Soon, I found the real
reason…
Figure 9
If you use the recent interest expense in Figure 8,2205/10764, you will get a very high 20.48% interest rate. But even if you use the
603.89/10764, you will get a 5.61% interest rate, which is still much higher
than the SAP’s cost of debt. So, the result makes sense. In the bond market, if
your coupon rate is higher than the market interest rate with the same maturity,
your bond will be traded at a premium. So, if your bond has a face value of
$1000 with a 6% coupon rate, while the market rate is 4%, the market value of
your bond will be higher than $1000 since buyers can receive more payment by
buying your bond than other securities in the market. In contrast, if your bond
has a 4% coupon payment while the market rate is 6%, your bond will be traded
at a discount.
These are
the jobs you need to do in financial analysis. I recall that when I took a
finance class from a professor called Oludamola Durodola
in the University Canada West. In each class, he will teach you several financial
formulas and accounting ratios, and give you an example, and say “now, do your
calculation and put your answer in the chatroom”. His whole class through the
semester has the process, and now you can tell that it is not a finance class.
That is just a primary or middle school mathematics class, depending on whether you
receive education in Asia or North America. In this industry, they are people
claiming that Finance is all about numbers and they are people talking about
ideas with no data backup on CNBC all the time. But to succeed in financial
analysis, you need the concept, data, and commonsense simultaneously.
If the interest coverage before adjustment has been
over 6.5, giving us a 0.69-0.89% default spread, then we don’t need to do an adjustment since the effects on the market value of debt will be few or none.
While I thought I finally got my work done, I found
that there is an issue in my calculation. My default spread, a measure of how
risky the company’s debt is related to the risk-free rate, does not consider the market value of operating leases. Then, I found I was in a dilemma:
To calculate the market value of operating leases, I
need the cost of debt.
To estimate the cost of debt, I need to know the interest
coverage ratio (operating income/interest expense).
To know my total interest coverage ratio, I need to
know not only my interest expense for loan but also for my operating leases.
If I want to know the interest expense for my
operating leases, I need to know the market value of operating leases.
It
is a similar situation for most international students in the US. If you want
to work in the US, you need work sponsorship from your employees. If you want US
employers to hire you, you must show that you don’t need work sponsorship…
But personal investment has more solutions than life! This link
from NYU professor Aswath Damodaran allows you to calculate all items above in
one Excel. Make sure to enable the option function in excel; otherwise, you will
be like an international graduate in the US!
With
all these adjustments above, I calculated the market value debt, median
debt/equity ratio, median cash ratio, and median beta for 22 companies similar
to Oracle, as the Figure below shows.
Figure 10 Oracle industry beta
The
Marginal tax rate is made of the marginal US federal and state corporate tax rates. Some companies, such as SAP and Infosys, have different tax rates
accordingly based on their headquarters. Hewlett Packard Enterprise does not
have state corporate tax since it is in the great state of Texas.
With this information, I can estimate Oracle’s beta based on the process in my
previous post.
(Search Beta Calculation to quickly locate the explanation). And
the Oracle beta I estimated is 1.24, while the beta on Yahoo Finance is 0.99.
If you read my previous post, you learned that if the beta equals 1, the
stock has the same risk as the overall market. If Beta is smaller than 1, the stock has a lower risk than the market and vice versa. I feel comfortable using the 1.24 beta since I calculated them by myself.
If I am wrong, I will not feel too bad since I have no one to blame. If my
estimation is closer to the fact, neither will I feel shy to take full credit.
In
fact, I can take a guess about why the beta on Yahoo Finance is 0.99. The past
decade of the stock market is mainly driven by technology stocks, while the beta
on financial institutions and platforms such as Yahoo Finance and CNBC is based on the regression of historical data. So, when you calculate the correlation between a
technology stock and a market that has been driven by technology stocks for a
decade, no wonder you will get a beta close to 1. But what happened in the past
does not predict the future.
Then, you may wonder how I can use the median beta from my samples. Because the law of
large numbers in statistics helps. Just like if you want to know the average
height of people. Someone is really tall, and someone is short. But they
canceled out when your sample size was large enough.
Figure 11 Oracle Beta Calculation
Final
reminders (Important!)
I
need to point out that you need to be very careful about selecting your
samples. Your sample should be representative and large enough. I remember once
I was watching an online representation in which Professor Oludamola Durodola
was evaluating a cannabis stock to the University Canada West board. (Don’t ask
me why he picked the cannabis stock). And he compared the stock to the Royal
Bank of Canada. I found something wrong, so I asked him why he compared the RBC
to the cannabis business when they have totally different business models, risks, and
profit margins. So, if a person has a little, tiny financial knowledge, he
still can cover his mistake in one way. He can say, “Because the bank stock has
over 30% weight of the Canadian stock market, I just use RBC to represent the
whole market to take a shortcut.” But he said, “Oh, yeah. So, when I compare, I
need to compare something different, right? So that’s exactly why I compare it
with RBC. Good job, Zachary.” Hearing this, I can conclude that he knows nothing
about finance, although the school board will never realize it.
So,
when you pick a sample, you must pick some companies that are similar to your
company. While Oracle is a technology company, not all technology companies are
Oracle’s peers. Oracle focuses on data management, enterprise systems, and
hardware. So, SAP, Microsoft, and Cisco can be their peers, but Amazon and Google may
not. By the way, that professor also brags about how he used to work for CIBC
as a financial advisor. So, next time, you can use a similar way to test your
financial advisors or asset managers. If your asset managers said something
like, “We don’t buy stocks like Walmart or Costco, their returns are too low. We
invest in high growth companies. You know, last year I helped another client
earn 100% on Tesla.”, then you can know that he knows nothing about investment,
and he is bluffing. You don’t need to try it with financial advisors, though, since they are salesman, and they are expected to know nothing about
investment.
Comments
Post a Comment