See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/260088646

The Valuation of Internet Companies

Article · January 2014

CITATIONS READS

0 406

1 author:

Ako Doffou
The Institute of International Studies
22 PUBLICATIONS   23 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Pricing Currency Options under Stochastic Interest Rates and Jump-Diffusion Processes View project

All content following this page was uploaded by Ako Doffou on 10 February 2014.

The user has requested enhancement of the downloaded file.

 The Journal of Applied Financial Research, Volume I 2014

The Valuation of Internet Companies

Ako Doffou, PhD, CFA

Visiting Full Professor of Finance, The Institute of International Studies,

Ramkhamhaeng University, Bangkok, Thailand

Abstract:

This paper uses cross-sectional quarterly data to estimate the parameters of the
Schwartz and Moon (2000) model and to price five well known internet companies. The
model is drawn from real options theory and capital budgeting techniques. Both
continuous and discrete time versions are proposed. The cross-sectional quarterly data
used are those available from a sample of twenty internet companies. Judgment and
knowledge of each company and its industry are critical to infer some of the parameters.
The model is solved by simulation. The value of an internet stock is highly dependent on
the initial conditions, the exact specification of the parameters chosen, and the assumed
growth rates in sales or revenues.

Key Words: Internet Companies; Real Options; Continuous Time; Discrete Time;
Simulation.

JEL Classification: G12

1. Introduction

Internet stocks are stocks of companies such as Google, Amazon, eBay, Yahoo,

and Facebook, just to name a few. These stocks have become very popular as

their fast growing values have made the entrepreneurs behind these firms as well

as many employees millionaires and billionaires. Yet, besides the high valuation,

some of these companies have generated substantial and sustained losses for

years.

Some money managers see this unrealistically high valuation of internet stocks

as a vivid example of a market bubble caused by day traders whose continuous

purchases bid upward the prices of these stocks. They fear that the real

economy will suffer severe consequences soon after this bubble bursts. But the

internet is also perceived by many as a powerful catalyst in the transformation of

business transactions. The internet has generated companies which have grown

very rapidly to dominate the market in their respective industries.

To appease the fear of the traditionalists, a correct valuation of internet stocks is

needed. The motivation of this paper is to show that the Schwartz and Moon

(2000) model can be extended to value the stock of even well matured internet

companies and to prove that such value is dictated by fundamentals, not by

market bubbles as the traditionalists may believe.

The proposed pricing model for internet companies uses real options theory and

modern capital budgeting techniques and is formulated both in continuous time

and discrete time. The model parameters are estimated using a cross-sectional

data from a sample of 20 internet companies which include Google, Amazon,

2
eBay, Yahoo, Facebook, and many more. Finally, the model is solved by

simulation and some sensitivity analyses are carried out.

2. The Model

Suppose the internet company to be valued generates sales at time t given by

S t . The sales at time t are also the instantaneous rate of revenues at time t . The

dynamics of these revenues are assumed to follow the stochastic differential

dS t
equation = µ t dt + σ t dZ1 , (1)
St

where the drift term µ t represents the expected rate of growth in sales, σ t the

volatility of the rate of sales growth, and dZ 1 is an increment of a Brownian

motion. The drift is assumed to exhibit mean reversion and is pulled over time to

its long-term mean value µ at a certain speed. This speed of adjustment is

dictated by the competitiveness within each industry. The initial super high

growth rates of the internet company will converge to a more reasonable and

lower sustainable rate of growth characteristic of each industry. Hence,

( )
dµ t = K µ − µ t dt + δ t dZ 2 (2)

where the initial volatility of the expected rates of growth in sales is given by δ 0

and K is the speed of adjustment. This speed of adjustment or speed of mean

reversion translates the rate at which the growth in sales is expected to converge

to its long-term mean value µ .

The unanticipated changes in the drift and the unanticipated changes in sales are

assumed to converge deterministically to zero and to a more normal level

3
respectively: dδ t = − K 2δ t dt ; . (3)

Furthermore, the unanticipated changes in the drift and those in the growth rate

of sales may be correlated: dZ1dZ 2 = ρdt . (4)

The after tax net rate of cash flow to the internet company, NRCFt , can be

expressed as follows NRCFt = (S t −C t )(1 − TRc ) , (5)

where C t is the costs at time t and TRc is the corporate tax rate. The internet

company pays taxes only when there is no loss carry-forward. When the loss

carry-forward is positive, the tax rate is zero and no tax is paid. The costs at time

t are composed of the cost of goods sold (COGS) which are assumed to be

proportional to sales and of other expenses with a fixed component and a

variable component proportional to sales:

C t = COGS t + OtherExpensest = αS t + (F + βS t ) = (α + β )S t + F , (6)

where α expresses COGS as a percentage of sales and β is the element in

other expenses that is variable.

The dynamics of the loss carry-forward, LCFt , can be formulated as follows:

dLCFt = − NRCFt dt if LCFt > 0 (7a)

or dLCFt = max (− NRCFt dt ,0) if LCFt = 0 . (7b)

The internet company has a certain amount of cash available at time t , CASH t ,

that follows the process dCASH t = NRCFt dt . (8)

Technically, the internet company is bankrupt when it has zero cash that is when

CASH t in the model reaches zero for the first time, assuming there is no

possibility of additional financing in the future. Furthermore, the firm’s operating

4
cash flow stays within the firm, yields the risk-free rate of interest, and is

available to shareholders in the long run at time T when the firm reverts to a

regular company with its industry growth rate. The internet company is not likely

to begin paying dividends until its cash flows are strongly positive and expected

to remain positive for the foreseeable future.

Under the equivalent martingale measure (risk-neutral probability measure) Θ ,

the value V0 of the internet company at the current time zero is the expectation of

the discounted net cash flow to the company at the risk-free rate:

(
V0 = E Θ CASH t e − rT , ) (9)

where the continuously compounded discount factor is given by e − rT . Equation 9

implicitly assumes that the internet company is liquidated at the horizon T and all

cash flows are distributed.

The model is characterized by two sources of uncertainty, the uncertainty related

to the changes in sales and the uncertainty related to the expected rate of growth

in sales. Using Brennan and Schwartz (1982) simplifying assumptions, the risk-

adjusted processes of the state variables under the risk-neutral probability

measure Θ are:

dS t
= (µ t − γ 1σ t )dt + σ t dZ1Θ , (10)
St

[( ) ]
dµ t = K µ − µ t − γ 2δ t dt + δ t dZ 2Θ , (11)

with dZ1Θ dZ 2Θ = ρdt , (12)

where the market prices of risk γ 1 and γ 2 are constant. The value of the internet

company at any time t depends on the state variables and time. The state

5
variables are sales, expected growth in sales, loss carry-forward, and cash

available). Hence, Internet Company Value = V ≡ V (S , µ , LCF , CASH , t ) . (13)

The dynamics of the value of the internet company is provided in equation (14)

below obtained by applying Ito’s lemma to equation 13:

1 1
dV = V S dS + Vµ dµ + V LCF dLCF + VCASH dCASH + Vt dt + VSS dS 2 + V µµ dµ 2 + VSµ dSdµ .
2 2

The volatility σ V2 of the value of the internet company is given by

2
V Vµ
2
1  dV   VS   V µ 
σ = var
2
 =  σS  + δ  + 2 S 2 Sσρδ . (15)
V   V
V
dt  V    V

As shown in equations 14 and 15, the model can be used to assess the value of

the company and its volatility. The cash available at any time and the loss carry-

forward are path dependent. These path dependencies can be easily accounted

for by using a Monte Carlo simulation to derive the value of the internet company.

To implement the Monte Carlo simulation, the discrete version of the risk-neutral

process given in equations 10-12 is used:

S t + ∆t = S t e
[ ( )]
{ µt −γ 1σ t − σ t2 / 2 ∆t +σ t ∆tω1 }
(16)

γ 2δ t  1 − e −2 K∆t
µ t + (1 − e )

and µ t + ∆t = e − K∆t − K∆t
µ − + δ t ∆tω 2 , (17)
 K  2K

( )
where σ t = σ 0 e − K1t + σ 1 − e − K1t , δ t = δ 0 e − K 2t , and ω i for i = 1,2 are correlated

standard normal variates with Corr(ω1 , ω 2 ) = ρ . The expressions of σ t and δ t

are obtained by integrating equation 3 and setting the initial values at σ 0 and δ 0 .

A discrete version similar to Equation 17 is proposed in Schwartz and Smith

(1997). Quarterly accounting data available from internet companies are used to

6
implement the model in discrete time.

Under the discrete version of the model, the after tax net cash flow is still

provided by equation 5 where sales and costs are measured within the period of

time ∆t . The dynamics of the loss carry-forward and the level of cash available

can be directly derived from equations 7 and 8 respectively.

3. Parameters Estimation

To implement the model, its parameters need to be estimated. Some of the

parameters are observable and those not observable are estimated using cross-

sectional data from a sample of twenty internet companies. The internet

companies considered in the sample include Google, eBay, Amazon, Yahoo,

Facebook, and fifteen other internet companies. As in Schwartz and Moon (2000),

the estimation of some parameters requires the use of judgment based on the

unique knowledge of each company and its industry. The initial sales value, S 0 ,

for each company is observable from current income statements. The initial loss

carry-forward LCF0 and the initial cash available CASH 0 are observable from

current balance sheets. The initial expected rate of growth in sales, µ 0 , is

estimated from past income statements and projections of future growths. The

standard deviation of the percentage change in sales over the recent past gives

the initial volatility σ 0 while the initial volatility of expected rates of growth in sales,

δ 0 , is inferred from the market volatility of the stock price. The correlation ρ

between the percentage change in sales and change in expected rate of growth

is estimated from past company and cross-sectional data. The rate of growth in

sales for a stable company in the same industry as the internet company being

7
valued is used as the long-term rate of growth in sales µ . The volatility of

percentage changes in sales for a stable company in the same industry as the

internet company being valued is used as a proxy for the long-term volatility σ of

the rate of growth in sales. The corporate tax rate TRc applicable to each internet

company is readily available from the tax code. The one year U.S. T-bill rate is

used as a proxy for the risk-free interest rate. The speed of mean reversion for

the rate of growth process K and that for the volatility of sales process K1 as

well as that for the volatility of the rate of growth process K 2 are estimated from

assumptions about the half-life of the process to µ , to σ , and to zero

respectively. The cost of goods sold (COGS) as a percent of sales α , the fixed

component of other expenses F , and the variable component of other expenses

β are estimated from Analysts’ future projections. The market price of risk for the

sales factor γ 1 is derived by multiplying the correlation between the percentage

changes in sales and return on aggregate wealth by the standard deviation of

aggregate wealth. The market price of risk for the expected rate of growth in

sales factor γ 2 is derived by multiplying the correlation between changes in

growth rates in sales and return on aggregate wealth by the standard deviation of

aggregate wealth. The estimation horizon T is an arbitrary long-run horizon at

which the internet company is assumed to be considered a “normal” company

growing at a stable and reasonable rate. Finally, the time increment ∆t for the

discrete version of the model is usually quarterly which is dictated by data

availability.

8
4. Simulation

The model is used to value five of the best-known, well managed, largest internet

companies in the world: Google, eBay, Amazon.com, Yahoo, and Facebook. The

basic data collected for each company include quarterly sales; COGS; gross

profit; selling, general, and administrative expenses; and operating profit before

taxes (EBITDA) from April 1, 2009 to April 1, 2013). Moreover, balance sheet

data are used to estimate the loss carry-forward and the cash available. The

parameters used in the basic valuation of Google, Amazon, Facebook, Yahoo,

and eBay are shown in Table 1. Some of the parameters used in the valuation of

these five internet companies come from the financial statements or are

otherwise directly observable. Some of these parameters are estimated from

past data, and some from future projections. Next, some sensitivity analyses are

carried out to assess how each internet company value varies following changes

in the estimated parameters. The initial expected rate of growth in sales is the

average growth rate over the last four quarters. The rate of growth over the next

four quarter is the average of the expectations of the analysts polled by Thomson

Reuters I/B/E/S International. The initial volatility in sales is the standard

deviation of past percentage changes in sales. The initial volatility of the

expected rate of growth in sales is derived from the observed stock price volatility.

The changes in expected growth rates and the percentage changes in sales are

assumed to be uncorrelated. The long-term rate of growth in sales is the industry

rate of growth which is set at 1.25 percent per quarter (5 percent per year). The

long-term volatility of the rate of growth in sales varies from 5 percent to 4

9
percent per quarter depending on the internet company. The mean reversion

coefficients or speed of adjustment coefficients are derived by assuming that the

half-life of the deviations can be approximated by 10 quarters. Based on the

available data for the last four quarters, the cost of goods sold (COGS)

parameter is assumed to be 74% of sales, 43% of sales, 32% of sales, 30% of

Table 1: Base Valuation Parameters

.

Parameter Notation Proposed Estimation Procedure .

Amazon Google Yahoo eBay Facebook

Initial Sales (billion $) S0 $16.07 $13.97 $1.14 $3.75 $1.46

Initial Loss Carry-Forward (million $) LCF0 $39.00 $0.00 $0.00 $0.00 $0.00

Initial Cash Balance Available (billion $) CASH 0 $4.481 $15.375 $1.175 $6.53 $2.325

Initial Expected Rate of Growth in Sales µ0 0.12 0.05 0.015 0.04 0.08

Initial Volatility in Sales σ0 0.10 0.08 0.08 0.08 0.10

Initial Volatility of Expected Rates of Growth in Sales δ0 0.03 0.025 0.025 0.025 0.03
Correlation between Percentage Change in
Sales and Change in Expected Rate of Growth ρ 0.00 0.00 0.00 0.00 0.00

Long-term Rate of Growth in Sales µ 0.0125 0.0125 0.0125 0.0125 0.0125

Long-term Volatility of the Rate of Growth in Sales σ 0.05 0.045 0.04 0.04 0.05
Internet Company’s Corporate Tax Rate TRC 0.35 0.35 0.35 0.35 0.35
Risk-free Interest Rate r 0.05 0.05 0.05 0.05 0.05
Speed of Adjustment for the Rate of Growth Process K 0.07 0.06 0.06 0.06 0.07
Speed of Adjustment for the Volatility of Sales Process K1 0.07 0.06 0.06 0.06 0.07

Speed of Adjustment for Volatility of the Rate of Growth P. K 2 0.07 0.06 0.06 0.06 0.07
Cost of Goods Sold (COGS) as an Element of Sales α 0.74 0.43 0.32 0.30 0.28
Fixed Component of other Expenses (billion $) F $3.89 $2.67 $0.406 $1.29 $0.48
Variable Component of other Expenses β 0.19 0.22 0.22 0.22 0.22

Market Price of Risk for the Sales Factor γ1 0.01 0.01 0.01 0.01 0.01
Market Price of Risk for the Expected Rate of
Growth in Sales Factor γ2 0.00 0.00 0.00 0.00 0.00
The Estimation Horizon (in years) T 25 25 25 25 25
Time Increment for the Discrete Version (in months) ∆t 3 3 3 3 3
.
Unless otherwise stated, all data reported here are per quarter. The risk-free interest rate is an annual rate.

sales, and 28% of sales for Amazon, Google, Yahoo, eBay, and Facebook

respectively. The fixed component of other expenses is derived from each

10
internet company’s income statement for the last four quarters. The variable

component of other expenses is a low percentage of sales (19% and 22%)

consistent with the historical dada of the last four quarters.

The two market price of risk factors are estimated using a 5 percent per quarter

standard deviation for aggregate wealth, a correlation of 0.2 between the return

on the aggregate wealth and the percentage changes in sales, and a zero

correlation between the aggregate wealth and the changes in growth rates.

Finally, the parameters are estimated over a 25 year horizon, with a time

increment of three months or one quarter. The terminal value for each internet

company at the 25 year horizon is assumed to be approximately equal to 10

times each company’s pretax operating profit (EBITDA), consistent with the

approach frequently used by finance practitioners.

The valuations of each internet company considered using the proposed model

require 100,000 simulations with an aggregate of 500,000 simulations for all five

companies. The base valuation using the parameters provided in Table 1 and as

of the 17th of May 2013, gives a value of $120.87 billion for Amazon, $300.76

billion for Google, $27.91 billion for Yahoo, $72.77 billion for eBay, and $62.91

billion for Facebook. These values are achieved even though there is a non zero

probability for each company to go bankrupt. The probability for bankruptcy per

year for the base valuation is listed in Table 2. Bankruptcies occur only in year 5

when each firm runs out of cash. No bankruptcy occurs after year 16 for Google,

year 18 for eBay, year 19 for Amazon, year 20 for Yahoo, and year 21 for

Facebook. The probability of bankruptcy is the highest in year 6 but decreases

11
slowly thereafter.

The effects of changed parameters on the value of each internet company are

displayed in Tables 3 to 7. The numbers are generated by creating a perturbation

which often here is a 10 percent increase in the value of a given parameter while

leaving all the remaining parameters unchanged, that is keeping them the same

Table 2: Probability of Bankruptcy per Year for the Base Valuation

.
Year . Bankruptcy .
Amazon Google Yahoo eBay Facebook .
1 0.0% 0.0% 0.0 0.0% 0.0%
2 0.0 0.0 0.0 0.0 0.0
3 0.0 0.0 0.0 0.0 0.0
4 0.0 0.0 0.0 0.0 0.0
5 3.4 2.3 3.5 3.3 3.6
6 5.2 2.7 5.3 5.1 6.1
7 3.7 3.1 3.8 3.7 5.3
8 2.1 1.1 2.3 2.2 3.8
9 0.8 0.7 1.1 0.7 2.9
10 0.6 0.5 0.8 0.5 2.4
11 0.4 0.3 0.5 0.4 1.5
12 0.3 0.2 0.4 0.3 1.2
13 0.3 0.2 0.4 0.3 0.7
14 0.2 0.1 0.3 0.2 0.5
15 0.2 0.1 0.3 0.2 0.3
16 0.2 0.1 0.3 0.1 0.3
17 0.1 0.0 0.2 0.1 0.2
18 0.1 0.0 0.2 0.1 0.2
19 0.1 0.0 0.2 0.0 0.2
20 0.0 0.0 0.1 0.0 0.1
21 0.0 0.0 0.0 0.0 0.1
22 0.0 0.0 0.0 0.0 0.0
23 0.0 0.0 0.0 0.0 0.0
24 0.0 0.0 0.0 0.0 0.0
25 0.0 0.0 0.0 0.0 0.0 .
Total 17.70% 11.4% 19.7% 17.20% 29.4% .

as in the base valuation model. The numbers in Tables 3-7 show that two

classes of parameters have a significant impact on the value of each internet

company examined. The first of these two classes is the variable component of

the cost function given by α + β , equation 6, which is proportional to sales. An

increase in either α or β generates the same effect on the cost function which in

turn affects the value of the internet company. An increase in either α or β

increases the costs function and decreases the value of each of the internet

12
companies examined. The sum α + β in the base valuation model is 93 percent

of sales for Amazon, leading to a profit margin of 7 percent of sales. If either α or

β is increased by 0.01 or 1 percent as shown in Table 3, the profit margin

decreases and the value of Amazon decreases from $120.87 billion to $94.68

billion (a 21.67% decrease). Table 4 shows the sensitivity of Google’s value to

Table 3: Effects of Changed Parameters on Amazon Market Value

.
Parameters Changed Value Total Amazon Market Standard Probability of
of Parameter Value ($ billions) Deviation Bankruptcy
.
Base Valuation Model $120.87 33% 17.70%
µ 0 (per quarter) 0.132 142.77 38 14.46
σ 0 (per quarter) 0.11 118.56 33 18.20

δ 0 (per quarter) 0.033 136.19 43 18.77

ρ 0.01 119.36 33 17.75

µ (per quarter) 0.0138 132.01 14 17.05

σ (per quarter) 0.055 118.36 33 18.06

K (per quarter) 0.077 93.22 23 18.95
K1 (per quarter) 0.077 120.98 32 17.58
K 2 (per quarter) 0.077 111.77 29 17.20
α 0.75 94.68 27 23.46
F (per quarter) $4.279 billion 114.36 33 22.51
β 0.20 94.68 27 23.46
γ 1 (per quarter) 0.011 118.19 32 17.77
γ 2 (per quarter) 0.001 118.06 32 17.77
T (in years) 26 122.35 34 17.84
.

changed parameters. The sum of the two variable cost parameters in the base

example is 65 percent of sales, giving a profit margin of 35 percent of sales for

Google. An increase of 1 percent in either of the variable cost parameters leads

to a decrease in profit margin and a decrease in the value of Google from

$300.76 billion to $235.44 billion (a 21.72% decrease). Similarly, the sum of the

variable costs in the base model is 54, 52, and 50 percent of sales for Yahoo,

13
eBay and Facebook respectively, leaving a profit margin of 46, 48, and 50

percent of sales. As shown in Tables 5-7, an increase of 1 percent in any of

these variable costs leads to a decrease in profit margin and a decrease in the

value of the internet company from $27.91 billion to $21.53 billion for Yahoo (a

22.86% decrease), from $72.77 billion to $56.93 billion for eBay (a decrease of

21.77%), and from $62.91 billion to $47.91 billion for Facebook (a decrease of

23.84%). It appears that the variable components of the cost function have a

significant impact on the valuation model. A poor assessment of these variable

costs leads to an incorrect valuation of the internet company.

Table 4: Effects of Changed Parameters on Google Market Value

.
Parameters Changed Value Total Google Market Standard Probability of
of Parameter Value ($ billions) Deviation Bankruptcy
.
Base Valuation Model $300.76 32% 11.40%
µ 0 (per quarter) 0.055 351.89 37 9.32
σ 0 (per quarter) 0.088 295.58 32 11.73
δ 0 (per quarter) 0.0275 336.96 41 12.09
ρ 0.01 297.20 32 11.43

µ (per quarter) 0.0138 326.33 13 10.98

σ (per quarter) 0.0495 293.69 32 11.63

K (per quarter) 0.066 232.02 23 12.20
K1 (per quarter) 0.066 300.88 31 11.34
K 2 (per quarter) 0.066 276.99 28 11.09
α 0.44 235.44 26 15.13
F (per quarter) $2.937 billion 284.41 32 14.46
β 0.23 235.44 26 15.07
γ 1 (per quarter) 0.011 292.83 31 11.41
γ 2 (per quarter) 0.001 292.51 31 11.41
T (in years) 26 303.07 33 11.45
.

The second class of parameters with a significant impact on the value of the

internet companies are the parameters associated with the stochastic process of

14
the changes in the growth rate in sales given by equation 2. Those of these

parameters that influence the future distribution of the rates of growth in sales

have even bigger effect on the value of the firm. A 10 percent increase in the

initial volatility of the rate of growth δ 0 increases the value of Amazon from

$120.87 billion to $136.19 billion. Similarly, a 10 percent increase in δ 0 increases

the value of Google from $300.76 billion to $336.96 billion, Yahoo from $27.91

billion to $31.47 billion, eBay from $72.77 billion to $81.91 billion, and Facebook

from $62.91 billion to $72.11 billion. A 10 percent increase in the speed of mean

reversion K decreases the value of Amazon from $120.87 billion to 93.22 billion,

Google from $300.76 billion to $232.02 billion, Yahoo from $27.91 billion to

$21.20 billion, eBay from $72.77 billion to $56.06 billion, and Facebook from

$62.91 billion to $49.34 billion. The deterministic speed of adjustment K 2 for the

volatility in the rate of sales growth also has a significant effect on the value of

the internet companies but at a lower degree than the impact of δ 0 and K .

The distribution of future rates of growth in sales is affected by these three

parameters. An increase in the initial volatility of the rate of growth in sales

increases the variance of this distribution. An increase in the speed of mean

reversion or the speed of adjustment for the volatility of this process reduces the

variance of this distribution. The variance of the distribution of future rates of

growth determines the “option value” of the internet company and as such is

critical in the valuation. Higher variance of future growth rates is associated with

higher probability both of very high growth rates and very low or even negative

growth rates. Higher growth rates generate larger cash flows leading to a more

15
 Table 5: Effects of Changed Parameters on Yahoo Market Value
.
Parameters Changed Value Total Yahoo Market Standard Probability of
of Parameter Value ($ billions) Deviation Bankruptcy
.
Base Valuation Model $27.91 34% 19.70%
µ 0 (per quarter) 0.0165 32.99 39 16.09
σ 0 (per quarter) 0.088 27.41 34 20.25
δ 0 (per quarter) 0.0275 31.47 44 20.88
ρ 0.01 27.38 34 19.75

µ (per quarter) 0.0138 30.04 14 18.97

σ (per quarter) 0.044 26.93 34 20.09

K (per quarter) 0.066 21.20 24 21.07
K1 (per quarter) 0.066 27.99 33 19.59
K 2 (per quarter) 0.066 25.42 30 19.17
α 0.33 21.53 28 26.15
F (per quarter) $0.4466 billion 26.00 34 25.09
β 0.23 21.53 28 26.15
γ 1 (per quarter) 0.011 26.82 33 19.81
γ 2 (per quarter) 0.001 26.79 33 19.81
T (in years) 26 28.88 35 19.88
.

valuable internet company. If the growth rates are considerably low, the internet

company may go bankrupt. Stockholders of the internet company have limited

liability which implies that the valuation model is nonlinear. Given a more variable

distribution of future rates of growth, this nonlinearity in the valuation function

leads to a more valuable internet company.

The variance of the distribution of future growth rates is critical to the valuation.

consequently, the parameters of the model should be jointly picked to give a

good picture of what is believed to be a reasonable distribution of future rates of

growth and future sales for a given internet company. The variance of the growth

rate in sales affects the option value of the internet company and the mean of

sales distribution. Based on Equation 16 and Jensen’s inequality, greater

volatility leads to greater mean sales.

16
 Table 6: Effects of Changed Parameters on eBay Market Value
.
Parameters Changed Value Total eBay Market Standard Probability of
of Parameter Value ($ billions) Deviation Bankruptcy
.
Base Valuation Model $72.77 33% 17.20%
µ 0 (per quarter) 0.044 85.87 38 14.06
σ 0 (per quarter) 0.088 71.31 33 17.69
δ 0 (per quarter) 0.0275 81.91 43 18.25
ρ 0.01 71.78 33 17.26

µ (per quarter) 0.0138 79.39 14 16.58

σ (per quarter) 0.044 71.18 33 17.57

K (per quarter) 0.066 56.06 23 18.43
K1 (per quarter) 0.066 72.88 32 17.14
K 2 (per quarter) 0.066 67.21 29 16.77
α 0.31 56.93 27 22.87
F (per quarter) $1.419 billion 68.76 33 21.95
β 0.23 56.93 27 22.87
γ 1 (per quarter) 0.011 71.07 32 17.32
γ 2 (per quarter) 0.001 70.99 32 17.32
T (in years) 26 73.57 34 17.38
.

5. Derivation of the Share Price

The determination of the share price of an internet company requires a close

examination of the capital structure of this company. The number of shares

outstanding, the number of shares likely to be issued to employees who hold

stock options, and the number of shares likely to be issued to holders of

convertible bonds must be clearly identified. Moreover, the cash flow that will be

available to the internet company shareholders after principal and coupon

payments are made to the bondholders must be known.

It is reasonable to assume, in the case the internet company does not go

bankrupt, that options will be exercised and convertible bonds will be converted

17
 Table 7: Effects of Changed Parameters on Facebook Market Value
.
Parameters Changed Value Total Facebook Market Standard Probability of
of Parameter Value ($ billions) Deviation Bankruptcy
.
Base Valuation Model $62.91 35% 29.40%
µ 0 (per quarter) 0.088 75.60 40 24.03
σ 0 (per quarter) 0.11 62.78 35 30.25
δ 0 (per quarter) 0.033 72.11 45 31.20
ρ 0.01 63.19 35 29.51

µ (per quarter) 0.0138 69.88 14 28.35

σ (per quarter) 0.055 62.65 35 30.04

K (per quarter) 0.077 49.34 25 31.51
K1 (per quarter) 0.077 62.97 34 29.30
K 2 (per quarter) 0.077 59.15 31 28.67
α 0.29 47.91 29 39.10
F (per quarter) $0.528 billion 60.53 35 37.52
β 0.23 47.91 29 39.10
γ 1 (per quarter) 0.011 62.55 34 29.61
γ 2 (per quarter) 0.001 62.48 34 29.61
T (in years) 26 64.75 36 29.72
.

into shares of common stocks. Consequently, the number of shares is adjusted

to translate the exercise of options and convertibles in the no-default scenarios of

the simulations. The cash flow available to all securityholders determines the

total value of the internet company. To compute the share price, the cash flow

available to shareholders is needed. It is obtained by subtracting the principal

and after-tax coupon payments on the debt from the cash flow available to all

securityholders and adding the payments by optionholders at the exercise of the

options. The exercise of options and convertibles take place at their maturity

dates because each internet company considered is assumed to pay no

dividends. This approach undervalues the options and convertibles and

overvalues the stock when optionholders exercise their options optimally.

18
Moreover, to be able to sell the underlying stock to diversify their portfolios,

employees more often exercise stock options prior to the maturity date if the

options are exercisable. It is also important to note that many employees leave

the company before they are vested in their stock options. Consequently, not all

the options will be exercised even if they are in the money. The impact on share

value of the number of new shares to be issued at exercise and conversion is

likely to be small if the number of these new shares is small relative to the total

number of shares outstanding.

All the four internet companies examined were valued on May 17, 2013. At this

valuation date, Amazon had 455,242,700 shares outstanding; Google had

331,771,500 shares; Yahoo had 1,082,579,200 shares; eBay had 1,297,831,100

shares; and Facebook had 2,417,904,800 shares outstanding. For each internet

company, the capital structure may consist of equity, convertible bonds, discount

notes, employee stock options, and many more. To arrive at a correct

assessment of the share price, each element in the capital structure must be fully

identified. The face value, coupon rate, maturity date, and the conversion ratio of

all convertible debt issues are determined. The equity value is directly observable

in each company’s balance sheet. The face value and the maturity of all senior

discount notes are clearly assessed. The employee stock options outstanding as

of the date of the valuation are obtained from each internet company’s 10-K form

and were adjusted for a subsequent stock split. The majority of the options

outstanding had average exercise prices far below the stock price of each

company on the valuation date. Consequently, these options are more likely to

19
be exercised if each of the internet company examined survives as a going

concern.

The simulation program is modified to account for shares issued following the

conversion of the convertibles and the exercise of the options and to assess

which part of the cash flow belongs to shareholders. The base valuation gives a

stock price of $248.31 for Amazon. This stock value appears lower than

the market price of $269.90 for Amazon when the market closed on the valuation

date of May 17, 2013. Hence, the model price is exactly 92% of the actual stock

price. The base valuation price obtained for Google is $850.09 which is 93.5% of

its market price of $909.18 at the close of May 17, 2013. A base valuation price

of $24.33 was obtained for Yahoo which is 91.74% of its market price of $26.52

at the close of May 17, 2013. For eBay, the base valuation price obtained is

$52.11 which is 91.89% of its market price of $56.71 at the close of May 17,

2013. Finally, the base valuation price obtained for Facebook is $23.78 or 90.59%

of its market price of $26.25 at the close of May 17, 2013. Overall, the model

pricing accuracy is acceptable.

The total cash flows available to all securityholders are implicitly assumed to be

independent of the capital structure. The model allows bankruptcy to occur when

the cash balances reach zero value. So, if the cash is depleted to pay off all

maturing debts, an equal amount of new debt is issued to restore the previous

cash balances.

Equation 15 gives the volatility of each internet company. The volatility of the

equity is obtained from the same equation 15 in which the company value is

20
simply replaced by the equity value. The partial derivatives of the company value

and equity value with respect to the expected growth rate in sales and with

respect to the level of sales are obtained by simulations. Basically, the initial

value of sales (the rate of growth in sales) is distressed to generate new values

of the company and equity from which these partial derivatives are calculated.

Given the parameters used in the base valuation, a volatility for the equity of

27.71% per annum is obtained for Amazon, 29.52% for Google, 43.89% for

Yahoo, 41.63% for eBay, and 72.34% for Facebook. These volatility numbers are

consistent with the observed historical volatilities of these internet companies in

the year prior to this valuation. These results are based on the volatility of

expected growth rate in sales.

The volatility of the expected rates of growth in sales, δ 0 , is a very critical

parameter in the valuation model and affects the stock price and its volatility. An

increase in δ 0 causes the stock price to increase dramatically and the volatility of

the stock price to increase linearly. All the internet companies valued here are

mature and almost all exhibit substantially higher profitability as opposed to what

their profitability was in their early stage. This high profitability allows the model

to generate prices and volatilities that are consistent with those observed in the

market.

6. Conclusions

The model used here is based on some simplistic assumptions about the optimal

exercise of American-type options. First, it assumes that bankruptcy is only a

function of the amount of cash balances and when that amount is zero, the value

21
of the internet company becomes zero. In fact, the value of the company

depends not only on the amount of cash balances but also on all the other

underlying assets such as the level of sales, the expected rate of growth in sales,

their volatilities, and the amount of the loss carry-forward. If the prospects of the

company are good, it can raise new fresh cash or merge with another company

to survive even though its cash balances are zero. Second, in assessing the

internet company equity value, it is assumed that the options are exercised and

the convertible bonds converted if the company survives as a going concern.

Moreover, the optimal exercise of these options depends on the company value

at the decision date.

To address the issues listed above, a least-squares Monte Carlo approach to

price American-type options by simulation developed by Longstaff and Schwartz

(1998) can be used. The point is to compare the value of immediate exercise

with the conditional expected value of continuation under the risk-neutral

measure for American-style options. The aim here is to assess the conditional

expected value of the internet company under the risk-neutral measure at each

point in time. This procedure optimally determines the stopping time of

bankruptcy and does not stop a particular path when the amount of the cash

balances is zero. If the conditional expected value of the company is less than or

equal to zero, the company is bankrupt and its value is zero. This process starts

from the horizon T and proceeds recursively in the same way up to the present

time, and produces the optimal stopping time for each path from which the

22
current value of the company can be calculated. A similar procedure is followed

to find the optimal exercise of the options and convertibles.

The valuation model proposed here is based on assumptions about the expected

growth rate of sales or revenues and on expectations about the cost structure of

the internet company. The model generates internet company values and stock

prices that can be very volatile because based on expectations which can

change on a continuous basis as new information is revealed to the market.

The implementation of the model requires several assumptions about possible

future financing, the horizon of estimation, future cash distribution to bondholders

and shareholders, and much more. The model is flexible enough to incorporate

alternative assumptions in the overall analysis. The use of this model requires

more reasonable and realistic assumptions dictated by a sophisticated

knowledge of the internet company and its industry.

The value of the internet company given by the model is sensitive to the initial

conditions and the exact specification of the parameters. The value of an exit

option for the internet companies is also investigated. The exit value in the model

is assumed to be zero but the option to abandon the internet company can be

valuable. Berger, Ofek, and Swary (1996) investigated investors’ ability to price

the option to abandon a company at its exit value and found that after controlling

for other variables, firm value does increase in exit.

The estimation of the parameters of the model remains quite challenging. The

application of the model is illustrated using data from twenty internet companies.

The analysis undertaken here is quite thorough because based on a cross-

23
 sectional data from a sample of twenty internet companies to estimate the model

parameters. Some judgment calls are made for the parameters for which no data

were available. Furthermore, seasonality in the sales or revenues of the internet

company in its industry is taken into account when estimating the model

parameters. Consequently, the risk of overestimating the volatility of the growth

rate in sales is minimized. The seasonality found in the sample internet

companies analyzed was not significant to warrant the use of seasonally

adjusted sales in the estimation process.

References

Berger, P.G., E. Ofek, and I. Swary. 1996. “Investor Valuation of the

Abandonment Option.” Journal of Financial Economics, Vol.42, No.2
(October):257-287.

Brennan, M.J., and E.S. Schwartz. 1982. “Consistent Regulatory Policy under
Uncertainty.” Bell Journal of Economics, Vol.13, No.2 (Autumn):507-521.

Longstaff, F.A., and E.S. Schwartz. 1998. “Valuing American Options by

Simulation: A Simple Least-Squares Approach.” Working paper, University of
California at Los Angeles (August).

Moon, M., and E.S. Schwartz. 2000. “Rational Pricing of Internet Companies.”
Financial Analysts Journal, Vol.56, No.3 (May/June):62-75.

Schwartz, E.S., and J.E. Smith. 1997. “Short-Term Variations and Long-Term
Dynamics in Commodity Prices.” Working paper, University of California at Los
Angeles (June).

24

View publication stats