Eurotel 3G License Valuation

developed by Mark Shay1, Robert Freund, and Georgia Perakis

©2021 Dynamic Ideas, LLC

Introduction
From her office in the rugged hills of Sophia Antipolis, France, Ana Bellario gazed out
over the mist-covered Mediterranean shore. The Director of Market Strategy for Eurotel,
Ana had spent the past few days revising NPV calculations for Eurotel’s planned
geographic expansion in the next fiscal year.2 She was not satisfied with the results
before her, and began wondering how she might direct her team of analysts to approach
the problem differently…

It is the autumn of 2002, and Eurotel has over the last three years built an empire out of
its successes in nearly a dozen 3G (“third-generation”) telecommunications license
award competitions. Eurotel’s 3G operating rights now stretched, as Winston Churchill
might have put it, “From Stettin in the Baltic to Trieste in the Adriatic…” 3

The Business Problem

After winning its first few 3G licenses in large, lucrative markets such as the U.K. and
Germany, Eurotel experienced the euphoria that swept the global telecommunications
industry at the time. Since then, Eurotel – and its competitors – have taken a more sober
approach to expansion, focusing on paying reasonable prices for 3G licenses without
overpaying. Eurotel in particular has been increasing its efforts in license valuation
modeling, for example by hiring Ana, a recent MIT Sloan MBA graduate.

Holding the rights to offer the next generation of high-speed voice and data
communications to mobile customers across the Continent has put Eurotel in an
enviable first position in terms of market access, but at a steep cost. Eurotel has had to
raise nearly $US 14 billion from debt and equity markets to cover the costs of licenses
1
Based on a consulting engagement in Denmark, 2001, in which Mark Shay (Sloan MBA 2004) was involved.
2
NPV = Net Present Value, the sum of all expected future cash flows from the project, minus the required investment,
discounted back to the present time at the opportunity cost of capital, or “discount rate.”
3
Churchill’s famous ”Iron Curtain” speech was given on March 5, 1946 at Westminster College in Fulton, Missouri, where
he said, “From Stettin in the Baltic to Trieste in the Adriatic an iron curtain has descended across the Continent. Behind
that line lie all the capitals of the ancient states of Central and Eastern Europe…”

1
and fixed asset investments – and the first 3G mobile phones and base stations have yet
to be delivered.

Future investment requirements are uncertain yet unavoidable, and future revenues are
the subject of industry-wide speculation. No one knows for sure whether 3G services will
command a premium over “traditional” mobile telephony services, and naysayers predict
that 3G will fail against substitutes such as “WiFi” access points, which give mobile
computer users low-cost wireless Internet access from many public places.

Now, as Eurotel positions itself for the upcoming 3G license auctions in Russia,
Hungary, and Turkey, it must choose its bids wisely: bid too low and Eurotel will not win
these strategic puzzle pieces; bid too high and it will be punished by the financial
markets and industry press, and will fail to recoup its investments – a potential recipe for
bankruptcy.

Eurotel’s Approach to License Valuation

In only six working days, Ana Bellario must issue a report to Eurotel’s CEO that specifies
at what level Eurotel should value the licenses in the 3G auctions in Russia, Hungary,
and Turkey. Eurotel has until now relied on a three-step process to generate license
valuations:

• Step 1 - NPV Analysis: In an iterative capital budgeting exercise, Eurotel first

estimates the required investment and total expected revenues associated with
the 3G expansion into a given country. The future cash flows are discounted
back to the present time at a discount rate of 16%, this being the standard rate
that Eurotel has used for all of its 3G projects. The budgeting is done initially
without the cost of the license -- to see whether the NPV is positive -- as a first
test. If the NPV is positive, then the analysis continues and a hypothetical license
price is added to the investment portion (i.e., the investment becomes larger by
the amount of the license price). The license price is incremented until the NPV
reduces to zero, yielding a break-even point. The license price for which the
project NPV is zero is used as Eurotel’s maximum price, since any higher price
implies a sub-par investment.

• Step 2 – Market Indicator Consideration: Eurotel’s management team checks

the value from Step 1 against a “top-down” indication of what the license value
could be. This estimate is a simple arithmetic mean of all previous 3G license
awards, expressed in terms of license price per capita (i.e., price paid per
license, divided by the number of people in the country). Eurotel management
uses this figure as a “ballpark” indicator, primarily to know whether the Step 1
value is on the right order of magnitude. Any substantial difference is reconciled
through additional research to explain the discrepancy, or to revise the estimates.

• Step 3 – Game Theory: In countries where 3G licenses are awarded by auction,

Eurotel uses game theory analysis and simulation to determine its auction bid (or
series of escalating bids). With the value from Step 1 as a maximum, the game
theory analysis attempts to generate a minimum bid value that is sufficient to win.
Since the auction formats vary significantly by country, this step can be quite
complex. The analysis must consider the auction format (e.g., English auction vs.
single bid auction) as well as the competitive landscape (e.g., who are the

2
 competing bidders, and what are their strategies and financial capabilities?). The
bidding strategy is submitted to Eurotel’s CEO and Executive Committee for final
approval.

In countries where auctions are not used, national telecommunications regulators run
“beauty contests” to determine the license winners. These contests require competitors
to submit detailed business plans and rollout strategies. The license value is usually
given as a fixed price by the regulator. Eurotel determines whether to bid in these
countries based on comparing its estimated license value (from Steps 1 and 2 above) to
the price given by each regulator. In some cases (e.g., Japan, Sweden, Finland) the
licenses have been offered free of charge. The beauty contest winners are chosen by
the national regulators based on financial strength, the likelihood of success, and
contribution to the overall good of the country.

In auctions and beauty contests alike, government regulators use the revenues from 3G
license sales to fund a wide variety of national budget expenditures. Because
governments own and can sell a natural resource (i.e., air) that costs them nothing,
these methods for raising state revenue have grown in popularity in recent decades. A
lively social and industry debate has grown concurrently, with many industry groups
arguing that fees for licenses pass costs on to consumers and threaten operators’
competitive and financial strength.

A New Approach: Multiple Linear Regression

Ana had reviewed the results of Eurotel’s recent auctions and beauty contests, and had
noticed a few vague patterns. It seemed like there were some connections between
where licenses were awarded, when the contests were held, and what the winning 3G
operators paid for the licenses.

Ana wondered if, as an alternative approach to the NPV method, she and her team
could collect reliable data about past 3G license contests and run a multiple regression
model as an estimation tool for future contests. If the regression model were sufficiently
robust, then data estimates for Russia, Hungary, and Turkey could be used to generate
approximate license valuations in these markets. Ana’s team could then use the
predicted valuations as inputs to the existing valuation/bidding process, giving Eurotel an
improved license valuation and – quite possibly – a much-needed competitive edge.

As the sun disappeared into the Provençal foothills, Ana turned to her PC and started
hammering out a work plan. She would need data – lots of data –, and she would have
to proceed quickly and thoughtfully to get the job done in such a short time-frame. Ana
was not even certain multiple regression would work, but she knew that the potential
payoff of more accurate license valuations was too great not to pursue this approach.

Data Gathering
Ana began planning the data search with an idea of the factors that were most likely to
drive license values up or down. She sketched the kinds of data she would need from
each license award contest and country:

• Industry – Sophistication of the countries’ mobile telephony industries; level of

competition among operators.

3
 • Spending power – Ability of consumers to spend money on
telecommunications.

• Size – Number of potential customers.

Knowing that the worldwide telecommunications market had fallen recently with the rest
of the global stock market, Ana also considered including an indicator of the value of the
industry as a whole, for example a stock market index.

Finally, Ana would need the outcome of each license award contest, including those in
which Eurotel chose not to participate.

With these first ideas in mind, Ana made a list of the data categories she thought would
be of interest:

Data Category Description

Country Mobile subscribers/Population
Contest type Auction or beauty contest
Date Month and year of the award
Population Country population, in millions
Mobile subscribers Active mobile phone
subscriptions in the country, in
millions
Licenses available Number of licenses being
awarded
Incumbent operators Number of existing mobile
phone operators in the country
License term Number of years the license
will be valid from the award
date
FTSE Global Telecom Index A global telecom stock index
value on the date of the award
GDP Gross domestic product, in
$US billion
ARPU Average revenue per user, the
average monthly phone bill for
existing mobile phone
subscribers
Price for all licenses Total revenue earned by the
government in the contest, in
US$ billion

Several of these categories, Ana noticed, might inflate the value of large countries at the
expense of smaller countries. For example, Spain probably has a much higher GDP than
Switzerland, but Switzerland is more “wealthy” in terms of individual spending power. A
license in Spain might be worth more since there are more people and greater total
wealth (GDP), but a license in Switzerland might be worth more since the target
customers have more money to spend.

4
Since it is hard to know for sure how these factors had affected license valuations, Ana
figured that normalizing several of the categories by country population would be a good
thing to do. She reasoned that she could decide in the process of the analysis which
normalized categories made sense to use and which ones were better left in terms of the
original numbers.

In addition to the original data categories listed above, Ana decided to include the
following normalized values:

Data Category Description

Mobile penetration Mobile subscribers/Population
GDP per capita GDP/Population
Total license price per capita Price for all licenses/Population

Ana would have preferred to conduct the research herself (to get a feel for how the data
fit together), but this would be cost-prohibitive. Instead, she would supervise her team of
three analysts and a part-time research assistant. While she could issue a purchase
order for a stack of comprehensive industry research reports she had seen advertised
for several thousand dollars each, this would be expensive and potentially time-
consuming. Ana would have her team collect the data using free, public sources.

Fortunately, as Ana began “surfing” a handful of familiar industry web sites, she noticed
that most of the information she needed was indeed publicly available. If, at the end of a
thorough search, there were gaps in the data, Ana figured she could interpolate and
provide a few educated guesses that would be adequate and keep the analysis moving.

After an hour or so of preliminary searching to get the team started, Ana had identified a
set of sources that provided most of the data she needed:

Type of Data Source

Country data (GDP, population) U.S. State Department
www.state.gov/countries/
Telecommunications International
demographics (mobile Telecommunications Union
subscribers) (ITU)
www.itu.int/ITU-D/ict/statistics/
Global telecom stock index FTSE index service
www.ftse.com
3G license contest descriptions Industry on-line web sites, e.g.,
and data www.3gnewsroom.com
ARPU Eurotel proprietary research

With a few additional hours of research, Ana was able to fill in most of the table in Exhibit
A using data from these sources.

In a few cases, however, data was ambiguous or not available at all. For example,
Eurotel’s own research did not include ARPU for all countries. In nearly a dozen such
cases, Ana needed to estimate what the ARPU was. To do so, she focused on known

5
ARPU data and looked for obvious relationships between ARPU and other categories.
There was a clear correlation between GDP per capita and ARPU, with evidence that
countries with high mobile penetrations tended to have higher ARPU values. Concerned
that a regression analysis to generate ARPU estimates would be too time-consuming,
Ana used the existing data to interpolate the missing values. She figured that her
estimates were likely to be correct to within 10% or 20%, which was probably sufficient
as a starting point.

Ana also found, for example, that GDP data varied across similar sources. The U.S.
State Department and the International Telecommunications Union each reported
slightly different GDP values for the same countries. Some variations were due to the
year for which GDP was reported (e.g., 2000 vs. 2001). Other variations were simply
due to different numbers being reported for the same year, perhaps due to rounding
differences or different methods for computing GDP. To simplify her data search and
adhere to a consistent GDP reporting method, Ana decided that she would use the most
recent data available from the U.S. State Department for each country.

Your Assignment
Using the data Ana and her team have collected (Exhibit A), create a multiple regression
model to estimate the license valuations for Russia, Hungary, and Turkey.4 If you require
data that are not given, make assumptions and state them clearly. Do not conduct
additional research to find data.

In a memo to Ana Bellario, describe your approach, your model, and your results, as
well as the implications for Eurotel’s endeavors in the three planned expansion markets.
To determine a recommendation for which new markets Eurotel might invest in, compare
your results with the hurdle valuations generated by the NPV analysis, i.e., the maximum
values Eurotel should bid in each market (see table below).

Country Hurdle Valuation Hurdle Valuation

($US Million) (Total Price per Capita)
Hungary $73.1 $7.24
Russia $4,060.0 $28.19
Turkey $4,891.0 $74.67

In the model, try to limit the number of independent variables to a reasonable number,
and use your intuition – not brute force – as your guide for determining which possible
independent variables to include. If you are not able to develop a robust regression
model (which may be the case), describe your approach to the problem and why you
think the regression exercise was not successful.

Describe in your memo reasons for and against using regression modeling as an
approach to the overall decision-making process at Eurotel. In your arguments you may
wish to comment on data quality and availability, time/cost constraints, or other issues in
the case that are relevant to management. Include relevant numerical results and
illustrations as an appendix to the memo.

4
While the data presented in Exhibit A were obtained principally from the given sources, numerical values have been
changed in several instances for teaching purposes. Stock market index values for the 2003 awards are based on Wall
Street projections given at the time of the analysis.

6
Further Reading
Klemperer, P. “How (Not) to Run Auctions: The European 3G Telecom Auctions,”
forthcoming in European Economic Review; this and other resources available at
www.paulklemperer.org.

Moldovanu, B. and Jehiel, P. “An Economic Perspective on Auctions,” working paper

June 2002, available at http://www.vwl.uni-mannheim.de/moldovan/epauc.pdf.

Articles and links relevant to radio spectrum auctions in the U.S. can be found at
www.fcc.gov.