This article was downloaded by: [115.113.198.

246] On: 01 January 2024, At: 07:14

Publisher: Institute for Operations Research and the Management Sciences (INFORMS)
INFORMS is located in Maryland, USA

INFORMS Transactions on Education

Publication details, including instructions for authors and subscription information:
http://pubsonline.informs.org

Case—Forecasting Beer Demand at Anadolu Efes

Murat Köksalan, Selin Özpeynirci, Haldun Süral,

To cite this article:

Murat Köksalan, Selin Özpeynirci, Haldun Süral, (2010) Case—Forecasting Beer Demand at Anadolu Efes. INFORMS
Transactions on Education 10(3):142-145. https://doi.org/10.1287/ited.1100.0048cs

Full terms and conditions of use: https://pubsonline.informs.org/Publications/Librarians-Portal/PubsOnLine-Terms-and-

Conditions

This article may be used only for the purposes of research, teaching, and/or private study. Commercial use
or systematic downloading (by robots or other automatic processes) is prohibited without explicit Publisher
approval, unless otherwise noted. For more information, contact [email protected].

The Publisher does not warrant or guarantee the article’s accuracy, completeness, merchantability, fitness
for a particular purpose, or non-infringement. Descriptions of, or references to, products or publications, or
inclusion of an advertisement in this article, neither constitutes nor implies a guarantee, endorsement, or
support of claims made of that product, publication, or service.

Copyright © 2010, INFORMS

Please scroll down for article—it is on subsequent pages

With 12,500 members from nearly 90 countries, INFORMS is the largest international association of operations research (O.R.)
and analytics professionals and students. INFORMS provides unique networking and learning opportunities for individual
professionals, and organizations of all types and sizes, to better understand and use O.R. and analytics tools and methods to
transform strategic visions and achieve better outcomes.
For more information on INFORMS, its publications, membership, or meetings visit http://www.informs.org
 Vol. 10, No. 3, May 2010, pp. 142–145
issn 1532-0545
10
1003
0142 informs ®

doi 10.1287/ited.1100.0048cs
I N F O R M S © 2010 INFORMS

Transactions on Education

Case
Downloaded from informs.org by [115.113.198.246] on 01 January 2024, at 07:14 . For personal use only, all rights reserved.

Forecasting Beer Demand at Anadolu Efes

Murat Köksalan
Department of Industrial Engineering, Middle East Technical University, Ankara 06531, Turkey
[email protected]
Selin Özpeynirci
Department of Industrial Systems Engineering, İzmir University of Economics, İzmir 35330, Turkey,
[email protected]

Haldun Süral
Department of Industrial Engineering, Middle East Technical University, Ankara 06531, Turkey
[email protected]

Key words: forecasting; regression model; demand estimation

History: Received: January 2009; accepted: February 2010.

Forecasting Beer Demand at Information about the beer production process, his-
Anadolu Efes tory of beer, and alcohol consumption in Turkey are
Efes Beverage Group is the beverage division of given in Figures 1–3.
Anadolu Holding, one of Turkey’s leading conglom-
erates. Anadolu Holding was founded in 1969 and Forecasting Beer Demand
has principal interests in the beverage, automotive, As a part of their yearly planning efforts, Efes fore-
finance, restaurant chain management, and office sup- casts the monthly demand for the coming year during
plies sectors. Efes Beverage Group produces and mar- the fall of the current year. Historically, high-level
kets beer, malt, and soft drinks across a geography sales managers, based on input from their sales per-
that consists of Turkey, Russia, the CIS countries, sonnel, have done this forecasting mostly subjectively.
Southeast Europe, and the Middle East. Anadolu Efes Because they have been in the sector for a long time,
(or Efes in short) is the beer division under the Efes their estimates have been pretty accurate in the past.
Beverage Group (see http://www.anadoluefes.com). There have been cases where the estimates were off
As of 2006, Efes was one of the leaders of the beer but these were rare instances. They already made
industry in the region with its 16 breweries, 6 malt their forecasts in the Fall of 1993. In January 1994, they
plants, and 1 hops processing facility in six countries. are discussing how they could formalize this process.
Efes has a brewing capacity of 3.3 billion liters and a The Managing Director, Serdar Bölükbaşı, men-
malt production capacity of 236,500 tons annually. tioned his desire to come up with a more formal
Efes has a share of about 78% in the Turkish beer way of forecasting to Levent Tomaç, the Vice Presi-
market. Efes Pilsen, Efes Light, Efes Dark, Efes Extra, dent of Marketing. He explained that such a study
Efes Ice, Marmara Kirmizi, Marmara Gold, Ritmix, would be useful to test their intuitions on the fac-
and Gusta are the popular flavors supplied by Efes. tors they thought had a significant effect on beer
Additionally, it brews and supplies Miller Genuine demand. Levent agreed immediately. He explained
Draft, Beck’s, and Foster’s under licence agreements. that he had been thinking on this subject as well and
Efes also offers a wide variety of local brands with they could still use their judgments after the formal
different tastes and appeals in international markets. analysis if they thought some aspects had been over-
Stary Melnik, Sokol, Bely Medved, and Krasny Vostok looked. They remembered the young employee in the
are the local brands and Warsteiner, Amsterdam Nav- Logistics department, Selin Baydar, who recently did
igator, Zlatopramen, Sol, and Bavaria are the brands a great job studying the distribution of beer and the
Efes brews under licence agreements in Russia. location of new breweries.
142
 Köksalan, Özpeynirci, and Süral: Case—Forecasting Beer Demand at Anadolu Efes
INFORMS Transactions on Education 10(3), pp. 142–145, © 2010 INFORMS 143

Figure 1 The Beer Production Process Figure 4 The Month of Ramadan

Beer Production Ramadan

Beer is produced by the fermentation of carbohydrates in cereals, It is considered the most blessed month of the Islamic year and is
such as wheat, corn, rice, and most commonly, barley. The main linked to the lunar calendar. Many Muslims fast during the
ingredients of beer are water, malted barley, hops, and yeast. month of Ramadan. Those who fast refrain from eating,
Malted barley is the processed grain that has begun drinking, and smoking during the day. At the end of Ramadan,
germination by being soaked in water, and provides beer its a three-day feast starts when friends and relatives visit each
Downloaded from informs.org by [115.113.198.246] on 01 January 2024, at 07:14 . For personal use only, all rights reserved.

body and color. Hops is used in small amounts as a other, and candies, sweet pastries, and sweet liquors are served.
preservative agent. It also gives beer a bitter flavour and a
pleasant aroma. Yeast is composed of microorganisms that
convert sugar in malt juice to alcohol and carbon dioxide. From always enjoyed studying and is currently enrolled in a
the start of the production process, it takes approximately 21
days until the beer is ready for consumption.
Masters program in an Industrial Engineering depart-
ment. She appreciates learning the details of many
approaches and having the opportunity to experi-
Figure 2 A Brief History of Beer ment with large-scale models. She recently worked
on large-scale regression case studies in a statistics
History of Beer course. She thought that regression would be useful
Beer is one of the world’s oldest
alcoholic beverages. As early as 6,000
in trying to understand and forecast beer demand.
B.C., people were brewing beer in She had not been much of a beer drinker, but hav-
Mesopotamia, the land between the ing worked for Efes for four years, she had some idea
Tigris and Euphrates, rivers that about what affects beer demand. Tourism, the rela-
originate in southeast Anatolia. The tive prices of beverages that could be substituted for
Hittites, one of the most sophisticated
civilizations 4,000 years ago in beer (hard liquor and soft drinks), increasing trend
Anatolia, were drinking beer in daily life as well as in sacred because of increasing population and increase in beer
rituals. Ancient Egyptians used beer as a remedy for various drinking habits, and Ramadan (the month of fast-
maladies. In the dark ages when Europe was in the throes of ing in Islam; see Figure 4 for further information)
various epidemics, beer saved many lives as a safe and
were some of the factors she immediately thought of.
nourishing drink.
After some discussions with her colleagues, she fur-
By the seventh century A.D., beer was produced by European
monasteries. Today, the brewing industry is a huge business
ther developed the specifics of some of the factors
worldwide. The global beer consumption was approximately (see Tables 1–4).1 Colleagues have observed and some
155 billion litres in 2005. surveys supported over the years that many con-
sumers would substitute soft drinks for beer and vice
versa, especially when it came to quench their thirst.
Figure 3 Alcohol Consumption in Turkey
They thought that this substitution occurred between
Alcoholic Beverages canned packages of soft drinks and beer. Hard liquor
Turkey has a diverse culture. It has been a secular state since the (especially the traditional Turkish drink Raki) was
establishment of the Republic in 1923 and there is no restriction also considered to be a substitute for beer. The claim
on alcohol consumption. Even during the Ottoman Empire, the was that, in restaurants and pubs, people would sub-
restrictions depended on the attitude of the reigning sultan. stitute draft beer and Raki for each other. There were
Sultan Suleyman the Magnificent (1520–1566) was a strict
prohibitionist, for instance. On the other hand, Sultan Ahmed
indications that these consumers were price sensi-
III (1718–1730) permitted alcohol consumption. tive in their substitutions. Another issue regarding
Surveys indicate that around 48% of the adult population in the price was the timing of price changes. Because of
Turkey consume alcoholic beverages, and around 42% of those relatively high inflation, Efes (as well as other com-
who do not drink state that religious beliefs are the main reason panies) occasionally adjusted their prices. The real
for abstaining. Many individuals stop drinking alcoholic prices, hence, would go down for several months, and
beverages during the month of Ramadan, an Islamic fasting
month.
then go back up after the price adjustment.
Selin was now ready for the meeting.
Raki is a traditional hard liquor produced and consumed in large
amounts. Wine and beer are other popular alcoholic beverages. Serdar Bölükbaşı: I understand you made some plans
about formalizing our beer forecasting.
Selin was excited about her new assignment. She Selin Baydar: Multiple regression is one of the
had studied food engineering before taking a job widely used statistical tools for forecasting. I have
at Efes in the Production department. After work- some experience with this and I think it would be
ing for several years, she completed a part-time useful in explaining beer demand to a large extent.
MBA program and started working in the Logistics
department. Oddly, to many of her friends’ tastes, she 1
Tables 3 and 4 are in the supplementary Excel file Tables 3–4.xls.
 Köksalan, Özpeynirci, and Süral: Case—Forecasting Beer Demand at Anadolu Efes
144 INFORMS Transactions on Education 10(3), pp. 142–145, © 2010 INFORMS

Table 1 First and Last Days of Ramadan in Years 1987–1994∗ Levent Tomaç: We previously got some detailed
information from the Ministry of Tourism for some
Year First day of Ramadan Last day of Ramadan
other purpose. They collect data on the number of
1987 April 29th May 28th tourists coming from each country and produce infor-
1988 April 18th May 16th mation on number of tourist nights spent in Turkey
1989 April 7th May 5th
1990 March 28th April 25th
by citizens of different countries during each month.
1991 March 17th April 15th We have these results for five major countries and the
Downloaded from informs.org by [115.113.198.246] on 01 January 2024, at 07:14 . For personal use only, all rights reserved.

1992 March 6th April 3rd rest are summed up under the “others” category.
1993 February 23rd March 23rd Serdar: That is great! Average beer consumption
1994 February 12th March 12th statistics of each country is known and we can come
∗
http://www.diyanet.gov.tr. up with a weighted factor for these five countries. We
can then take the tourist nights of “others” and treat
Selin explained what she had in mind and men- it as the sixth country weighting with the average of
tioned the factors she thought of using. They agreed the average beer consumptions of others. You know
to use the total monthly demand figures of the whole which country’s tourists I would like to see in Turkey
country for the period 1987–1993. Efes had a huge most. Several Czechs could consume a lot of beer!
market share (approximately 78%). Other compa- Levent: Using weighted number of tourists is a good
nies had similar policies in terms of price adjust- idea. We will come up with a variable that stands for
ments, marketing of different beer types, etc. They the approximate amount of beer we expect tourists
decided not to differentiate between different types to consume in a given month. However, I am a little
and packages and defined an aggregated litre of beer sceptical about the claim on the substitution of soft
to define the consumption amounts. They agreed to drinks for beer. I would be surprised if this amounts
use appropriate weights to calculate average prices. to anything sizeable but let us try and see. For the
Selin explained her difficulties in quantifying effects substitute beverage effects, we could directly use their
of tourism, population, and substitutable beverages. prices relative to our prices.
Selin: I think those would work well. Do you
have any suggestions on how to incorporate the
Table 2 Average per Capita Beer Consumption of Some
Countries population?
Serdar: I think increase in population and increase
Beer consumption in beer drinking habits might have similar effects. In
Country (liters/capita)∗
Turkey, there has been a steady growth in both. I don’t
Czechoslovakia 163 see why that would change in the near future.
Germany 138 Selin: Great, I can use a trend factor that may cap-
Denmark 126 ture those effects. I think I am ready to try these ideas.
Austria 118
Ireland 113
Serdar: Let us know as soon as you get some mean-
Belgium 108 ingful results.
United Kingdom 100
After a week of hard work, Selin obtained her
Luxembourg 104
Australia 96 first results. Though the factors she used seemed
United States 86 to explain a significant part of the variation in the
The Netherlands 85 monthly beer demand, the residual graphs did not
Finland 86 look perfect. She wanted to see if some transforma-
Venezuela 77
tions would work better to justify the assumptions
Canada 71
Sweden 64 of linear regression. After some transformations, she
Spain 67 had alternative models that seemed to approximately
Portugal 64 satisfy the assumptions. She thought it was time to
Japan 56 present her results to Serdar Bölükbaşıand Levent
South Africa 55
Norway 50
Tomaç. Selin briefly explained the results and talked
Mexico 47 about how she checked the assumptions of regression.
Columbia 41 To her surprise, the bosses were interested even in
France 39 these details.
Korea 34
China Hong Kong 28 Serdar: These are enlightening. We have always
Italy 25 thought that people drink more beer when it is hot.
China Taiwan 23 Our sales figures do increase in summer months. Did
China 10
your model verify this observation?
∗
Approximate values as of 1993. Selin: Oops! How could I totally ignore this fact?
 Köksalan, Özpeynirci, and Süral: Case—Forecasting Beer Demand at Anadolu Efes
INFORMS Transactions on Education 10(3), pp. 142–145, © 2010 INFORMS 145

Levent: Don’t worry. This is why we have these Selin: I hope so too. This was a quick meeting. I bet-
meetings. Sometimes one gets too carried away with ter start writing immediately.
details.
Selin prepared her report explaining the dependent
Selin: I will figure out a way to incorporate this sea-
and independent variables in detail. She explained
sonality effect.
how she conducted the experiments to exclude the
Serdar: I like this “seasonality effect” term. Shall we
variables that turned out to be statistically insignif-
postpone our meeting for tomorrow then?
icant. She discussed the meaningful variables and
Downloaded from informs.org by [115.113.198.246] on 01 January 2024, at 07:14 . For personal use only, all rights reserved.

Selin: Sure, I should be able to resolve this quickly.

Serdar: I would like to see a detailed explanation the magnitudes of their effects. She discussed and
of how you checked those assumptions of regression demonstrated how the assumptions of linear regres-
you mentioned. I learned about these years ago and sion were approximately satisfied. She explained how
this will be a good refresher for me. the results could be used and she demonstrated some
predictions and prediction intervals.
Selin was still embarrassed about overlooking the After finishing her report, Selin was tired but satis-
seasonality. She decided to do a detailed analysis to fied that the project turned out to be meaningful. She
see the relative effect of each season. She thought she could not wait to see the model put to use.2
had sufficient data points and she could even test the
seasonality on a monthly basis if she wanted to do
so. She quickly incorporated the necessary variables Supplementary Files
and solved the model. The results cheered her up. An electronic companion to this paper is available at
The model seemed to explain a high percentage of the http://ite.pubs.informs.org/.
variation in beer demand. She was now ready for the
meeting. Acknowledgments
Serdar: Things are looking great. I look forward to This case won first prize in the 2007 Case Competition of
getting your final report soon. Then we can have our the Institute for Operations Research and the Management
different departments discuss it. Hopefully, we will Sciences (INFORMS).
be able to incorporate this in our forecasting activities
starting from this year. Perhaps we will eventually be
able to use these as a decision tool as well by testing Reference
how beer demand might be affected when we change Efes Beverage Group, Anadolu Efes. Accessed on January 22, 2010,
the prices a little bit. http://www.anadoluefes.com.

2
The company and beer market information are accurate as col-
lected and cited from various references. Serdar Bölükbaşıand Lev-
ent Tomaç are the real managers at the positions mentioned in the
case. We would like to thank them for their comments about the
case and for kindly allowing us to use their names. The story devel-
oped for the case is realistic but fictional.