11035624-Dissertation-MsC Information Technology (Final)

FORECAST PRACTICE IN
MANUFACTURING FIRM AND THE
ROLE OF INFORMATION TECHNOLOGY
Dissertation UBLLY7-60-M
Vy Quoc Tran
Student ID: 11035624
MsC Information Technology
University of the West of England
Date: 26 November 2015
Word counts: 16,643
Vy Quoc Tran
Student ID: 11035624
MsC Information Technology
Supervisor:
Dr. Hisham Ihshaish

Forecast Practice in Manufacturing Firm and the Role of Information Technology
1 | P a g e
Table of Contents
Table of Figures.......................................................................................................................................3
Acknowledgement..................................................................................................................................4
Abstract...................................................................................................................................................5
Chapter 1: Introduction & Research Methodology ................................................................................6
1.1 Introduction ..................................................................................................................................6
1.2 Research scope and context .........................................................................................................6
1.3 The problem..................................................................................................................................7
1.4 Research Aim ................................................................................................................................8
1.5 Research Objectives......................................................................................................................8
1.6 Dissertation Structure...................................................................................................................8
1.7 Research Methodology.................................................................................................................8
1.8 Ethical............................................................................................................................................9
Chapter 2: Forecasting – A Literature review.......................................................................................11
2.1 Introduction ................................................................................................................................11
2.2 The Role and Nature of Forecasting in business.........................................................................11
2.3 Forecasting types........................................................................................................................12
2.4 Forecasting process.....................................................................................................................14
2.5 Forecasting model.......................................................................................................................17
2.6 Forecasting method....................................................................................................................18
2.7 Forecasting accuracy & error: Statistical Vs Judgmental method. .............................................20
2.7.1 Accuracy & Error ..................................................................................................................20
2.7.2 Statistical method’s error ....................................................................................................22
2.7.3 Judgmental method’s error .................................................................................................23
2.8 Improve forecast accuracy: Integrate statistical and judgmental method.................................25
2.9 Conclusion...................................................................................................................................27
2.9.1 Summarize Chapter 2...........................................................................................................27
2.9.2 The remaining questions......................................................................................................28
Chapter 3: How Information Technology supports forecast practice in manufacturing firm ..............29
3.1 Introduction ................................................................................................................................29
3.2 How firm organizes forecast function.........................................................................................29
3.2.1 Role and position .................................................................................................................29
3.2.2 Staff......................................................................................................................................31
3.2.3 Forecast practice..................................................................................................................32
3.3 Forecasting in manufacture firm: Focus in Demand forecast to support supply chain..............36

2 | P a g e
3.3.1 Why supply chain and how it links to demand forecast? ....................................................36
3.3.2 The role of demand forecast in supply chain.......................................................................37
3.4 How IT support demand forecast. ..............................................................................................39
3.4.1 Spreadsheet tools and Forecasting software package ........................................................39
3.4.2 Information system with a forecast function ......................................................................41
3.4.3 Data mining technology and Big data..................................................................................41
3.5 The Cost of IT investment in forecast .........................................................................................42
3.6 Conclusion...................................................................................................................................44
3.6.1 Summarize Chapter 3...........................................................................................................44
3.6.2 The remaining question .......................................................................................................44
Chapter 4: Research Finding and Analysis ............................................................................................46
4.1 Introduction ................................................................................................................................46
4.2 Scavi Viet Nam - Overview..........................................................................................................46
4.3 Information Technology level of Scavi........................................................................................47
4.3.1 Hardware and infrastructure:..............................................................................................48
4.3.2 Network Components:.........................................................................................................48
4.3.3 Basic Software Architecture:................................................................................................49
4.3.4 Information system..............................................................................................................49
4.4 Outline the forecasting practice .................................................................................................50
4.4.1 Forecasting function in Scavi ...............................................................................................50
4.4.2 Forecasting process links directly to supply chain through ScaX & Scala............................52
4.4.3 Forecasting methods of Scavi: .............................................................................................53
4.5 Issue finding and Discussion .......................................................................................................54
4.5.1 Statistical forecast methods are not useful in manufacture firm........................................54
4.5.2 The support of Information technology to the demand forecast of manufacture firm is
limited...........................................................................................................................................55
4.5.3 The lacking of statistical practice and forecasting technology limits the growing ability of
the firm. ........................................................................................................................................55
Chapter 5: Conclusion...........................................................................................................................57
5.1 Conclusion...................................................................................................................................57
5.2 Research limitations....................................................................................................................58
Reference..............................................................................................................................................59
Appendix A............................................................................................................................................64
Appendix B............................................................................................................................................67
Appendix C............................................................................................................................................69

3 | P a g e
Table of Figures
Figure 1. The five step of forecasting process ......................................................................................15
Figure 2. Where forecasting function resides.......................................................................................30
Figure 3. Highest Academic Degree Acquired by Forecasters ..............................................................31
Figure 4. Forecasters. Major Field of Study in University.....................................................................32
Figure 5. Business Background of Forecasters......................................................................................32
Figure 6. Modules Used in Forecasting.................................................................................................33
Figure 7. Times Series Models Used .....................................................................................................34
Figure 8. Forecasting Horizon ...............................................................................................................34
Figure 9. Forecast Buckets ....................................................................................................................35
Figure 10. Cause and Effect Models......................................................................................................35
Figure 11. Judgemental Models Used...................................................................................................36
Figure 12. The important of Sales Forecasting .....................................................................................38
Figure 13. Business Areas that use Sales Forecast Information ...........................................................38
Figure 14. Market Share of Different Forecasting Software Packages.................................................40
Figure 15. Market Shared of Forecasting Packages Vs Spreadsheet Packages ....................................40
Figure 16. Market Shared of Different Forecasting Systems................................................................41
Figure 17. Organization Structure of Scavi Viet Nam ...........................................................................47
Figure 18. The Information System of Scavi .........................................................................................50
Figure 19. Commercial Department of Scavi ........................................................................................51
Figure 20. Flow Chart of Demand Forecast and Taking Order process ................................................52

4 | P a g e
Acknowledgement
My deepest gratitude goes to my big family in Viet Nam, especially my parents, for always
motivation and supporting me to overcome difficulties during the dissertation.
I would like to thank Scavi Viet Nam, The Board of Director, Ms. Nguyen Thi Xuan Dai, Ms.
Nguyen Thi Hong Chau, Mr. Tran Quoc Nam and all members in the Commercial Department
and IT Department, for fully supporting me throughout the two month of conducting
research.
I specially thanks to Ernst and Young Viet Nam, the department of IT Audit Risk & Assessment,
for letting me spending two month of internship in the firm. The working experience at EY
had provided many useful information for my research.
And lastly, my dissertation would never have been completed without my supervisor, Dr.
Hisham Ihshaish. I thank him for the guidance and encouragement to me while helping me to
identify this topic. I know that I could not have done my dissertation without his help.
.

5 | P a g e
Abstract
Forecasting in business has been developed strongly in the past decade in both fields of business and
computer. Following the development of more and more statistical method there have a shift to using
more computer and technology into the forecasting practice of many organizations.
In business, manufacturing firm is described as the muscle of the economy and demand forecast has
been mentioned and studied as one of the biggest problems of the modern manufacture industry. In
the recent years, information technology has been proving to be a powerful tool to support the
demand forecast. However, in developing countries, where the overall condition is different, and the
IT level is lower, how the company implements IT to support demand forecast practice?
In this research, we found out that in developing country, manufacturer rarely use statistical method
in demand forecast practice. And as a consequence, the role of Information Technology in support
demand forecast is not significant.

6 | P a g e
Chapter 1: Introduction & Research Methodology
1.1 Introduction
Throughout many researches (Tony Hines, 2013; Charles,2013; Fildes & Goodwin, 2008; MyerHoltz &
Caffrey, 2014), demand forecast has been mentioned and studied as one of the biggest problems of
the modern manufacture industry. This issue attracts the attention of, not only in the term of business
management but also of information technology application.
In term of business, a manufacturing firm that have an accurate demand forecast system, will provide
a better supply chain performance (Myerholtz & Caffrey, 2014). In the manufacturing industry, the
role of the supply chain is a crucial one (Tony Hines, 2013). In other words, the purpose of the business
and the supply chain are to provide good/product/service to meet the demand. If there are no
demand, there will be no supply, and then, there will be no business. Therefore, forecasting demand
is an important information that allows a firm to maintenance, or push, their business (Tony Hines,
2013).
In term of information technology, demand forecasting is one of the biggest difficulty while trying to
apply IT in business (Dreischmeier et al., 2014). The fast developing of technology has created many
new approach theories and practical tools that allow firms to develop more accurate demand
prediction systems (Nenni et al., 2013). However, each manufacturing company must deal with a
unique market area and situation. Therefore, how information technology can fit into each
organization and provide the best support to the forecast demand tasks, is always a difficult question
(Charler W. Chase, 2013).
Filling this gap between business and IT is always a challenge for both business and IT leader. In this
research, we believe that looking into this issue and providing a better understanding of demand
forecast, will contribute to a more efficient way of applying information technology to the
manufacturing industry.
1.2 Research scope and context
Forecasting in business has been developed strongly in the past decade in both fields of business and
computer. There are more and more techniques, and technologies have been designed to support the
practice of forecast method (Fildes & Goodwin, 2008). However, in this research, we will not focus on
the evaluation of forecasting technique, nor forecasting technology. Rather than that, we will
concentrate on the assessment of the implementation of forecast in business, as well as analyze the
using of informatics to support the forecasting process.

7 | P a g e
In business, manufacturing firm is described as the muscle of the economy (Friedman David, 2006).
Manufacture companies are the factor that directly produce wealth to the economy. That is the reason
the role of the manufacturer is crucial for the whole economy of a country. Nowadays, the company
in developed countries tend to outsource its production function to developing countries, where the
level of information technology application is lower. Clearly, there will be different in the forecasting
practice between the manufacturers in these countries. In this research, to reduce the scope of the
study, we will try to evaluate the forecasting practice in a manufacturing firm in developing countries,
in term of forecast practice and forecast technology.
1.3 The problem
Various forecast tools and methods have been developed to help forecaster (Nenni et al., 2013). Many
of them is developed using information technology and statistic method (Fildes & Goodwin, 2008).
However, as some research has been published, after the forecast has been provided by these IT and
statistical methods always got a re-adjustments (Fildes & Goodwin, 2008). In fact, for many
manufacture firm, demand forecasts are always conducted by both computer and human.
In many papers about forecasting (Nenni et al., 2013; Fumi et al., 2013; Charles, 2013;…), researcher
just focus on the technical aspect of forecasting. Researchers try to optimize the forecast process by
using the development of technology and by reducing the involvement of human factors. It is a good
effort to improve the forecast accuracy. However, in another way, it increases the gap between
technology and a real case business.
Many business consultant and data provider firms published the success in the application of new
modern forecasting technology such as big data (Charles W. Chase, 2013; Kotlik et al., 2015;…). But
these papers are based on big corporates in developed countries. These firms already have the power,
in both financial and technology resources, to obtain and apply the newest forecast technology (Jonh
E. Hanke & Dean W. Wichern, 2005). Hence, the question is: how about others medium-to-small
manufacturers in others less-developed countries?
I think that It is not realistic if we just focus on analyzing the successful of big corporate with the
enormous resource, data, and information assets. In a complex market nowadays, the application’s
situation will be entirely different in a small-medium firm with limited resources and IT condition (Luna
et al., 2014). Plus, the gap in technology between developed and developing countries is also a
significant barrier that restrict the application of new forecast technology (Barker et al., 1987; Issa et
al., 2009). Therefore, an in-depth study of forecast technology, in an emerging market of manufacture
industry, is needed.

8 | P a g e
1.4 Research Aim
The aim of this research is to reduce the gap between Business and IT in actual practical. The
application of IT in a real case business always exists many blind-spots that are hardly detected. And
Demand Forecasting is a crucial and promising field for both Business and IT practical. Thus, this
research will contribute to the development of demand forecast practical in the future.
1.5 Research Objectives
 Find out how company organize its forecast function? What is the most used forecast
practice?
 Find out what is the current level of forecast technology? How can information technology
support the forecasting process in an organization?
 Find out how the manufacturer firm in developing countries implement forecasting practice?
 Find out how the information technology has been applied to support the forecasting
practice?
1.6 Dissertation Structure
This paper will be divided into five chapters:
 Chapter 1: “Introduction and Research Methodology”.
 Chapter 2: “Forecasting – A Literature review”. This chapter will provide an overview of
forecasting in business. It will help the reader understand more about some basic knowledge
of forecasting, as well as its development in research in the past years.
 Chapter 3: “How Information Technology supports forecast practice in manufacturing firm”.
This chapter will help us answer the first two questions that have been mentioned in the
Research Objectives.
 Chapter 5: “Research Finding and Analysis”. This chapter will answer the last two questions in
the Research Objectives.
 Chapter 6: “Conclusion”. Conclusion and Limitations of the research.
1.7 Research Methodology
In Chapter 2 is a literature review part. Therefore, we will gather all academic resources, from the past
until today, and summarize all the essential knowledge of forecasting. It will provide the base
knowledge to the reader before we move to the next part of the research.
In chapter 3, to answer the first two question of the research objectives, the research method will use
a qualitative approach. This chapter will gather all available document, secondary data, business
report and survey, to outline all the information that can help us achieve the research objectives.

9 | P a g e
In chapter 4, to answer the last two questions, we will conduct a qualitative research method in a real
business case. The primary research tools will be used: Interview and Observation.
The case study will focus on one manufacturer firm only: SCAVI Joint Stock Co. SCAVI, established in
1988, is a long-life textile and garment manufacturer in Viet Nam, a developing country. This firm
belongs to the fashion industry, which is, by Fisher’s study (1997), having a high implied demand
uncertainty. Therefore, an in-depth research of demand forecast process and application in this firm
will be a good example to achieve this research’s objectives. Furthermore, the result of this research
may become a useful real-case benchmark for further research and study.
This study will use a non-participant observation method for two months. Which means that the
observer will not involve in the forecast process of the firm. The observation will focus on the forecast
demand meetings of the top managers and directors. The process and result of the meeting will be
noted down, as well as record (audio & video) if possible. Plus, a study of all the documents that
provided by the firm will be combined so that the observer will have a full understanding of the
forecasting process as well as the role of IT in this process. The data and information, both formal &
informal, will be collected and analyzed at the end of the observation period. The analysis will focus
on answer the main objectives of the research that were mentioned above. Finally, the research will
produce an observation report that will reflect the whole demand forecast process of Scavi, as well as
analyze the level of information technology that involves in the process.
The interview method will focus on all the directors and managers who get involve and who have the
decision-making power in the demand forecast process of the firm. A general interview guide
approach and a semi-formal interview method will be conducted. This approach is intended to ensure
that the same general areas of demand forecast are collected from each interviewee, but still allows
a degree of freedom and adaptability in getting the information. Also, for each interviewee, especially
the one who belong to the IT department, a topic about IT and IT application in the demand forecast
process will be asked. All the data and information will be noted, or recorded if possible, and analyzed.
A comparison between interviewee’s ideas on the demand forecast process of the firm will be
conducted. The common issues will be highlighted, and the different opinions will be critically
analyzed for its value to reduce a biased result.
1.8 Ethical
For the observation, the access’s right will be provided directly by the Board of Directors and the
research’s purpose and the result will be fully reported to the Board at the end of the research period.

10 | P a g e
For the interview, all participants will be notified about the purpose of the study and will be asked for
the permission before conducting the interview. The data, information, and record after the interview
will be showed to the interviewee who will have the right to withdraw, fix or add to answer.

11 | P a g e
Chapter 2: Forecasting – A Literature review
2.1 Introduction
Since the ancient times of human history, people had always tried to predict the future to support
decision making and to make a plan for future action (Granger, 1980). Most of the forecasting
practices at that times are spiritual ways. The forecasters are usually calling for many different ways:
prophecy, sorceress, or fortune teller… The way of the forecast was the using of a series of past
coincident data, information or event, to set up relevant rules. And then, based on this regulation, the
current situation, and personal judgment, the forecasters will give out the future’s prediction.
However, as times pass, the development of natural science had changed the way people forecast the
future in which a more logical, numerical and statistical method of manipulating data has been applied
to increase the forecast’s accuracy (Granger, 1980). Today, forecasting has become an interesting
topic that attracted the attention of many researchers, authors and business person. Therefore, in this
chapter, to provide a good basic understanding, we will have a detailed review of forecasting: its
development over time and its essential characteristics.
2.2 The Role and Nature of Forecasting in business
In business, forecasting is not a new activity and organization always needs to forecast. (Ashton &
Simister, 1970). The reason is that group always operate in an atmosphere of uncertainty, but the
decision must be made today that affect the future of the organization. And to be more precise,
forecasting is one of the most critical aspects of planning (Nada, 1997 and Zinki, 1970). In 1970, the
book “The role of forecasting in corporate planning”, edited by David Ashton and Leslie Simister, has
collected many authors and journals that discussed this topic. Colin Robinson, one of the contribute
authors, stated that forecasting is something we all understand and do it quite naturally. Forecasting
is about making the prediction of the future based on the experience of the past. And in an
organization’s operation, the process of doing decision, planning and implying action is based on these
forecast. Therefore, forecasting is naturally a part of the business planning process.
Maurice Zinki, another author, also specified a critical nature of forecast: probability. The forecast is
not about absolute accuracy. It is estimated by probability, and it relies on the law of large number
and different possibility. The forecast is the statement of what we think the future is likely to look like,
rather than what it should look like (Armstrong, 1985). Likewise, E. J. Davis added: “Forecast is the
result of prediction covering variables that can be measured and qualified; and the knowledge of the
market which allows other factors to be brought into consideration to provide the best forward
estimate in the situation.”

12 | P a g e
In the early age, forecast’s practice in business has three objectives: the outcome, the time and the
change over time (Granger, 1980). There are three questions that a forecast must answer. To begin
with: What are the possible outcome (result) of the event that is likely to happen? Next: When it is
liable to happen? And finally: What may have change over time? This information can be achieved by
looking into the record/data and then, manipulate it’s by using different rules and methods to
generate relevant information. The quality of the forecast will be limited only by the availability of the
data, technology and the cost of gathering data.
However, as the role of forecasting in business increased over time, the corporation’s forecast has
developed three more addition objective’s requirement: Usability, Accuracy & Timely and Cost-Benefit
(Hanke & Wichern, 2005; Kalchschmidt, 2008).
Firstly, a modern forecast must provide high-value information to the forecast users/manager (Hanke
& Wichern, 2005). It is not only about what is the information a forecast can provide. The forecast
must answer a series of question to prove it usability value: What is the information? Who is the user?
What is the meaning and purpose of the information? To solve what need or problem? It is the best
information that can support the user to address the issue in the most efficiency way?
Secondly, the forecast must provide a high quality of accuracy and high quality of timely
(Kalchschmidt, 2008). It means that the forecast must provide an accurate prediction of the
outcome/result/error’s probability in the shortest perform time as possible (as fast as possible).
Because the global market has become more competitive and the technology gap has become easier
to catch up, a crucial mission of an organization is to control the operation in the most efficiency way.
And an accurate and timely forecast can provide many advantages for a firm’s operation control:
reduce waste and cost, maximize distribution network’s efficiency, control material using, reduce
storage level… (Tony Hines, 2013). Nevertheless, the timely and accurate forecast will be the key to
this problem.
Lastly, the forecasting practice must provide the best cost-effective benefit to the organization (Hanke
& Wichern, 2005). As James Morell mentioned in his paper (1970), the ultimate objective of business
is to maximize the profits. Therefore, all forecasting should bend toward this goal; and the forecaster
must provide the information that may lead to a profit situation in which, the firm can maximize the
benefits/profits by using the cheapest forecasting method if possible.
2.3 Forecasting types
In 1970, the structure of the book “The role of forecasting in corporate planning” (edited by Ashton &
Simister) divided its contents into five types of forecasting: Environment forecasting, financial

13 | P a g e
forecasting, technological forecasting, and Sale forecasting and human forecasting. It reflects that the
type of forecast was divided based on its purposes or function in a corporation.
In 1980, Granger classified forecasting types based on times-length. There will have two type of
forecast: short-term forecast and long-term forecast.
In 2005, Hanke & Wichern provided a more specific classify of forecasting types:
 Classified based on time-length: Short (daily, weekly or monthly forecast), Medium (termly,
seasonal or yearly forecast) and Long-term (more than two years forecast).
 Classified in term of their position on a micro-macro continuum: Small details forecast or large
summary values.
o Small detail’s example: number of sales in days, the production cost of one unit…
o Large summary’s example: The total sales in the markets, the economic situation…
 Classified according to their methods: Quantitative (statistical & numerical) or Qualitative
(interview, expert judgment…).
In 2003, Jonh H. Vanston suggested classifying the forecast based on the type of the forecaster and
their view of the future. There are five types of forecasters: Extrapolators, Pattern, Goal Analyst,
Counter-Puncher, and Intuitors.
An extrapolate forecaster believes that the future will represent a logical extension of the past. This
type of forecast is suitable in a situation where the control factors (environment, market, operation…)
are well defined and relatively constant. In this situation, quantitative method is useful, and it requires
relevant and accurate data to generate a good forecast. However, this type of forecast will not be
suitable if the environment is unstable and the driving change force is strong.
A pattern forecaster believes that the future will replicate past’s event. The type of forecast can be
applied when there already have an analogous/identical event in the past from which, the
data/information can be used for the new event/situation. A well characteristic’s analysis of both past
and new event must be carried on carefully to guarantee that the old data/information are well
understood and development to use in the new event/situation. So apparently, the central issue in
this type of forecast lays on the ability to recognize the dissimilarities between the old and new event.
A Goal analysis forecaster believes that the action and belief will determine the future. This type of
forecast is useful in a situation where some key action/factors of an organization, or environment, can
have a tremendous impact on the future outcome. For example: if the population of the world
continue to increase (environment factors), there will likely lead to a rise in foods demand; so if we
decide to invest strongly in the agriculture industry (organization action), we may likely get the change

14 | P a g e
to satisfy the foods need and then gain profits from its. However, the identification of the critical
factors/action will be the main problem.
A counter-Punch forecaster believes that the future will result from unpredictable events and
activities. The forecaster is identical to a risk analysis. It works best in a situation where the
environment is high volatile, unstable and contains many hazards. Thus, the forecaster needs to be
highly flexible because forecasting requires continuous updates and changes. And in such case, it is
hard to establish a long-term planning.
An intuitors believes that the future shaped by inexorable forces, random events and actions of
individuals and institutions. It is a mixed type of forecast. It is useful in the situation where the overall
situation is poor defined, but the change driving force is established. This kind of forecaster is flexible
and uses both quantitative and qualitative method to deal with the situation. Nevertheless, building
a forecast model, that combines a reasonable level of both quantitative and qualitative method, is
always the most challenge task.
2.4 Forecasting process
To produce an accurate forecast in the most efficiency way, the forecasting has to be carried out
systematically (Robinson, 1970). An accurate forecast will be generated not only by the using of the
powerful forecast technique, method or IT support tools; it is also the result of the whole forecasting
process that can run in the most efficiency way (Hanke & Wichern, 2005). As mentioned above, the
history of forecast’s practice changed from a spiritual way to a more systematic and logical way. The
development included the effort to generate a forecasting process that can help forecasters to carry
out the task more systematically and efficiently.
Early in 1947, G. Clark Thompson introduced a general approach for sale forecasting including many
steps. These process only focus on sales forecasting. Therefore, it cannot be used as a guideline
process for others types of forecast. As times past, the forecasting process has been continuous
developed and represented be many forecasting researchers. However, in 2005, Hanke & Wichern
described the five simple steps of forecasting process that can be easily understood and applied in
most types of forecast (figure 1):
 Problem formulation and data collection.
 Data manipulation and cleaning.
 Model building and evaluation.
 Model implementation.
 Forecast evaluation.

15 | P a g e
Step 1, the forecaster has to complete two tasks: identify and analyze the problem; and then, gather
needed data to solve the problem. Identify the key problem is a crucial task. The reason is that the
cost (financial, time, human resource) of gathering data is usually high (Granger, 1980); hence, firm
needs to identify the right problem that have the most effect on its performance and collect the
appropriate and relevant data to save the cost of collecting data.
Step 2, the gathered data must be cleaned, organized and cleaned. These data are not always useful,
complete and accurate. Some data may not be appropriate and redundant for forecasting. Thus, these
data needs to be clean down to save cost and to reduce noise information. Other data may be
unavailable or incomplete. Hence, it cannot be used in a quantitative method. In such case, the data
needs to be re-estimated or assumed to fit with the chosen quantitative method or to be used with
another qualitative method.
Step 3, firm needs to fit the collected data into a forecasting model that is appropriate in terms of
maximizing forecasting accuracy and of minimizing forecasting error. Each organization will have its
unique characteristics (Porter, 1985). Therefore, the firm needs to construct a model that is unique
and best fit its condition. The chosen model should be carefully modified and balanced in term of
complexity (Hanke & Wichern, 2005) in order to maximize cost-benefit and to adequately support
forecast’s user (manager).
Step 4, the firm implement the chosen model into the actual forecasting environment. In this step,
forecaster needs to observe carefully the result produced by the model, as well as to control the
process of gathering and consuming new data of the model. The forecasting error is then observed
and summarized because it will be used in the next step.
Step 5, forecaster will compare and analyze the different between the forecast values generated by
the model and the actual history values. The critical factor in this analysis will be the summarized error
in step 4 and the changing data that affected by the change in environment. If the error excess the
Figure 1. The five step of forecasting process

16 | P a g e
predefined acceptance level, then the process will turn back to the first step where the problem needs
to be defined and where the forecasting model needs to be modified.
The next question is: what happens after step 5? If the forecasting result is highly accurate and can
satisfy the need of the manager, then the model and forecasting process will remain to be applied.
But if the result is not good enough and, therefore, the forecasting process and method need to be
changed to generate a better outcome, what is the correct course of action? Changing of the
forecasting process is always a difficult task because it leads to a complex changing in the culture of
the organization operation, the IT system, as well as the functionality department within a firm
(Montgomery, 2006). In a paper written in 2006, Davis Montgomery suggested five keys points that
company can consider while trying to improve the forecasting process.
The first thing, the firm should do is trying to use more of the statistical forecast. This idea is proved
to be the trend of the modern forecast. With the development of IT that strongly support this method,
the statistical forecast has become a powerful tool that gradually replace the role of judgment
forecast. Further review of this topic will be discussed in another part of this paper.
The next thing to improve the forecasting process’s efficiency is to combine multiple forecasts into
one. It likely increases the accuracy of the forecast result (Hanke & Wichern, 2005). For example,
combining sales forecast input with production forecast and the market share forecast will provide a
complete view of the overall situation of demand-supply that may happen in the future. Thus,
nowadays there already have various software to support that job.
Another way of improvement is to appoint approximate effort of forecasting based on the value
contribution of the product and its forecast ability. The firm should focus their effort on their most
valuable product/service because, clearly, they are the direct income revenue of their business. For
small value, or unknown value product (new product), the effort should be limited, cut-off or be
invested carefully to reduce the loss risk. If the product has a high forecast ability (for example: good
data and information source, stable market environment…), the firm should consider the using of
statistical method to bring out the most benefit of accuracy and efficiency. However, in the case of
low forecast ability (ex: an unreliable source of data and information, volatile market…), the firm
should consider the using of the judgmental method as well as improve the customer communication
to attain the most relevant and reliable information.
The last two keys point to improve forecasting process focus on the development of a robust reporting
tools and the management of data. In such, a reliable reporting is defined by its ability to deliver the
right, and accurate, information to the right people as quick as possible. Plus, the correct input will

17 | P a g e
increase the accuracy of the forecast (Remus et al., 1998). Therefore, the firm must also focus their
effort on the management and control of collecting the right, accurate and newest data as quick as
possible.
2.5 Forecasting model
After many years of researching, in 2006, Chaman L. Jain benchmarked forecasting model into three
types:
The first group of forecasting model is time-series models. The firm that apply this model assumes
that the pattern will continue in the future. This model is used when the environment is expected to
be stable, and therefore, the old data can be used as the input for the model to forecast the future.
The most using methods in this models are the statistical methods: Average technique (simple and
moving), Simple Trend, Exponential Smoothing, Decomposition, Box-Jenkins (ARIA – Auto-Regressive
Integrated Moving Average)…
The second group is Cause-Effect models. In this models, a cause, or a driver factor (independent
variable), will create an effect (dependent variable). A firm that apply this model assumes that the
future will result from a certain specific conditions or events. Some techniques can be used in this
Models: Regression, Econometrics, Neural Network…
The last group is judgmental models. Although the statistical method has slowly replaced judgmental
method as the most used method, however, the using of the judgmental method can never be
replaced completely (Fildes & Goodwin, 2008). Furthermore, in some case of the environment, such
as the highly volatile environment or the conducting of a forecast far into the future, the judgmental
forecast is believed to produce a better forecast quality than the statistical one (Armstrong, 1985).
Some techniques that can be used: Analog, Delphi, Diffusion, PERT (Performance evaluation review
technique), survey, interview…
Furthermore, in this paper, Chaman also summarized some fundamental rules of the forecasting
model.
 Accuracy: the result produces by the forecasting model no need to be 100% accuracy. There
always have a place for error and the error allowance is decided by the manager. “Actual =
pattern + error” is the simple formula that can be used to calculate the error: the error is equal
to different between the forecast and the actual result.
 More data and more sophisticated models do not guarantee a better forecast result
(Lawrence et al., 2000). In fact, many researcher even suggested firm construct a model with
the formula as simple as possible (Hanke & Wichern, 2005).

18 | P a g e
 There are no perfect models that can be used in every environment over time. The
environment change with time causing forecast model to age with time. Therefore, forecast
model always needs to be controlled, observed and updated if necessary.
 Each model has its data requirement. Each organization has its unique characteristic and has
its environment (Porter, 1985). This unique characteristic and environment will generate
unique data for each firm. Hence, the company needs to develop their forecast model that
consumes its unique data.
 The forecast must be prepared from various stakeholder (forecaster or firm’s staff) to provide
a good forecast. This practice reduces the error causing by bias evaluation as well as reduce
the loss causing by undefined risk.
Also, Chaman also suggested that statistical forecast is no more than a baseline forecast. The result
produced by statistical forecast needs to be combined with another forecasting method, such as
judgmental adjustment, to improve the accuracy of the forecast. We will discuss this topic in the next
part of this chapter that focus on the review of different forecasting methods, primarily statistical and
judgmental.
2.6 Forecasting method
Over time, the definition and description of forecasting method are complicated. Rather than trying
to create systematic forecasting categories, forecaster usually only consider the choosing of different
forecast technique or formula, and then use it separately calculate and produce the forecast. The
category of forecasting method is not clear and is hard to gather all methods into different types of
the group with the same characteristic or purpose.
However, the development of forecasting technique and methodology had encouraged researcher
and forecaster to organize and gather different forecasting technique into a group of the same
characteristic.
In 1966, in a survey conducting of sale forecasting in America, Reichard held forecasting method into
four types: Executive judgmental, statistic method, sales force estimate and economic method. This
classification describes the forecasting practice in a firm only. It cannot clearly distinguish the different
between each method. In 1970, Simister mentioned two types of forecasting method that is statistical
methods and economic methods. The author completely ignored the value of judgmental methods.
In the same years, Davis decide classified forecasting method, based on its level of forecasting, into
three types: projection, prediction, and forecasting. The differences in these 3 level are the difference
in the volume of consumed data, the manipulation of data and the complexity level of the formula.

19 | P a g e
Although forecasting method classification is a difficult task, Davis’s paper (1970) had given a hint of
the key factor that will distinguish the different between different forecast methods: the availability
of data and information. In the book “Forecasting in Business and Economic”, Granger (1980) also
agreed with this idea and explored: “The forecast method can vary greatly and will depend on
data availability, the quality of model availability, and the kinds of assumptions made.”
In 1985, J. Scott Armstrong introduced a way to classify different types of forecasting method.
In his book, Armstrong organized forecasting method into three groups of opposite approach:
 Subjective Vs Objective.
 Naïve Vs Casual.
 Linear Vs Classification.
In the first group, forecast methods are classified based on the data availability. The forecast method
that dealt with well-specified data will be classified as an objective method (for ex: explicit, statistical,
formal method). In contrast, the method that dealt with non-specified data will be classified as a
subjective method (ex: implicit, informal, clinical, experienced-based, intuitive method, guesstimates,
wild-assed guesses, gut feeling).
In the second group, the method will be classified based on its complexity of the formula/models and
the amount of consuming data. The naïve is considered as a simple approach in which, the result is
assumed to be the same /identical with the past. In this case, the old data can be used to produce the
forecast. On the other hand, casual is a gathering of more complex methods, and it assumes that the
future result will not be the same as the past. Therefore, in order to produce the forecast, casual
methods required more data and more relationship’s evaluation of the data over the forecast horizon.
In the third group, the method is classified based on the formula/rule that defined the relationship
between the input (data, information) and the output (result, event). The linear method is a simple
way of defining the relationship between input and output. For example: if X change than Y will change
following a certain rule. However, a linear method is useful only when the relationship between input
and out is simple and then, a statistical technique can be used. For a more complex relationship,
forecaster usually uses classification method. The classification methods, in a more complex way, find
the behavioral units that respond in the same way to certain groups of other units. For example: the
total demand of the market may be effect by a group of input including GDP, interest rate, inflation
rate…
Beside this 3 group of forecasting method’s classification, Armstrong also divided forecasting method
based on two key factors: human or number. The method, in which the human factor keeps the key

20 | P a g e
role in the forecast process by incorporating intuitive judgment, opinions, and subjective estimates to
produce the forecast, is called judgmental forecast or qualitative method. On the other hand, the
method in which the result of the forecast is calculated mathematically from a set of well-defined data
number is call extrapolation method, statistical method or quantitative method.
The classification concept of judgmental and statistical method is widely accepted and used as a
framework for many modern researchers. While modern forecasters seem to focus their effort on
developing more technique, analysis and implementation method for the statistical method, the value
of the judgmental method, although it is reduced in the recent time, is proving to be unable to replace
entirely (Fildes & Goodwin, 2008). The main topic of discussion over these two methods is that: which
one is more accurate? In what situation? And how we can reduce the error and improve the forecast
accuracy? This topic will be reviewed more detail in the next parts.
2.7 Forecasting accuracy & error: Statistical Vs Judgmental method.
2.7.1 Accuracy & Error
As mentioned in the first part of this chapter, forecasting accuracy is estimates of probability (Zinki,
1970). The main challenge of forecasting is to deal with the uncertainty surrounding the future, and
therefore, forecasts should always be expressed in term of probabilities (Robinson, 1970). The
forecaster’s goal is to reduce the uncertainty, not eliminate uncertainty (Robinson, 1970). Therefore,
in the forecast, there no needs for a 100% accuracy (Chaman, 2006). Furthermore, the statistical
method will never be able to produce the absolute accuracy. Therefore, measure error is the crucial
key (Armstrong, 1985).
“What is not measured, never gets improved” (Chaman, 2007)
The key factor to evaluate the accuracy of a forecast is a forecasting error. Forecasting error is
calculated by the differences between the forecast’s result and the actual outcome (Chaman, 2006)..
In the forecasting process, we need to have a control system that will say when the forecasts are going
wrong (Zinki, 1970). Therefore, in the final step of forecasting process (forecast result evaluation), the
primary task is to evaluate the errors and then decide wherever the result of the forecast model is
accurate or not. (Hanke & Wichern, 2005). So the question here is what is the level of error’s
allowance? Or, how much error firm can afford to absorb?
Summarize the paper of Chaman in 2007, the level of errors that a firm can afford depends on:
 The cost of error: the higher the cost of error, the fewer errors a firm can afford. If the cost
(financial, time, resources…) created by the error is too high, then a company cannot afford

21 | P a g e
to absorb all the loss. Hence, in order to keep the cost as the minimum level, the firm must
improve the accuracy level by controlling the error level as low as possible.
 Adjustment ability: the quicker a firm can adjust the error, the larger error the firm can afford.
In other words, when an error appears, if a firm can quickly apply the solution to reduce the
error, then the number of errors that a firm can afford will increase.
 Industry benchmark: the amount of error a company can afford varies from industry to
industry. Therefore, to benchmark an error ruler, the firm needs to compare itself with others
companies in the same industry and then calculate the mean of error within the industry. If a
firm’s error is higher than the average, then an improvement plan must be carried on.
The next question is: what are the typical characteristics of the forecast error?
The first common characteristic is that: The farther in the future, the higher the error (Chaman, 2007).
Forecasting always content one crucial component: timeline period; and an increase in the timeline
period will lead the forecast into becoming more inaccurate (Simister, 1970). The reason is that the
using information and data will become less relevant as the time past by (Granger, 1980). And the
using of outdated, irrelevant or inaccurate data is proved to be one of the primary reason for forecast
fail (Vanston, 2003).
The second characteristic: the more detail the forecast, the higher the error (Chaman, 2007; Lawrence
et al., 2000). As a forecaster trying to conduct a more precise detail of the forecast, it will increase
the amount of the requirement data as well as the level of forecasting model’s complexity (Hanke &
Wichern, 2005). Such case will increase the chance of error. The reason is that, firstly, an increase in
data consuming will lead to a rise in data management difficulty and data’s inaccurate level (Chaman,
2006). Secondly, the using of a sophisticated forecast models are not always producing a better
forecast result (Chaman, 2006). In fact, research that published by Vanston (2003), had proved that
one of the two reason of forecasting failure is the using and combining of the inappropriate forecasting
model.
The last characteristic of error is that the errors of a company vary from industry to industry. The
reason is that each industry will have their unique nature, culture and characteristic (Porter, 1985).
That’s why the error allowance will differ from industry to industry. However, in the same industry, a
firm can still calculate the average level of error allowance (Chaman, 2007). It serves the purpose of
comparison the firm’s error with the industry average level and from that, conduct a plan for further
improvement.

22 | P a g e
2.7.2 Statistical method’s error
Since 1970, there was being a common idea about the accuracy characteristic of the statistical
forecasting method: statistical method is accuracy and useful in short-term and medium-term forecast
only (Simister, 1970; Granger, 1980; Hanke & Wichern, 2005). In the long term, as the time-length
increase, the statistical method will likely produce more error (Armstrong, 1985). The reason is caused
by the way statistical method using the data and by the using of inappropriate forecasting model
(Armstrong, 1985).
Firstly, incorrect or No information/data will reduce the accuracy of a forecast (Remus et al., 1998). In
statistical methods, there are three types of data: historical data, data in an analogous situation, and
simulated data (Armstrong, 1985). In similar interactive, there will have three reasons that may cause
the forecast failure in the long-term: Outdated data, irrelevant data and inaccurate data (Vanston,
2003).
In the case of historical data, the accuracy of the statistical method is affected by two major
conditions: the accuracy of historical data and the extent to which underlying conditions will change
in the future (Armstrong, 1985). In long-term, the older the data, the less accuracy it will become. The
reason is that the situation, condition, and environment will always change, and therefore, old data
and information will not hold much value in the long future (Granger, 1980). In a simpler meaning, the
data is outdated.
In the case that historical data cannot be obtained, after carefully evaluating, firm can consider the
choice of using the data from another similar/analogous situation or event (Armstrong, 1985). In such
case, there is a critical issue: how to distinguish the differences, as well as the similarities, between
old and new situations? (Vanston, 2003). The data from another event/situation needs to be carefully
analyzed and modified before being able to use in the new situation. A wrong or weak analysis will
create a bad set of data using the new model. Plus, in a long-term period where the situation and
environment are different, the evaluation of similarity will become more difficult. Hence, it leads to a
situation where the data in the analogous situation is inaccurate and irrelevant to use in the current
situation.
Simulated data is applied when we cannot obtain neither historical nor analogous data (Armstrong,
1985). There are two types of simulated data: laboratory simulations (data obtained in the lab) and
field simulations (data obtained in the real world). Simulated data is mainly used in the case of a new
product or new market where the environment, situation, and condition is totally new. This type of
data has an inaccurate risk in both short and long-term period (Armstrong, 1985). The biggest cause

23 | P a g e
of inaccurate data is not about the timeline period of the forecast, but is about the bias of the lab
researcher (or the researched market) that may subjectively affect the simulated data.
Secondly, forecast model aged with time and will need to be updated in the long-term, otherwise, it
will produce an error (Chaman, 2006). The future environment will not always be the same, and the
pattern may change at any time (Chaman, 2007). If the firm doesn't have a process to deal with such
change, the use of inappropriate forecasting model may produce inaccurate forecast result. Most
statistical methods used historical data to calculate the result (Robinson, 1970) and these methods
will never be able to produce the absolute accuracy (Armstrong, 1985, Hanke & Wichern, 2005). As
time length increase, the environment will change, and the data/information of historical data will
reduce its usability (Granger, 1980). Therefore, firm need to be ready to develop a new model that
can best use the nearest and newest data and reduce the error.
2.7.3 Judgmental method’s error
When quantitative method had not been widely developed and used, the judgmental method was the
most used forecasting practice in the past (Nada, 1997). And throughout its development history,
there are two main reasons that were causing an error in judgmental result: bias judgment and lacking
in communication.
Firstly, lacking in communication also reduce the accuracy of the judgmental forecast (Nada, 1997).
The two types of communication lacking that cause error in judgmental forecast is:
 Lacking in communication with the market and customer (Nada, 1997): the market forecast is
one of the most important jobs of a forecaster (Davis, 1970). The primary objective of the
market forecast is to calculate and compare supply and demand’s ratio of a product, or
service, in a target market. In order to forecast the demand and supply’s ratio, the firm has
two sources of information: internal information (historical data of sales, production
capability,…) and external information (competitors, market, customers…). While internal
information is a task that a firm can actively manage and take control of, external
information’s management was remaining a critical challenge nowadays (Myerholtz &
Caffrey, 2014), the market is hard to predict because of the overload information:
various promotion, variation in large customer purchase, competitor prediction
difficulty… As consequence, instead of gathering useful and relevant information for
the judgmental method, a lacking in market communication will gather noises and
irrelevant Intel that may mislead the forecaster and cause an error to the forecast

24 | P a g e
result. Hence, firm needs to improve market communication to get better information about
the environment, competitors, as well as customer’s needs.
 Lacking in communication within the forecasting process’s stakeholders (Brown, 2011): in his
research, Brown discovered that, because of communication problem, there have some
conflicts of using the judgment forecast within a firm. The first conflict is caused by the
pressure of establishing a forecasting practice balance between cost and benefit. In this
conflict, the pressure of minimizing cost and maximizing revenue may affect the forecast
accuracy. For ex: forecaster may decide to sacrifice the forecast accuracy to reduce the cost
of the forecast. Hence, firm needs to improve its communication between top manager and
forecaster to ensure that a reasonable cost-benefit balance is well defined. The next conflict
is caused by the lack of communication between different roles or departments within affirm,
for example: sale role and forecast role. In this case, forecaster may tend to under-forecast
the sale target to improve the accuracy credits from the top manager; while a sale manager
may increase the sale predict to improve their performance rating. Once again, this type of
conflict can be reduced by increasing the communication between the different department
and by producing forecast result that is combining multi-stakeholders (Chaman, 2007).
Bias is the second reason that has been quoted the most by many authors that cause the error to
judgmental forecast (Armstrong, 1985; Nada, 1997; Fildes et al., 2009; Lawrence et al., 2000;
Makrkadis, 1986). In reality, there are two types of judgmental forecast’s bias: over-forecast (positive
or optimism) and under-forecast (negative or pessimism) (Armstrong, 1985; Nada, 1997; Fildes et al.,
2009). While over-forecaster is likely to produce a result higher than the actual outcome; under-
forecaster is likely to produce a lower result. The motivation of these bias decision depends on the
forecaster. However, in a survey in 1997, Nada showed that the majority of manufacturers firm
preferred to under-forecast (58.4%) while the number of over-forecast practice is only 15.8%; and the
rest (25.8%) preferred neither direction. Fildes et al. (2009) stated that negative, judgmental forecast
produces a higher level of accuracy than positive forecast. The reason has been explained furthermore
by Charlotte Brown in 2011 that forecaster will tend to under-forecast result in order to improve the
accuracy credits from the top manager.
Then how to reduce the bias in judgmental practice and increase the accuracy of a forecast? The most
popular method is to combine both statistical methods and judgmental methods (Strong, 1956;
Makrkadis, 1986; Hanke & Wichern, 2005; Montgomery, 2006; Chaman, 2008; Fildes & Goodwin,
2008; Nada, 1997; …). We will have further review of this topic in the next part.

25 | P a g e
2.8 Improve forecast accuracy: Integrate statistical and judgmental method.
As mentioned in the last part, a statistical method is accurate and useful in the short-term and
medium-term forecast (Simister, 1970; Granger, 1980; Hanke & Wichern, 2005). But, on the other
hand, the farther the future, the less accurate forecast which the statistical method using historical
data will produce (Armstrong, 1985). For long-term forecast in the far future, where there are few
number of historical data and where there has a high level of uncertainty and unusual, a judgmental
method is the better tool comparing with statistical method (Hanke & Wichern, 2005; Armstrong,
1985; Makrkadis, 1986). Some judgmental methods, such as scenario writing, may encourage the long-
range thinking of the top managers to prepare a plan for recognizing and reacting to unusual
environment changes (Hanke & Wichern, 2005). Furthermore, different level of uncertainty will
produce the various level of using the judgmental method: the more uncertainty, the higher
judgmental method level that the firm will rely on (Charlotte Brown, 2011, Davis, 1970). Similarly, in
a situation of low uncertainty, where forecaster believe that the future will follow a pattern and result
in the same with the past, the firm should focus on the using of statistical method to produce the best
performance (Hanke & Wichern, 2005).
Nevertheless, there is some paper that raised the doubt about the accuracy level of judgmental
forecast in the far future (Connor, 1993, Makrkadis, 1986), as well as the accuracy of statistical
methods in the near future (Wichern & Hanke, 2005). Wichern & Hanke stated that: “In some
situations, such as unusual circumstance, history data may not be an accurate predictor of the future.
The amount of judgment injected into the forecasting process is increased if the historical data are
few or are judged to be partially irrelevant”. He also suggested that using a computer to conduct
statistical practice can only provide the numbers but hardly provide an in-depth view of the true
nature and quality of the forecast. On the other hand, Connor (1993) and Makrkadis (1986), both
concluded that in time of change, human judgmental perform worse than statistical method. Then, to
improve forecast accuracy, they suggested that forecaster should conduct judgmental forecast after
using the statistical result as the baseline. These two opinions had one common idea: in any situations,
short or long term, stable or high uncertainty, integrating statistical and judgmental methods together
is the right practice to improve the forecast accuracy.
In fact, the suggestion of combining different forecast methods together to improve the accuracy of
the forecast in not a new idea. In 1956, based on a survey, Lydia Strong showed that there had s shift
to using computer and technology (on data gathering, storing and analyzing) into the forecast.
However, human factor remained an important role in the forecast process. Davis (1970) also stated
that the using of statistical alone is impossible because not every information can be put into an
equation to calculate the forecast; and therefore, a combination of both statistical and judgmental

26 | P a g e
method is necessary. In fact, although the statistical method and new technology have been strongly
developed in the past two decades, the using of judgmental forecast still play a significant role in
forecasting practice (Nada, 1997; Armstrong, 1985; Makrkadis, 1986; Fildes et al., 2009). The
combining of both judgmental and quantitative method are necessary for a forecast practice (Hanke
& Wichern, 2005). Then the raising question is: what is the most efficient and appropriate way to
combine statistical and judgmental method? (Montgomery, 2006)
Many researchers suggested that the ration between statistical method and judgmental method in a
firm’s forecasting practice should be based on the level of uncertainty in the future (Hanke & Wichern,
2005; Armstrong, 1985). The more uncertainty the future, the more judgmental method will be used
(Charlie Brown, 2011). That the reason, as mentioned above, many authors suggested that
judgmental will be a more powerful tool to forecast in the long-term (Hanke & Wichern, 2005;
Armstrong, 1985; Makrkadis, 1986). Nevertheless, in 1993, Connor proved that even in a situation of
high uncertainty, the judgmental method produces a worse performance than statistical method
(Connor, 1993). He suggested that forecaster should use the statistical result as the base firstly and
then conduct a judgmental adjustment later. This suggestion also gets supported by Chaman (2008),
Goodwin (2000), Fildes et al. (2009) and Michael at al. (2006). And in reality, this practice has been
reported to be the most used in the modern forecasting process (Fildes & Goodwin, 2008; Fildes et
al. , 2009).
Still, there remains one issue, judgmental adjustment of statistical method’s result often reduce the
accuracy of the forecast (Fildes & Goodwin, 2008). To solve this problem, Paul Goodwin (2000, 2002)
suggested that firm needs to have a control policy over the process of using judgmental methods to
adjust the result produced by the statistical method. There are two ways to integrate judgmental with
statistical method: voluntary integrate method (Goodwin, 2000), where forecaster is free to ignore,
accept or adjust the result produced by statistical methods; and mechanical integrate method (Lim &
Connor, 1995), where the statistical result will be corrected or combined (calculate the mean of the
result or error) with a separate result produced by the judgmental method.
In the case of voluntary integrate, the outcome is likely to accept by both forecaster and manager
because forecaster has the change to modify the result to be most acceptable by different
stakeholders. But, “it may lead to inefficiency and downgrade of the forecast’s accuracy”-Goodwin
written in 2002. To reduce the level of bias in voluntary integrate method, the process of using
judgmental adjustment can be controlled by three solutions. Firstly, by issuing an adjustment request
form, the company can improve responsibility of stakeholders who involves in the forecasting process.
Secondly, the reason of adjustment must be justified and explained. And lastly, forecaster must be

27 | P a g e
continuous review actual outcome with forecast result (Goodwin, 2000). On the other hand,
mechanical integrate method will likely to produce more accuracy forecast in many situations
(Goodwin, 2002). The needed condition is that there is a separation between forecaster and forecast
users in order to eliminate bias and mutual affection between different roles.
2.9 Conclusion
2.9.1 Summarize Chapter 2
In business, forecasting is a natural part of the business process (Colin Robinson, 1970). It serves as an
input factor for the operational planning stage (Nada, 1997; Zinki, 1970). Forecasting is all about
dealing with uncertainty in the future (Armstrong, 1985). Hence it needs to be addressed in term of
probability (Zinki, 1997). Forecasting practice in business has three objectives: the predicted event in
the future, the time when the event will likely to occur and the changes in outcome that may occur
(Granger, 1980). Also, a forecasting in business has three others requirements that are: usability,
accuracy (in outcome and in time) and cost-effective (Hanke & Wichern, 2005).
The forecasting process has five simple steps (Hanke & Wichern, 2005):
 Problem formulation and data collection.
 Data manipulation and cleaning.
 Model building and evaluation.
 Model implementation.
 Forecast evaluation.
The type of forecasting can be classified based on: Time-length, the scope of forecasting or forecasting
methods (Hanke & Wichern, 2005); and the point of view of the forecaster: Extrapolators, Pattern,
Goal Analyst, Counter-Puncher or Intuitors (Vanston, 2003). In modern forecasting practice, there are
three types of forecasting model: Time-series models, Cause-effect models and judgmental models
(Chaman, 2006).
In 1985, Armstrong suggests dividing forecasting techniques into two main methods: judgmental
methods (human approach, qualitative analysis, subjective, based on individual experience and
opinion) and statistical methods (mathematic approach, quantitative evaluation, objective, based on
logical and the using/calculating of the number).
Statistical methods are useful in short-medium time forecast (Simister, 1970; Granger, 1980; Hanke &
Wichern, 2005). However, the farther the future a firm try to forecast, the more error and the less
accuracy the methods will be (Armstrong, 1985). The reason is that: firstly, the data using in statistical
will be irrelevant, inaccuracy and outdated in the far future (Vanston, 2003); and secondly, the model

28 | P a g e
using to calculate in the statistical method will be inappropriate as the future environment changing
(Chaman 2007). Furthermore, the statistical result cannot provide an in-depth understanding of the
forecast and the root-cause of the changing (Hanke & Wichern, 2005).
Judgmental methods is said to be useful in the far future or in a time of high instability (Armstrong,
1985). However, some research has proved that even in time of change and in the far future, the using
of judgmental forecast performs worse than the using of statistical methods.
To improve the accuracy level of a forecast, statistical and judgmental methods should be used
together (Hanke & Wichern, 2005). Chaman (2006) suggested that the best way of combining both
methods is to use statistical result as a baseline for any addition judgmental adjustment. Plus, the ratio
between statistical methods and judgmental methods should be based on the level of uncertainty and
the availability level of the data: the more uncertainty and low level of data availability, the more
judgmental methods should be applied (Hanke & Wichern, 2005, Armstrong, 1985).
2.9.2 The remaining questions
 How firm imply these forecasting theory in practice?
 What is the role of information technology and how it can support the forecasting practice?
 In emerging markets (developing countries) where there a high level of uncertainty and low
IT level to gather data/information, how firm combine judgmental and statistical together to
increase forecast accuracy?

29 | P a g e
Chapter 3: How Information Technology supports forecast practice in
manufacturing firm
3.1 Introduction
In the last chapter, this paper represented a detail literature review about forecasting and its
development until today. However, for me, there remains two question. Firstly, how firm imply these
forecasting theory in practice. And secondly, what is the role of information technology and how it
can support the forecasting practice? This chapter will try to explore and answer these questions.
The structure of this chapter will be divided into four parts: Firstly, how firm organize its forecast
function? Secondly, why demand forecast are so necessary for manufacturing company? Thirdly, what
is the role of IT and how it can support the forecasting practice? And lastly, we will consider the cost-
benefit aspect while deciding to invest the forecasting technology.
3.2 How firm organizes forecast function
3.2.1 Role and position
Forecasting is a natural part of any business organization (Ashton & Simister, 1970). Nowadays, based
on a survey conducted in 2007 by Chaman L. Jian, 99% of the company’s top manager recognized the
important role of forecasting and supported the establishment of a forecast process. Among them,
57% of the company has already established a forecast function within its operation. Based on this
survey, the top 6 department, in which the forecast function resides, are: Operation and production
(27%), forecasting department (19%), marketing (12%), sales (10%), logistic (7%) and finance (7%)
(Figure 2.).

30 | P a g e
Figure 2. Where forecasting function resides
Each firm, depend on its uniqueness, will organize its forecasting function different comparing to
others firms. Following Hanke and Wichern (2005), the role and location of the forecast function in a
company will depend on three conditions:
 The size of the firm: because of the limitation in resources, the forecasting task in a small-
medium company also be carried out by the forecast user. For example: sales manager will
take responsibility for sales forecast, production manager will take responsibility for
production forecast… While big corporate, relying on its resources capability, can invest
strongly into the forecast function and separate the forecast responsibility into an
independent forecasting department.
 The nature of the firm’s management style. For example: if the firm’s strategy is to focus on
the creation of new production/service, then the firm can reduce its forecast function and
invest more in research and development department.
 The important level of the forecast that related to the decision-making process and
production. For example, if technology forecast is evaluated as not important for a firm in
short future, or for the improvement of production, then the firm can decide to reduce the
invest and effort on technology forecast.

31 | P a g e
3.2.2 Staff
Large company use forecasting specialist more common than the small-medium company (Hanke &
Wichern, 2005). And as mentioned above, in many businesses, forecasting tasks are produced, not
only by forecaster specialist but also by the manager or forecast users. For example: sale manager or
marketing manager will forecast sale forecast, production manager will forecast production forecast…
This situation happened is most of small-medium organization where the firm cannot afford to hire
an expert in the forecast. But in a large group, where an expert can be hired and forecast department
can be organized, the forecaster can hold crucial role with a high salary (Chaman, 2007). These experts
can be used to all department within the organization to generate an accurate and adequate forecast’s
result. However, a lacking in communication and corporation between forecasting expert and forecast
users (managers) will reduce the quality of the forecast (Hanke & Wichern, 2005).
In 2006, Chaman L. Jian conducted a survey to benchmark the background of forecasters. In this
investigation (Figure 3), a combination of 93% of the forecasters earned the degree of Bachelor or
Master; the rest 8% finished the high school and only 1% of the forecaster achieved the Ph.D. The
major field that was most studied are business; in which 31% are focus on supply chain (production,
distribution, and logistics), another 36% are specialise in sale and marketing (sale, product and market
knowledge, marketing research), and only 8% had an information technology background (Figure 5).
It clearly described the gap between information technology and business in the field of forecasting.
A forecaster, despite their knowledge in the business field, lacks the knowledge in the information
technology field. Therefore in reality, the most firm still experienced difficulties while applying new
information technology into the forecast.
Figure 3. Highest Academic Degree Acquired by Forecasters

32 | P a g e
Figure 4. Forecasters. Major Field of Study in University
Figure 5. Business Background of Forecasters
3.2.3 Forecast practice
In 1997, Sanders Nada conducted a survey that was gathered from 86 manufacturing firm in the US
with an average annual sale range of $5 million to $10 million, to outline the picture of forecasting
practice in business. The survey showed that judgment forecast is the most used method. In fact,
despite the substantial development of the statistical method, a judgmental method is still used in
almost every companies as a regular basis. The judgmental process was organized in 2 ways:
structure/formal approach and unstructured/informal approach. The ratio between the using of these
two approaches by manufacturer firm was somewhat a little favorited to the informal approach:
54.2% over 43.1%. Also, the survey also showed that 52.8% of the firm conducted the judgmental

33 | P a g e
forecast as a group of more than two people; the rest: 45.6%, Conducted by individual and 1.6% could
not decide the preferred method.
Because of the development of the statistical method, there was an increase of the complexity level
by combining multiple forecasting techniques. Surprisingly, the most popular quantitative technique
is not the newest complex development one, but the most simple approach: naïve method; where
firm assume that the future will act the same with the past and produce a forecast by using only the
historical data. Still, the most firms admitted that naïve method was not employed on a regular basis
because of its risk of causing an error. Another interesting point is that to conduct statistical method,
a total of 89% of the firm had used some types of forecasting software.
Later on, in 2006, a survey that was conducted by Chaman L. Jian, showed a big difference (Figure 6)
between the using of forecasting models from judgmental methods (11%) and the using of statistical
approaches like times-series models (72%). Nevertheless, the using of judgmental forecast still keeps
a major role in the practice of many organizations. As showing in a paper, written by Robert Fildes et
al. (2009), 80% of the companies still using judgmental adjustment on the statistical forecast’s results.
Figure 6. Modules Used in Forecasting
In times-series model (Figure 7), the most used method was Average/Simple Trend (60%). How far in
the future firm want to forecast? Based on another survey of Chaman in 2008, while carrying this
model, the most company will try to forecast one year (or more) ahead (Figure 8). Plus, the forecast
cycle will be carried mostly on a monthly basis (Figure 9).

34 | P a g e
Figure 7. Times Series Models Used
Figure 8. Forecasting Horizon

35 | P a g e
Figure 9. Forecast Buckets
In cause and effect model, the most used method was Regression (80%); while the most used method
in the judgmental model was surveyed (50%) (Figure 10 & 11).
Figure 10. Cause and Effect Models

36 | P a g e
Figure 11. Judgemental Models Used
3.3 Forecasting in manufacture firm: Focus in Demand forecast to support supply chain.
In the survey conduct by Chaman in 2007, we can also see clearly why demand forecast and supply
chain are so important for business (Figure 2). In the top 6 departments in which the forecast function
is positioned, four of them are related to the supply chain that are: Operation & Production (27%),
logistic (7%), marketing (12%) and sale (10%). It means that a total of 56% of the companies (all
industries combined) put their forecasting effort on supply chain (logistic, operation and production)
and demand forecast (sales and marketing).
In the recent years, demand forecast has been mentioned and studies as one of the most important
and biggest challenge of the modern manufacture industry (Tony Hines, 2013; Charles, 2013; Fildes &
Goodwin, 2008; MyerHoltz & Caffrey, 2014; Right 90 Inc., 2010). Then the question is: why demand
forecast, but not another type of forecasting, attracted the attention of so many manufacturers? The
answer that has been explained by many authors and forecast’s experts is that: the firm must improve
the demand forecast accuracy to improve the supply chain process.
3.3.1 Why supply chain and how it links to demand forecast?
Supply chain, in the term of business, is a system that transfer a product or service from supplier to
customers. In Supply chain, through organization, labor force, activities and process, natural resources

37 | P a g e
and raw materials have been transformed into a finished product that will be delivered to the end
customer. And In the modern manufacture industry, supply chain play a crucial role in the operation
of a whole organization.
In the manufacturing industry, the role of the supply chain is a crucial one (Tony Hines, 2013). The
purpose of the business and the supply chain are to provide the supply (product or service) to meet
the demand of the customer. To maximize revenue and minimize cost/waste of production, producing
a right balance between sale and inventory (or between demands forecast and supply chain) is an
important task that manufacturer must pay close attention (Charles W. Chase, 2014). If there are no
demand, there will be no supply needed; and similarly, if there are no supply chain, there will be no
product for the consumer. Therefore, the linking and management between forecasting demand and
supply chain management are a crucial requirement for any manufacturing firm if they want to
maintenance, or push, their business.
3.3.2 The role of demand forecast in supply chain
The process of supply chain can be simply described as following order (Chaman, 2008):
 Forecasting the demand/market.
 Discussing the result with the operation & production manager.
 Analyzing the balance between demand (market opportunity) and supply (production ability).
 Planning the production schedule and process.
 Implementing the production plan.
As we can see, the information provided by a sales forecast can be used mainly for the planning
purpose of the production (Lydia Strong, 1956; Fildes et al., 2006). In a report produced by Right 90
Inc. (2010), 74% of manufacturers surveyed consider demand forecast as critical to achieving their
business objectives (Figure 12). “Sales forecast data is used in many business-critical decisions made
by key operational areas such as Finance, Corporate Management, Operations, and Marketing. The
sales forecast informs management decisions on nearly every aspect of a manufacturer’s business,
including budgeting, cash flow, expansion, investments for capital equipment and raw materials
purchases, inventory management, product positioning and placement, production planning and
manufacturing scheduling, and HR planning, staffing, and hiring.” (Figure 13).

38 | P a g e
Figure 12. The important of Sales Forecasting
Figure 13. Business Areas that use Sales Forecast Information
As the time pass, supply chain theory and practice has also evolved from an operation focus into more
strategic focus (Tony Hines, 2013). The question of how to make supply chain work in the most
efficiency way is the fundamental concept of an operation focus. On the other hand, a strategic focus
will deal with the future problem such as how to make the supply chain work in the most effective
way. And the supply chain’s strategic concept of customer focused and market driven has been widely
shared my many researchers (Tony Hines, 2013). This concept features the role of demand forecast as

39 | P a g e
the driver factor for the whole supply chain system behind where the entire production line is planned
and based on. That’s why, understanding customer and predicting the demand become an important
task for any supply chain director. In another word, demand forecast has become a crucial part of the
modern supply chain process.
Apparently, the role of operation planning and forecast is improving significantly in the recent years
(Chaman, 2008). That the reason in the past year, developers in both the business and IT fields had
been put their effort on the development of new technology that can improve the quality of demand
forecast and supply chain management.
3.4 How IT support demand forecast.
In the recent time, information technology has been proving to be a powerful tool to support the
demand forecast (Fildes & Goodwin, 2008). Following the development of more and more statistical
method, there clearly have a shift to using more computer and technology into the forecasting
practice of many organizations in term of data gathering, storing, calculating and analyzing (Lydia
Strong, 1956). It fits perfectly with the introduction of DIKW hierarchy (data, information, knowledge,
and wisdom) where information technology will be significantly useful in two aspects (Ackoff Russell,
1989): Gathering and storing data at the data level; automate cleaning, calculating and analyzing data
into information with meaning at the information level. Furthermore, modern information technology
also provides the ability of understanding and taking entirely the value of the data (Charles W. Chase,
2013). A good information technology implementation can allow the firm to improve organization
performance, as well as to develop a strategy for achieving competitive advantage (Dewett & Jones,
2001).
In demand forecast, the three IT solutions that can be used to improve forecast performance are:
3.4.1 Spreadsheet tools and Forecasting software package
Forecasting software package is a stand-alone software that has been developed for forecasting
purposes of an organization (Chaman, 2008). In this software, where various forecasting models were
included for choosing, a forecaster (or firm) can decide and apply the most appropriate model that
will be used in their forecasting system. In the case firm cannot say which models will be used, these
software’s built-in expert system can suggest some forecasting solution that can best-fit the
company’s situation. The figure 14 below represented some most used forecasting software in USA
(Chaman, 2007):

40 | P a g e
Figure 14. Market Share of Different Forecasting Software Packages
Alongside with this forecasting software, a high number of the firm still develop its forecasting model
based on some spreadsheet tools such as Microsoft Excel or IBM Lotus. These tools had been
developed in the 90’s (Lotus: 1983, Excel: 1993) and were familiar with many firm and forecaster at
that time. That is the reason the using of these spreadsheets still seize a significant role in many
organizations (Figure 15) (Chaman, 2007).
Figure 15. Market Shared of Forecasting Packages Vs Spreadsheet Packages

41 | P a g e
3.4.2 Information system with a forecast function
In demand forecast, the likely method to be used is the statistical method by which historical data will
be utilized (Robinson, 1970). However, to produce a good forecast, forecaster needs to understand
the meaning behind the using of these data. Therefore in most organization, demand forecast is part
of an information system that links directly to supply chain management (Fildes et al., 2006).
Understand this requirement, many technology firms had tried to integrate its MIS products
(information management system) with an addition function of the forecasting tool. By using this
information system, the firm can improve the forecasting performance by reducing the manual labor
and improving the speed, volume, and quality of the internal data (Jaana Auramo et al., 2008). Some
forecasting systems are showed below (Figure 16) (Chaman, 2007):
Figure 16. Market Shared of Different Forecasting Systems
3.4.3 Data mining technology and Big data
As mentioned above, a new concept of demand-driven strategy has been widely shared as the new
solution for the management of supply chain (Tony Hines, 2013). This concept requires forecast
demand to act as the key driver for the whole production process and therefore, it requires the firm
to imply a new forecasting approach: predictive analytics (Charles W. Chase, 2014).In this new
approach, firm are required to develop a dynamic demand forecast that not only use the historical

42 | P a g e
data of the past, but also use a predictive system in which the data are gathered, cleaned and accessed
directly from the vast of downstream customer’s demand (end consumer) (Charles W. Chase, 2014).
In the past, the most used method of collecting downstream data from the customer was surveyed
(Davis, 1970). Nowadays, the development of modern technology in communication and network
(internet, smartphone, wireless, laser scan, GPS…) has allowed firm to conduct more method to collect
data such as: retailer point-of-sale (POS), syndicated scanner sources, consumer panel, social media
(Facebook, Twitter) and Internet of things. However, it also creates new problems: data overload, un-
uniform data, and noise data. Hence, to solve these issues, a new technology has been developed and
attracted much attention in the recent years: Big Data.
In his paper in 2013, Charles W. Chase described Big Data as technology tool with the ability to gather,
store and cleanse both structure and un-structure data from the internet or organization internal
information. The current technologies that can support Big Data implantation are Hadoop and cloud
computing. The Application of Big Data can bring out some real advantage such as: automatic sensing
demand signal and shaping the trend of demand in the future; mining loyal customer data; generating
retail coupons at the point of sales (POS); sending purchase recommendation at the right time to
consumer; analyzing data from social media; evaluating root cause of failure.
3.5 The Cost of IT investment in forecast
“The ultimate objective of business is to maximize the profits. All forecasting should be bent toward
this end, and the forecaster must consider the function of the firm and the way its activities leads up
to a profit situation” – James Morell, 1970.
The investment in IT has become necessary for any organization. However, strong investment in
information technology doesn't always guarantee the high return of benefits (Bruce et al., 2006). In
the paper, Bruce and the other authors demonstrated that: after controlling all other factors, IT
investment is proved to be positive and significant related to the increasing amount of error in
financial. In another word, a firm with a high level of IT investment will be hard to predict its future
earning and financial situation. The paper suggested that because of the risk of high failure level of IT
project, the firm must carefully consider its budget on IT. Otherwise, the business may suffer a
significant loss in financial.
In another paper, Charles W. Chase (2014) pointed out three main challenges while trying to apply
innovation to business forecast. Firstly, because the new technology develops too fast, the cost of
purchasing new forecast technology is not yet saturated and relatively too high for the most small-
medium organization. Secondly, some forecast technologies (such as big data) are defined as “too new

43 | P a g e
to apply”. These new technologies require more time to evaluate, decide and develop an appropriate
way of using in the real business. An inappropriate implementation may cost firm more than just
financial loss. Thirdly, the cost of data of data is also a significant challenge. New forecast technology
means that there will be a new requirement of consuming data (new amount, type, definition, quality).
These cost of data includes the cost of gathering new data from the market or within the firm; and the
cost of integrating data between old and new forecast system.
To calculate an appropriate budget for IT investment in the forecasting process, firstly, the firm must
understand the three primary cost of a forecasting system: development cost, maintenance cost, and
operation cost (Armstrong, 1985).
 Development cost. The three main factors that affect the development cost of a new
forecasting system are the amount of data needs, the complexity of the method and the
implementation cost. The more complex of the choosing forecast method is, the more
consumed data will need to be gathered, and, therefore, it will require substantially more
resources (financial, time and workforce) to implement new forecasting system.
 Maintenance cost. After the firm finishing the implement of the forecasting system,
maintenance cost is also important. Similarly to development cost, the cost of maintenance
will be positive related with the level of forecasting’s complexity and the amount of consuming
data.
 Operation cost: when the forecasting system is fully developed and run smoothly, the
operation cost will be stable and has little change over time, unless firm decides to update or
develop a new forecasting system.
Secondly, the company must consider the costs of IT investment. There are two types of IT investment
cost: Direct-cost and addition cost (Hanke & Wichern, 2005).
 Direct-cost: direct cost a financial cost that firm must spend while applying new IT system.
This cost includes computing hardware cost, forecasting software and IT staffs. These costs
can be track and quickly identify by carefully researching and comparing the IT market.
 Other addition costs include Time cost, regulatory cost, process changing cost, data gathering
cost. These costs are hard to identify because of the uniqueness of each organization (current
environment, adaptability, working culture, staff, the forecasting requirement…). Still, these
costs play a crucial role in many aspects such as consistent strategy, long-term financial
planning, competitive advantage, the cost-benefit balance… Therefore, the top managers
must carefully consider all of these aspects before making a decision regarding the budget for
IT implement.

44 | P a g e
3.6 Conclusion
3.6.1 Summarize Chapter 3
To answer the first question of how the company conducts the forecasting process in real practice.
Firstly, base in many surveys, we can see that manufacturer spends most of their forecast efforts into
two parts: Supply chain management (operation, production, inventory, purchase & logistic) and
predict the future demand (marketing, sales, and production knowledge) (Chaman, 2007). In fact,
these two parts are linked together with the demand forecast as the forefront input for the whole
supply chain following behind (Tony Hines, 2013). And with the rising of the new supply chain concept
of demand-driven strategy, the role of demand forecast has become an important factor in improving
business’s performance (Right 90 Inc., 2010).
Second thing, we can see that clearly, there a shift of using more statistical method in the modern
forecast, especially the using of time-series forecast models (Chaman, 2006). However, judgmental
methods cannot be replaced entirely by the statistical method. 80% of the statistical results still get a
judgmental adjustment by the top manager (Files et al., 2009).
Modern information technology is a powerful factor that can support demand forecast not only by
gathering, storing and cleaning the data (Lydia Strong, 1956) but also by providing the ability of
understanding and taking entirely the value of the data (Charles W, Chase, 2013). Some information
technology tools that are available for demand forecast are stand-alone forecasting software, classical
spreadsheet tools, information system integrating forecasting function, and new technology that can
support data mining (such as internet, smartphone, wireless, laser scan, GPS, and Big Data).
However, while deciding to invest in forecast technology, the firm must carefully consider the cost-
benefit aspect of forecasting. The reason is that: implement information technology into the forecast
is very costly (Charles W. Chase, 2013); and new IT project has a high risk of failure (Hanke & Wichern,
2005). There is two type of forecasting cost: the cost of developing and applying a new forecasting
system (Armstrong, 1985) and the cost of investment in IT (Hanke & Wichern, 2005). Then top
managers should sit down together, consider all these costs and carefully set up an appropriate
strategy, plan and budget for the IT investment.
3.6.2 The remaining question
Most of the survey finding is conducted in developed countries where the level of IT development is
high, and where there are many big corporate that have powerful resources (financial, staffs &
infrastructure) to invest in the newest forecast technology. However in small-medium companies,
where the resources are limited, and in emerging markets (developing countries), where the

45 | P a g e
environment is different, and the IT level is lower, how forecasting has been carried out? What are
the problem and challenges? How firm deal with it?

46 | P a g e
Chapter 4: Research Finding and Analysis
4.1 Introduction
In the last chapter (chapter 3), we raised two question: How manufacture firms implemente
forecasting practice in reality? And how information technology has been used to support the
forecasting process? Furthermore, in chapter 2, there remains another question: In an emerging
markets (developing countries) where there a high level of uncertainty and low level of IT
implementation, how manufacture firm combine judgmental and statistical together to increase
forecast accuracy?
To answer all these questions, we will conduct a qualitative analysis and in-depth research of Scavi
Viet Nam, our main study case in this chapter. Scavi Viet Nam satisfies all conditions of our questions:
firstly, it is a manufacturing firm; secondly, it resides in Viet Nam that is a developing country with an
emerging market; and lastly its products specialized in the textile garment and fashion industry where
there is a high level of forecasting uncertainty.
4.2 Scavi Viet Nam - Overview
Scavi Joint Stock was established in 1988 as the first foreign invested company in Viet Nam’s textile
garment industry. Scavi belongs to Corele International group in France that is one of the top 3 leading
company in French Lingerie Industry. Its service specialized in the production of textile garment and
luxury lingerie such as luxury underwear, core underwear, nightwear, home wear and children wear.
After 27 years of development, this is the current information of Scavi in 2015:
 5 manufacture factories: 4 in Viet Nam (Bien Hoa, Bao Loc, Hue and Da Nang) and 1 in Laos
(Vientiane).
 2 Represent office in France and China.
 Headquarter: Scavi Bien Hoa (Bien Hoa city)
 Appropriate 10,000 employees in Viet Nam, Laos, and France.
 Main market share: France, European region, United States and Viet Nam
 Service: Make-to-order. Produce good at the requirement of the customer.
 Customer: commercial customer, wholesale customer, fashion company and end consumer
 Annual revenue: approximate $80,000,000.
 Annual profit after tax: approximate $2,000,000.

47 | P a g e
 90% revenue come from oversea export, commercial customer, wholesale customer and
fashion firm with a big order. The rest 10% come from retailer and end-consumer in the
domestic market.
The organization structure below (Figure 17) was taken from the document “SCAVI Organization &
Overall Information system”, provided by Mr. Nam Quoc Tran – Head of Scavi IT Department - in 2015.
Scavi has 3 departments: Commercial & Supply Chain, IT department, and Finance/HR/Administration.
Figure 17. Organization Structure of Scavi Viet Nam
4.3 Information Technology level of Scavi
The IT department of Scavi has 14 staffs and divide into three groups: Application development,
System/Network Maintenance and Help desks/support team.
 IT operation cost: ~$70,000/year. (Scavi, 2013)
 Future investment budget planning: ~$200,000/year (Appendix. C)

48 | P a g e
The current information technology level below was provided by Mr. Nam Quoc Tran-Head of IT
department- in two document: “SCAVI Organization & Overall Information system” (2015) and
“Information system: As-is & To-be.” (2014).
4.3.1 Hardware and infrastructure:
4.3.2 Network Components:
 Lan system: A VLAN
 Internet connection: connect to internet based on 2 Fiber cable from ISPs: VDC and FPT.

49 | P a g e
 Router: Cisco.
 Switch: Cisco Linksys & Dell Powerlink
 Protocol: TCP/IP protocol
 Cable: Fiber & Cat 5,6 cable.
 User: 500 users
 Server: 11 severs
4.3.3 Basic Software Architecture:
Operating Systems:
 OS for server: Window Server 2008 Enterprise
 OS for clients: Window XP SP2, Window 7 Pro
Messaging:
 Mail server: Exchange POP 3, Exchange Server 2008 Ent
 Mail client: Microsoft Outlook 2007,2010
Web:
 Web Browser: Internet Explorer 7.0, 8.0
 Web Server: Internet Information Server (IIS).
System and Network Management:
 Firewall: Astaro Firewall, TMG 2010
 Database Server: SQL Server 2008 Enterprise
 Antivirus: Kaspersky Anti-virus Security 2010.
4.3.4 Information system
The information system of Scavi is designed based on the concept of an ERP system (Enterprise
resource planning - ERP) is a business-management software that an organization can use to collect,
store, manage and centralize data from many business activities into one source of data flow (Sheilds,
2011). ERP system may include product planning, budgeting, production management, service
delivery, marketing and sales, inventory management, logistic, transaction, and payment. This ERP is
considered as the best IT tool to support the supply chain management.
In Scavi, the core of the ERP system is IScala, and it implemented in 2006 (Figure 18). However, Scavi
purchased only two function of Scala: Finance and Logistic (Sale order, purchase and inventory

50 | P a g e
control). Later on, because of the need of management, others function has been developed and
integrated into IScala:
 ScaX, an in-house developed app that can integrate with IScala, has been designed especially
for the commercial department.
 Human resource and payroll apps, developed by a third party, was using for the HR & finance
department.
 GPRO is used at the manufacturing factory for production management.
Figure 18. The Information System of Scavi
4.4 Outline the forecasting practice
4.4.1 Forecasting function in Scavi
The forecasting function of Scavi is located in the Commercial Branch of Commercial & Supply Chain
department (Figure 19).

51 | P a g e
Figure 19. Commercial Department of Scavi
In commercial and supply chain department, there is 3 vice-president of Commercial and 13
commercial teams. The 13 commercial teams are divided base on the current number of the customer
and its order’s volume. Business customer with the big order will be taken care by a whole commercial
team while customers with the small order will be gathered into a group of 2-4 customers and will be
assigned to an appropriate commercial team.
Each commercial team will have one Senior Commercial Manager and approximate 10 team members.
Each vice-president will manage 3-4 commercial team and will only work directly with the senior
commercial manager.
The responsibilities of Senor commercial manager and commercial team are:
 Managing commercial team.
 Driving the whole business process and supply chain from A-Z: contacting with customer,
pricing, taking customer order, issuing production order to the factory, calculating and
purchasing material, checking factory’s production schedule, quality control, delivery, and
payment.
 Representing Scavi to manage the communication relationship with the strategic customer.
 Forecast the demand of each commercial customer.

52 | P a g e
The responsibilities of the vice president of commercial includes:
 Taking charge of Strategic sales planning (SSP) and Materials requirement planning (MPS).
 Controlling and validating strategic issue, ensuring the coherence with Scavi’s strategy.
 Checking sale target assigned to the commercial team and support the commercial manager.
 Reviewing and evaluating the forecast demand that was produced by the commercial team.
 Summarizing and Forecasting the demand for the whole firm.
4.4.2 Forecasting process links directly to supply chain through ScaX & Scala
The business and supply chain process of Scavi is very complex. There are no formal document that
describes the forecasting process and forecast practice in Scavi (Appendix B). Therefore, it will require
more time of detailed observation to fully understand the whole process. However, in two months of
observation, the process of forecasting demand and taking order can be simply outlined in the figure
below:
Figure 20. Flow Chart of Demand Forecast and Taking Order process
The flow chart above demonstrates the first two step of forecasting demand and of taking order that
will become the trigger for the whole supply chain progress following behind (calculating and
purchasing material, checking factory’s production schedule, quality control, delivery, and payment).

53 | P a g e
In this flow chart, before issuing an offer to customer, the demand forecast must pass through 3 level
(Appendix. B):
 The first level will be performed by the commercial team as a group. The forecast must provide
these information: What is the product? What is the quality requirement? What is the
quantity of the order? When the customer will need this product? What is the cost of
production and the price that we can offer?
 The second level will be carried out individually by the vice-president of commercial. Then the
VP can have the right to make any adjustment if necessary, or to require the commercial team
to re-do the forecasting work.
 The last level will be decided by the customer. If the forecast demand is correct and the price
offer is reasonable; the customer will accept the offer and make the order. It means that the
demand forecast was right. However, a wrong demand forecast may lead to the situation
where the price is unreasonable, and where the order volume is too high (over-forecast) or
too low (under-forecast). In this case, the customer may decide to reject the offer. Then the
commercial team must report back to the vice-president and re-do the demand forecast.
After confirming the order, the commercial team issues a production order by using ScaX. ScaX is an
in-house application that the IT department of Scavi developed and designed especially for the
commercial team. ScaX can integrate the data into IScala. Nevertheless, it doesn't have any forecasting
function. ScaX’s main function is to act as an intermediate step to transfer customer’s order data into
IScala. Then, these data will be integrated with IScala and used with others apps (GPRO, HR & payroll)
in the ERP system to manage the supply chain process.
4.4.3 Forecasting methods of Scavi:
Surprisingly, the commercial department uses the judgmental method only in forecasting practice
(Appendix. A & B). There a five factors will be considered while producing a demand forecast:
 Order History data: for existence customer, historical data of order is an important source of
information. However, Scavi didn’t develop any statistical methods to using this data. The
using method is a naïve method that assumes that the demand for this year will be somehow
identical to last year. The demand for last year will act as the baseline for further adjustment
while considering the others factors.
 Current environment & situation: any changing in the environment that the team analyzed as
a significant changing factor will be considered in the forecast. For example: economy,
national policy, the international relationship between countries,…

54 | P a g e
 Relationship with the customer: maintains an excellent communication and relationship with
customer will allows the team to gather more useful information that can support the
forecasting process greatly. Some value information may include the client strategy, others
suppliers of the customers (competitors), the expectation level,…
 The number of the document provide by the customer: each customer will have a unique set
of document that can provide to its suppliers. These documents may include many different
pieces of information: the trend of fashion, the catalog of mode, the demand forecast data of
end consumer, the requirement of product’s quality, the production guide, material
expectation.... Because most manufacture firm’s service is “made-to-order”, therefore these
information is crucially necessary.
 Production ability: The team must also the balancing between customer demand and
production capacity of the factory. A mistake in taking an order without considering
production capability may cause serious consequences. For example: factory overload with
the order (when demand is greater than supply), or factory cannot perform 100% its
production capability (when supply is greater than demand).
Both vice president and the senior commercial manager will use this same approach of considering all
these factors while producing and reviewing the forecast. What makes the difference in the forecast
result are: the level of forecasting experience between vice-president and senior manager; and the
leadership and empowerment style of each vice-president to their assigned commercial team
(Appendix A & B).
4.5 Issue finding and Discussion
4.5.1 Statistical forecast methods are not useful in manufacture firm.
In general, the business nature of manufacture firm is B-2-B. The manufacturer must deal with a
customer with a mass unit order (units/order from hundreds to thousands, or even millions), not with
end consumer with low order (1-2 unit/order). In the case of Scavi, 90% of the annual revenue came
from commercial customer and wholesaler with a big order. Therefore, the main task of the
manufacturer is to focus mainly on forecast the demand of the business customer. In others words,
conducting qualitative forecast method and maintaining good relationship/communication with the
business customer is the most important task.
For manufacture firm, the number of business customer is low in number, for example Scavi has only
20 business customer in 2015 (Appendix. A).In this situation, the using of a quantitative or statistical
method is not relevant because the number of sample/data is too few.

55 | P a g e
The firm can save the cost of implementing a statistical method by taking advantage the information
of business customers. Manufacture company normally doesn't need to spend effort on the retailers
function and brand building to end consumer (Scavi don’t have retailer and brand building function).
Retailer and brand building is the problem of the customer, not the manufacturer. Therefore, it is not
necessary for the manufacturer to invest into a statistical method that needs a vast number of data
from downstream customer’s demand is a task. Instead, the firm should focus on the relationship
building with the client and improving information sharing process.
4.5.2 The support of Information technology to the demand forecast of manufacture firm
is limited.
Not using statistical methods also lead to the limitation of IT solution that can support forecast. As
mentioned in chapter 3 of this paper, the shift into using statistical method encourages the
development of IT in the forecast. Most of the new forecasting technology were developed to support
statistical methods (Charles W. Chase, 2013), not judgmental method. In the case of Scavi, there is
almost no mark of information technology in the forecasting process, except for the mention of ScaX
and IScala. However, these two apps are just a parts of the ERP information system, and they don’t
have any forecasting function. It serves as the intermediate system between the commercial team
and the supply chain/production team. In an organization where judgmental forecasting methods
were dominant like Scavi, the only strong point of information technology is that it support the
communication between the firm and its oversea customers. Clearly, in manufacture firm, the need
for information technology in supporting forecasting process is not significant.
There is another reason manufacture company decides not to invest into new forecasting technology:
The cost is too high. Although Scavi is planning to increase the budget for information up to an amount
of $200,000/years, these investment is focus mainly on the upgrade of the ERP system as the top
priority of the business (by replacing IScala), rather than the upgrade of the forecasting technology
(Appendix. C). Furthermore, there are some others solutions that are better, less expensive and
contents less risk than investment in a new forecasting technology (Appendix. C). For example: hiring
forecasting expert, outsourcing forecast process, or buying data and information from a market
research company.
4.5.3 The lacking of statistical practice and forecasting technology limits the growing
ability of the firm.
Although investing in a statistical forecast method and new forecasting technology is not necessary
for manufacturer, it will also restrict the development of the company in some aspects:

56 | P a g e
If firm wants to expand the business into the retailer field. For example: Scavi had experienced a
massive failure while trying to explore the domestic market by introducing a new retailer brand-name.
The firm forecast that the demand of the market will be 1,000,000 units of clothes each year. But in
fact, it costs Scavi 2 years to sale-off all 1,000,000 units with a low price (Appendix A). The reason for
this failure is the inexperience in conduct the demand forecast of end consumer as well as the
habitude of using judgmental forecast.
It limits the firm’s ability to sense and discover new business opportunity. In the era of electronic
information, data is a greats assets for any business. And the firm that depends too much on the data
provided by the customer will have a limited source of information. This information will only provide
a “local view” of the environment and market around the customer. The firm who don’t invest in new
forecasting technology and statistical practice will hardly achieve a bigger (global) view of the whole
market and from the, discover new business opportunity. For example in the case of Scavi, exploring
the potential market and finding new business customer are two of the biggest challenges of the
commercial department (Appendix A).

57 | P a g e
Chapter 5: Conclusion
5.1 Conclusion
This paper finds out two critical issues regarding the demand forecasting practice of manufacture firm
in developing country. Firstly, in contrast to many academic researches and business reports,
manufacture firm’s forecasting practice focus mainly in the using of judgmental method rather than
statistical methods. Secondly, the using of new information technology to support forecasting practice
in manufacture is not significant.
There is reason why manufacture firm prefers using judgmental methods rather than statistical
methods: low number of historical data. In the manufacturing industry, the main form of business is
“Business to Business” and “make-to-order” where the most of the revenue is coming from business
customers (commercial customer, wholesaler customer and fashion firm) with a big batch of
production units (from thousands to millions of units) in each order. In reality, a manufacturing firm
only has a limited number of business customers. Hence, it makes the using of statistical and
quantitative methods ineffective because the number of customer’s historical data are few in volume
and sample.
In reality, the relationship between manufacturer and business customer is the key factor that affects
the quality of the forecast. A good communication and relationship with the customer will
substantially support the gathering of valuable data and information for the judgmental forecast. This
information may include: the long-term strategy of the customer; the expectation of quality, price and
delivery service; the current situation of the market; the current financial situation; the information
of others supplier/competitors; the opportunity of the market… Furthermore, by taking advantage of
the information provided by the customer, manufacturer can also save the cost and effort on the
investment into forecasting system, retailer system, distributed system and brand development.
Little use of statistical methods limits the using of information technology in demand forecast. Most
of the newest forecasting technologies was developed to support statistical forecasting methods. The
era of internet and electronic data provide the firm the opportunity of mining the massive volume of
data all around the world without time-delay. The development of smartphone, social network,
internet of thing and big data are the great example of these technologies. It allows the firm to track
down and gather the data directly from the downstream customers demand. After that, the company
may use some of the newest forecasting software, take full advantage of all these valuable data, and
produce the final forecast result. However, these forecasting technology cannot support the
manufacturing company that focuses mainly on the using of judgmental practice. Furthermore, the

58 | P a g e
high cost of expense and the high failure risk are also one of the main reasons why manufacture firm
doesn't want to invest in these new forecasting technology project.
However, the lacking of statistical practice and investment in new forecasting technology may affect
manufacture in the far future. Firstly, it reduces the chance of expanding the firm's business into the
retail field because the firm doesn't own any detail data and information about the end consumer.
And secondly, it reduces the ability to sense and explore new business opportunity because firm only
has a limited source of information from the current customer and don't own any detail data and
information about the global market.
5.2 Research limitations
The first limitation of this research is the using of only one case study. Each organization will have a
unique characteristic. The research finding in this paper may not be able to apply to all companies in
the manufacturing industry. Therefore, while using this case as a base for further research, the reader
is suggested to compare the difference between Scavi and the other companies within the
manufacturing industry carefully.
The second limitation of this study is the limited time in the observation process. A period of two
months of observation cannot guarantee a full and in-depth understanding of the forecasting process.
Furthermore, the observation target may react differently when compared with normal daily
activities. Hence, if possible, observation research should be conducted over a longer period.
The last limitation is the bias opinion of the interviewee while conducting the interview. Although the
interview has been carried using the semi-structure approach to encourage the interviewee to share
the information in the most open way, there may remain a bias opinion. All three interviewee are of
the top manager of Scavi. Some sensitive information that may create a bad impression on the
company top management team will not be shared. Hence, some under surface issues may not be
found.

59 | P a g e
Reference
Ackoff Russell L. (1989) From Data to Wisdom. Journal of Applies Systems Analysis.
Armstrong J. S. (1989) Combining Forecasts: The End of the Beginning or the Beginning of the
End. International Journal of Forecasting 5.
Barker C. E., Croft M. R., Green A. T. & Long A. F. (1987) Information technology in developing
countries. Health Policy and Planning, 2(3), pg.251-254. Oxford University Press.
Bruce Dehning, Glenn M. Pfeiffer & Vernon J. Richardson (2006) Analyst’ forecast and
investment in information technology. International Journal of Accounting Information
Systems 7.
Chaman L. Jain (2006) Benchmarking Background of Forecasters. The Journal of Business
Forecasting.
Chaman L. Jain (2006) Benchmarking forecasting process. The Journal of Business Forecasting.
Chaman L. Jain (2006) Benchmarking Forecasting models. The Journal of Business Forecasting.
Chaman L. Jain (2007) Benchmarking Forecasting errors. The Journal of Business Forecasting.
Chaman L. Jain (2007) Benchmarking Forecasting Software and System. The Journal of
Business Forecasting
Chaman L. Jain (2008) Benchmarking Forecasting process. The Journal of Business Forecasting.
Charles W. Chase, Jr. (2013) Using Big Data to Enhance Demand-Driven Forecasting and
Planning. Journal of Business Forecasting.
Charles W. Chase, Jr (2014) Innovation in Business Forecasting: Predictive Analytics. Journal
of Business Forecasting.
Charlotte Brown (2011) Rationality and Foolishness: Alternative Forecasting Systems in a
Manufacturing Firm. Researching the Future in Information Systems.
Clemen R. T. (1989) Combining Forecasts: A Review and Annotated Bibliography. International
Journal of Forecasting 5.

60 | P a g e
Colin Robinson (1970) The forecasting system. The Role of Forecasting in Corporate Planning.
Cox & Wyman Ltd. London.
C. W. J. Granger (1980) Forecasting in Business and Economics. Academic Press Inc. London.
David Ashton & Leslie Simister (1970) The Role of Forecasting in corporate planning. Staple
Press Ltd. London.
David Montgomery (2006) Flashpoints for changing your forecasting process. The Journal of
Business Forecasting.
Dreischmeier R., Lawecki P., Deutscher S. A. & Arcuri A. (2014) The CIO’s Choice@ Adapt or
Fail. The Boston Consulting Group Inc.
E. J. Davis (1970) Sales Forecasting. The Role of Forecasting in Corporate Planning. Cox &
Wyman Ltd. London.
Fildes R. & Goodwin P. (2008) Forecasting. Strategic Direction, 24(5), pg.42-43.
Fisher, M. L. (1997). What is the right supply chain for product? Harvard Business Review,
75(2), pg.105-116.
Friedman David (2006) No Light at the End of the Tunnel. Los Angeles Times.
Fumi A., Pepe A., Scarabotti L. & Schiraldi M. M. (2013) Fourier Analysis for Demand
Forecasting in a Fashion Company. International Journal of Engineering Business
Management, 5. Special Issue on Innovations in Fashion Industry.
G. Clark Thompson (1947) Forecasting Sales, a Conference Board Report. Studies in Business
Policy. No.25. National Industrial Conference Board Inc. New York.
Issa G. F., Hussain S. M. & Al-Bahadili H. (2009) Economic Efficiency Analysis for Information
Technology in Developing Countries. Journal of Computer Science, 5(10), pg.751-759.
Jaana Auramo et al. (2008) The role of information technology in supply chain management.
Department of Industrial Engineering and Management. Helsinki University of Technology.
Finland.
James Morell (1970) Environment forecasting. The Role of Forecasting in Corporate Planning.
Cox & Wyman Ltd. London.

61 | P a g e
John E. Hanke & Dean W. Wicherm (2005) Business Forecasting. 8th ed. Pearson Education,
Inc. New Jersey.
John H. Vanston (2003) Better Forecasts, Better Plans, Better Results.
J. Scott Armstrong (1985) Long-range forecasting from Crystal ball to computer. John Wiley &
Sons. New York.
Kotlik L., Greiser C. & Brocca M. (2015) Making Big Data Work: Supply Chain Management.
The Boston Consulting Group Inc.
Leslie Simister (1970) Techniques: An Introduction. The Role of Forecasting in Corporate
Planning. Cox & Wyman Ltd. London.
Lim J. & O’Connor M. (1995) Judgmental adjustment of initial forecasts – its effectiveness and
biases. Journal of Behavioral Decision Making 8.
Luna D. R., Mayan J. C., Garcia M. J. Almerares A. A., Househ M. (2014) Challenges and
Potential Solutions for Big Data Implementations in Developing Countries. IMIA Yearbook of
Medical Informatics.
Lydia Strong (1956) Management Review. American Management Association
Mahoud E. (1989) Combining Forecasts: Some Managerial Issues. International Journal of
Forecasting 5.
Makridakis S. (1986) The Art and Science of Forecasting. International Journal of Forecasting.
Matteo Kalchschmidt (2008) The Impact of Forecasting on Manufacturing Performances.
Marcus O’ Connor (1993) Judgmental forecasting in times of change. International Journal of
Forecasting 9.
Marcus O Connor, William Remus & Kenneth Griggs (2000) Does updating judgmental
forecasts improve forecast accuracy. International Journal of Forecast 16.
Maurice Zinki (1970) Corporate Planning and Forecasting. The Role of Forecasting in
Corporate Planning. Cox & Wyman Ltd. London.
Michael E. Porter (1985) Competitive Advantage. Free Press.

62 | P a g e
Michael Lawrence & Marcus O’ Connor (1992) Exploring judgmental forecasting. International
Journal of Forecasting 8.
Michael Lawrence, Marcus O’ Connor & Bob Edmundson (2000) A field study of sales
forecasting accuracy and processes. European Journal of Operational Research 122.
Michael Lawrence, Paul Godwin, Marcus O’ Connor & Dilek Onkal (2006) Judgmental
forecasting: a review of progress over the last 25 years. International Journal of Forecasting
22.
Myerholtz B. & Caffrey H. (2014) Demand Forecasting: The Key to Better Supply-Chain
Performance. The Boston Consulting Group Inc.
Nam Quoc Tran & Van Ai Huyen (2008) Apply New Process into Manufacturing Stage to Adapt
With Expanding Plans of Scavi. College of Business Administration for Managers-VCCI.
Nam Quoc Tran (2014) Information System: “AS-IS” & “TO-BE”. Scavi Joint Stock.
Nam Quoc Tran (2015) SCAVI Organization & Overall Information System. Scavi Joint Stock.
Nenni M. E., Giustiniano L. & Pirolo L. (2013) Demand Forecasting in Fashion Industry: A
Review. International Journal of Engineering Business Management, 5. Special Issue on
Innovations in Fashion Industry.
Parente F. J. & J. K. Anderson-Parente (1987) Delphi Inquiry Systems. Judgmental Forecasting.
John Wiley & Son. New York.
Paul Godwin (2000) Correct of Combine? Mechanically integrating judgmental forecasts with
statistical methods. International Journal of Forecasting 16.
Paul Goodwin (2000) Improving the voluntary integration of statistical forecasts and
judgment. International Journal of Forecasting 16.
Paul Godwin (2002) Integrate management judgmental and statistical methods to improve
short-term forecasts. The International Journal of Management Science 30.
Right 90 Inc. (2010) Answering the Sales Forecasting Challenge for Manufacturers.

63 | P a g e
Robert Fildes, Paul Goodwin, Michael Lawrence & Konstantinos Nikolopoulos (2009) Effective
forecasting and judgmental adjustments: an empirical evaluation and strategies for
improvement in supply chain planning. International Journal of Forecasting 25.
Robert Fildes, Paul Goodwin & Michael Lawrence (2006) The design features of forecasting
support systems and their effectiveness. Decision Support Systems 42.
Rowe G. & G. Write (1999) The Delphi Technique as a Forecasting Tool: Issues and Analysis.
International Journal of Forecasting 15.
R. S. Reichard (1966) Practical Techniques of Sales Forecasting. McGraw Hill.
Sanders Nada R. (1997) The status of forecasting in manufacturing firms. Production &
Inventory Management Journal. American Production & Inventory Control Society.
Scavi Joint Stock Company (2013) Consolidated Financial Statements for the year ended 31
December 2013.
Sheilds Mureell G. (2001) E-Business and ERP: Rapid Implementation and project planning.
John Wiley and Sons Inc.
Todd Dewett & Gareth R. Jones (2001) The role of information technology in the organization:
a review, model and assessment. Journal of Management.
Tony Hines (2013) Supply chain strategies: demand driven and customer focused. 2nd Ed.
Routledge.
William Remus, Marcus O’ Connor & Kenneth Griggs (1998) The impact of information of
unknown correctness on the judgmental forecasting process. International Journal of
Forecasting 14.

64 | P a g e
Appendix A
Interviewee: Ms. Dai Thi Xuan Nguyen
Position: Vice-president of Commercial
Note: Because of the semi-structure nature of the interview, this part below is only the summary of the
main key point that is related to the research, not the full detail of the interview. The content of this
Appendix has been reviewed and accepted by the interviewee.
Question: Can you give us some overview of the Commercial Department?
The Commercial department are lining directly with the supply chain. Our department not only focus
on sales and market research, we also get involve in the supply chain process. In fact, except the
production function in the factory, we take care all of the others task in the supply chain management
such as: tracking production, production schedule, raw material buying, delivery and payment.
We have 3 vice president of commercial and 13 commercial teams with approximate 150 staffs to take
care of around 20 business customers and some others small client. Each forecast team has two main
task: Development and Purchase & Production. The development activities includes: research the
market, forecast the demand of the customer, prepare offer and take order, develop new product,
choosing of material. While the Purchase & Production is mainly involve with the factory and take care
of the supply chain management.
Question: What is your responsibility in the forecasting process?
Normally, each vice president like me will manage and take charge of around 3 to 4 commercial team.
We will cover all the tasks in the commercial department if necessary. But there are some task that
only the vice president can do:
 Report directly to the CEO and the Board of Director
 Keep contact and organize meeting with the top manager of important customer.
 Contribute to the strategy development of the whole firm.
 Ensure the Commercial department follow the firm strategy strictly.
 Ensure sale target.
 Working directly with senior commercial manager and review the forecast demand report
from the commercial team.

65 | P a g e
Question: What is the demand forecast method in Scavi? Statistical, Judgmental or both?
In Scavi, although we do have a data record of customer’s history order, but I think that we mainly use
judgmental method in our demand forecast. To produce a demand forecast we consider 4 factors:
 Historical data: We use historical data as the base of the forecast only. For forecast the
demand of each customer, we will take the historical order of the last year, take its number
and use it at the base for further adjustment.
 Environment: We will consider some overview information of the current environment to
predict whether the demand will of the client will increase or decrease for the next year. It
included many factors: the economy, the competitors, the current price of material, the
business situation of the customer, tax,…
 The relationship with the top manager of the client. For me, this is the most crucial factor. The
“B-2-B” business, relationship, communication and trust are crucial. It profile us some valuable
information that will affect our business greatly. For ex: The long-term strategy of the
customer, the commitment with our business, the information of its other suppliers (our
competitors).
 The document provided by the customer. The more document we have from the customer
the better. These information may include: the trend of fashion, the catalog of mode, the
demand forecast data of end consumer, the requirement of product’s quality, the production
guide, material expectation.... Because our main service is “made-to-order”, therefore these
information is necessary
For the practice, we have some regulation in forecast:
 Fashion industry is hard to predict therefore and an error is about +/-5% or +/- 10% (depend
on the experience with the customer) is acceptable for us. Loyalty customer will have a low
error ratio while new customer will have a high one.
 The sale target demand forecast will be divided into 3 classes: Sure Target, Reachable Target
(target A), and Unsure Target (target B).
 VP of commercial is not the one who directly produce the forecast. The one who directly
produce the demand forecast of the customer is the commercial team. They did it in group
meeting and the senior commercial manager will decide the final version. Then he/she will
bring this forecast to me to have the last review. I have right to accept, adjust or even reject
the result and ask the team to re-do the work.

66 | P a g e
 Each VP will not intercept with the work and commercial team of each other’s. We will have
a meeting just to summarize the overall demand forecast of the whole group, decide the sales
strategy and report directly to the CEO and Board of Director.
Question: Do Scavi have any document to guideline the forecast process?
Unfortunately, at the moment, we don’t have any formal document that describes our demand
forecast process. We also don’t have any document for demand forecast training. Everything is based
on our working traditional and the knowledge are passing down to the new employee by direct
communication, direct observation and direct practice. However, as I known, we are currently have a
plan to upgrade our information system and the IT department did mentioned about the necessary of
developing these document. I am currently in the process of create a developing team for this task.
Question: Have you experience any business failure because of inaccurate demand forecast?
I did experience a huge failure in 2014 because of the inaccurate demand forecast. In 2013, we had a
plan to expand our business in to the domestic market in Viet Nam by introducing a new fashion brand:
MARA. We forecast that in the end of 2014, we can achieve the sales target of 1,000,000 units of
clothes. All the production has been finish and we already have 1,000,000 products in our inventory.
But at the end of 2014, we can sell only 300,000 units. And at the moment, after spending a lot of
effort, we just finish selling off all 1,000,000 units. But with a cheap price.
The reason of the failure are:
 Wrong demand forecast. Our information and data of the domestic end consumer is are very
limited. The forecast are produced by our two sale representative in North and South of Viet
Nam. Both of them forecast the demand base only on their judgment of the market and their
personal relationship with some domestic customer. There is no statistical methods has been
conducted.
 We don’t have any expert in field of retailer, distribution and brand building. And if you look
into the Scavi structure, we don’t have such department.
Question: Are you satisfy with the current forecast practice in Scavi?
I am somehow satisfy with the current forecast practice. However, when looking back into our failure,
it sure that we still lack a lot of thing. These lacking in forecast practice make a huge difficult to us. We
cannot expanding our business and finding new customer. In the future, maybe we will consider some
addition options to deal with these problem. For example: hire forecaster expert or outsource our
demand forecast to a market research company.

67 | P a g e
Appendix B
Interviewee: Ms. Chau Thi Hong Nguyen.
Position: Senior Commercial Manager.
Question: What is your responsibility in the forecasting process?
My role as a Senior Commercial Manager included:
 Internal role: Drives the supply chain process from A to Z: contact customer, forecast the
demand, set up the target, produce the offer, set up production plan, material purchase plan,
delivery plan, quality control,payment method… everything except the direct production of
the product.
 External role: represents the only contact for the customer that has been assigned to my team
 Forecasting role: after contact customer and gather information for the forecast. I will sit
down with the whole team to discuss and produce the demand forecast of our clients. The
content of the forecast includes: what product? The quality requirement? The quantity of
demand? When the product is needed? What is the price’s expectation?. And from that, we
will set up a sale target and offer.
 The forecast must pass 3 levels:
o First level: I will be the one who have the final review of our team forecast before
submit it to the VP of commercial.
o Second level: The VP will review. She/he will have the final adjustment (if necessary)
before we begin to set up a business offer to the customer. But if the forecast is too
bad, our team will have to re-do the forecast.
o Last level: The customer. This is just an indirect level to check wherever our demand
forecast of the customer is. If the customer accept the offer, it means that our
forecasts are correct. If the customer reject the offer, it means our forecasts are
wrong.

68 | P a g e
Question: How you can produce a demand forecast? what is the method of traning in the
Commercial team?
Basically I will provide to you the same information as my Vice-president. However, I would like to add
another factor that we must consider when making a forecast: Production capability of the factories.
We want to achieve a right balance between or order’s quantity and the production capacity. Any gap
between these demand and supply is a bad new. If the customer’s order is too few, we cannot cover
the operation cost (fix cost). But if the production cannot satisfy the order, it will cause negative
consequence: delay in delivery, delay in payment, trust’s damage, and customer dissatisfaction.
Question: It is that true that Scavi doesn't use any statistical forecast methods? How you can reduce
the error of the forecast?
Yes, we don’t use any statistical method in forecast. We did used historical data, but the method using
is naïve and we use it as the star-base only. I think that the only way we can control the error of the
forecast is the experience of the senior commercial team and the vice-president of commercial. The
more experience you have about the customer, the market, the environment, the fewer error you will
have. For me, the relationship with the customer is the most important factor of our forecast practice.
Question: Do you use any information tool or software to support the forecast practice?
I think no, we don’t use any. We did use MS Excel and ScaX. But MS Excel is just for the documentation
only: there are no statistical methods has been developed in Excel. While ScaX is just an app which we
use to upload the customer order to our information system IScala. This data will be used for the
production management and tracking. I hear that there is some forecasting software that oversea
company use. But in Scavi, we have none of that type.
Question: what are the main challenges of the commercial team?
I think maintain relationship with customer is our biggest challenge. In the manufacturing industry,
mutual trust is everything. If the customer distrust you, they can easily find another suppliers to
replace. This is the rigor of competition in the manufacturing industry.

69 | P a g e
Appendix C
Interviewee: Mr. Nam Quoc Tran
Position: Head of IT Department
Question: Can you give us an overview of the IT department?
There are many thing to say about technical stuff. So to safe time, i will email to you some useful
document that can provide a good overview of the current IT’s level in Scavi.
Files:
 Nam Quoc Tran & Van Ai Huyen (2008) Apply New Process into Manufacturing Stage
to Adapt With Expanding Plans of Scavi.
 Nam Quoc Tran (2014) Information System: “AS-IS” & “TO-BE”.
 Nam Quoc Tran (2015) SCAVI Organization & Overall Information System.
Question: Did you know anything about the forecasting practice in Scavi? And did the IT department
involve in the forecasting process?
I did known about the forecasting process in Scavi. Currently, the business of Scavi is developing too
fast. Thus, we has a plan to upgrade our ERP system in the future to maximize the performance of our
supply chain management. Forecasting process is part of our supply chain management. Therefore, to
prepare for our ERP project, understand the whole process of supply chain, including forecasting
process, is a part of my job.
However, unfortunately, the IT department didn’t much support the forecast practice. The only thing
we get involve in are ScaX, an in-house app that we developed for the use of commercial department
only. But ScaX don’t have any forecasting function.
Question: What are the current challenges of the IT department?
The Board of Director had invested a lot in IT since 2006. Each year, Scavi spend approximate
$70,000/year for the IT department. The Board is also in the process of consider an increase in the
total budget up to $200,000/year in order to expand our IT infrastructure. This budget will be used

70 | P a g e
mainly for the development of the ERP system. It will helps us solve our top priority challenge of supply
chain management.
Question: Do you have any idea how IT department can support the forecasting process?
Now as you ask, i has never think seriously about this issue. But I can think of some solution:
 Develop a website/app to conduct survey to customer and consumer.
 Develop a mobile app like ebay, but specialize in fashion product to collect the data.
 Develop a Business Intelligent system.
But I think that these solutions are hardly implement in Scavi, because the cost is too high. Thus, I
don’t quite sure that the commercial team will need and use it. Furthermore, there are some others
non-IT options that are better in term of Cost-Benefit like hiring a market research company, or hiring
a forecaster expert.

11035624-Dissertation-MsC Information Technology (Final)

More Related Content

What's hot (20)

Viewers also liked (6)

Similar to 11035624-Dissertation-MsC Information Technology (Final) (20)

11035624-Dissertation-MsC Information Technology (Final)