Forecasting in the South African Automotive Industry

(1)

Forecasting in the

South African Automotive Industry

‘A case study at Nissan South Africa’

©Bart Wichgers S1300792

(2)

Forecasting in the

South African Automotive Industry

‘A case study at Nissan South Africa’

Master Thesis on behalf of the Rijksuniversiteit Groningen; Faculty of

Management and Organization

The Author is responsible for the contents of this thesis;

the copyright of this thesis belongs to the author

Agency: University of Groningen

Supervisors:

University of Groningen: Drs. B. Yang (1st supervisor)

Dr. J.T. van der Vaart (2nd supervisor) Dr. M. Broekhuis (3rd supervisor)

Nissan South Africa: P. Sanders (Senior Manager sales planning)

R. De Vries (Director Marketing&Sales)

Date: August 2007

City/Country: Paris (France)

(3)

Abstract

This research has been concerned with the improvement and development of sales forecasting processes of Nissan South Africa. A few problems stated by the marketing and sales director have been the reason for this research:

 no clear view of future demand  high inventory levels

 no clear processes on planning and forecasting  no clear measurement of forecasting performance

Together with the sales planning manager and the M&S director, the following research questions have been defined:

Which are the current forecasting methods and processes at Nissan South Africa, and which improvements can be made to be able to produce a more accurate and measurable demand forecast for Nissan South Africa and the South African market?

The research can be divided in three main parts:  Pre-research

 Diagnosis and Analysis

 Conclusion and recommendations

The pre-research gave a clear view on NSA and its current processes. It made clear that NSA did not had a clear forecasting process as stated by the marketing and sales director. NSA was not aware of the importance of forecasting and the benefits of it. Thereby NSA could not distinguish forecasting from planning.

For the diagnosis and analysis a framework of ‘World Class Forecasting’ is used as a benchmark (Moon 2004). This framework is made up out of many researches from over the past 30 years . It combines empirical and theoretical researches into one framework for analyzing forecasting processes for improvement. The WCF is built up from four main elements; Functional integration, performance measurement, systems and approach. The WCF framework is used while NSA did not had a clear view of their own processes and NSA was searching for overall forecasting process improvements. According to the elements of the WCF a diagnosis and analysis has been made. The main conclusion is that forecasting is a process which should be supported by the whole organization, which starts by having the process formalized and creating awareness throughout the company. Once this is completed, it is important to think about steps to improve the forecasting process content on approach, functional integration, systems and performance measurements on an ongoing basis. Finally, the focus should be on the continuous learning aspect of forecasting.

The main recommendations of this research are listed below:

 Prioritise forecasting throughout the company by developing a forecast strategy and goals  Create a clear demand forecast process with roles and responsibilities

 Set up a forecast performance measurement and set acceptable bandwidths  Provide training on forecasting

 Employ skilled people for forecasting specific

(4)

 Set up a track and trace workbook e.g. to explain forecasts and events/occasions that happened each period

 Use dealer forecasts and outstanding orders into forecasting  Create a rolling forecast

The main contributions of this research to NSA are:

 Implementation and development of a new sales forecasting process, including roles and responsibility chart and performance measurement

 Creating awareness on the importance of forecasting throughout NSA by giving workshops and organizing brainstorm meetings on forecasting

 Initial steps on creating a technical forecast model for NSA by practising several forecast techniques on the FY 2006 MTP, BP and Carflow/RSP

(5)

Preface

This report is based on the research I have conducted at the Marketing and Sales department of Nissan South Africa. As a graduation thesis, this is the final piece of my study in Business Administration at the University of Groningen in the Netherlands.

I would like to thank my supervisors from Nissan South Africa, Roel de Vries and Paul Sanders, who gave me all access to the ´Nissan world´ and supported me during my intern period in South Africa.

Also I would like to thank Fatima, Ingrid, Spike and all other colleagues involved during my research within the organization of Nissan.

I would like to thank my supervisors at the University, especially Dr. Yang. His patience with me and his willingness to help out were necessary were outstanding. Especially during the final stage of the project.

Last but not least, I would like to mention I had an extraordinary great time in South Africa, with Nissan, my colleagues, the students I lived together and the wonderful South African people I have met during my stay. Many good memories and friendships will remain. Paris

August 2007 Bart Wichgers S1300792

(6)

Chapter 1 Introduction and initial management question

This introduction chapter will present an overview of Nissan, its environment and demand forecasting in the South African automotive industry. The chapter will start with an overview of the South African automotive industry, followed by demand forecasting. After this Nissan and Nissan South Africa (the case study company) will be introduced, together with an overview of forecasting at Nissan South Africa. The chapter will finish with the initial management question and the research objective.

1.1 Demand forecasting and the Automotive Industry

The automotive industry is a broadly and in-depth researched industry and highly developed. The topic of demand forecasting in the automotive industry is often researched, however, not in the South Africa. A lot of research is done on demand planning, order fulfilment and lead-times, and these outcomes brought this world-wide industry to its current level. Out of these researches appeared that, at least for the short future, production and delivery lead-times cannot decrease any further (Holweg 2000; Meyr 2004). This emphasises the importance of demand forecasting. Nowadays customers want their vehicles at the moment they are in the showroom, which implicates that no matter what lead-time there is, it will always be too long for the customer. Beside the customer requirement of having their vehicles in time, a proper demand forecasting process will also help to reduce costs in the manufacturers supply chain (figure 1.1). The more accurate these forecast are, the less inventory needs to be kept on all the products (Kahn 2003). Another issue is the, non-automotive specific, role of demand forecasting in decision making. Decisions based on demand forecast are made by all stakeholders, ranging from dealers/competitors to governmental stakeholders (figure 1.2). All long term decisions are made on forecasted demand, which is the input for operational, tactical and strategic company plans. Therefore these forecasts need to be as accurate as possible, reflecting the true potential market out in the field.

3. Lost 'Sales' Costs

4. Lost Companion Product Sales 5. Reduced Customer Satisfaction 5. Reduced Margin

1. Order Expediting Costs 2. Higher Product Costs

Cost of Forecast Error

Costs of Over-Forecasting Costs of Under-forecasting

1. Excess Inventory 2. Inventory Holding Costs 3. Transshipment Costs 4. Obsolescence

(9)

Importance of Marketing/Demand forecasts

Market, segment&model forecasts

Actions by Key-decision makers

* Competitors

* Supliers, Distributors * Goverment

* Company action(Marketing Mix)

Costs

Financial and other outcomes Actual Sales

Actual Market Share

Figure 1.2 The importance of demand forecasting

1.2 Nissan Japan(NML)

Nissan Motor Company Japan was officially founded in 1933, Yokohama Japan. Today Nissan has over a 186,336 employees world-wide (Sustainability report 2007) and has sold and manufactured nearly 3,5 million cars (FY 2006). Nissans’ corporate vision is ‘Enriching peoples lives’, and the Nissan global mission is ‘Providing unique and innovative automotive products that deliver superior measurable values to all stakeholders in alliance with Renault’.

Structure of the organisation

With the Nissan Global Headquarters in Japan Nissan is divided in (international) regional activities and functional activities, as indicated in figure 2.

(10)

Tokyo Nisan Motor Sales Co., Ltd.

Nissan Prince Tokyo Sales Co., Ltd.

Dealers in Japan 5 Aichi Nissan Motor Co., Ltd. Parts Manufacturers

9 Aichi Machine Industry Co. Ltd. JATCO Ltd.

Calsonic Kansei Corporation Sales Finance Companies

14 Nissan Financial Service Co., Ltd. 15 Nissan Motor Acceptance Corporation 16 NRF Mexico, S.A. de CV

12 Nissan Motor Iberica, S.A. 11 Nissan Motor Manufacturing (UK) Ltd.

7 Nissan Shatai Co., Ltd. 8 Nissan Diesel Motor Co., Ltd.

Nissan South East Asia Co., Ltd. Etc

6 Nissan Canada, Inc. 10 Nissan Mexicana, S.A. de CV

Dong Feng Motor Co., Ltd.

13 Nissan Motor Company South Africa (Pty) Ltd.

2 Nissan North America, Inc. 3 Nissan Europe S.A.S. 4 Nissan Asia Pacific Pte, Ltd. Nissan Middle East FZE

Partners Renault S.A.

Headquarters/Regional Headquarters/ Regional Companies

Vehicle Manufacturers&Distributors/

Distributors Vehicle Manufacturers 1 Nissan Motor Co., Ltd.

1 2 3 4

Nissan Motor Co. Ltd. Nissan N.A. Inc. Nissan Europe S.A.S. GOM

14 15, 16 7, 8 ,9 10 11, 12 13 5 6 Functional Activities Regional Activities Japan Management Committee General Overseas Market Management Committee North America Management Committee Europe Managment Committee Sales/Marketing Product Planning Technology/R&D Manufacturing Purchasing Finance/Treasury Human Resources Corporate Support Sales Finance Global Nissan Headquarters

Figure 2 Nissan Global Organisational Structure(Source; Derived from Nissan Sustainability Report 2005)

There are 4 regions distinguished; Japan, North America, Europe and the General Overseas Markets (GOM). Each region is responsible to work out the Global strategy, mission and goals within their region. The goals for each region are fixed in regional Business Plans. Within each region each Nissan-daughter is responsible to reach their targets as agreed upon in their individual Business Plans.

(11)

1.3 Nissan South Africa(NSA)

Nissan South Africa sold his first cars in South Africa in the early 1960s. It was owned by Automakers, in which Nissan was a minority shareholder until it acquired a 37% share from the Samlan Group in 2000 enabling it to take over the company. In 2001 Automakers changed its name to Nissan South Africa, becoming a subsidiary of Nissan Motor Company.

Organisational structure

NSA is organised in 9 functional departments. Besides manufacturing, the main focus process of NSA is on marketing&sales. The M&S department is sub-divided into several other departments and functions (figure 3). Marketing is split up into functions belonging to the NSA model line-up, sales regions and sales channels. The forecasting&research department is responsible for the forecasting and planning activities.

The manufacturing site is located in Rosslyn, where NSA produces 3 of his models. The manufacturing site has three lines; two single model lines and one multi-model line. Besides the three Nissan models, NSA also produces a Fiat model on the multi-model line. All the other Nissan models are imported from overseas.

M a n a g in g D ire c to r M a n u fa c tu rin g E x p o rts C o rp o r a te Q u a lity A s s u ra n c e C o r p o ra te P la n n in g C o r p o r a te F in a n c e H u m a n R e s o u r c e s S u p p ly C h a in M a n a g e m e n t A fte rs a le s

P r o d u c t E n g in e e rin g C o r p o ra te A ffa ir s & C o m m u n ic a tio n

P u rc h a s in g M a r k e tin g & S a le s

Administration Officer Cilla Trexler

Sales Pro Baurbette McDonald Ast Brand Manager

Valid Value Hannli Strydom

Regional Retail Specialist Tersia Human Regional Retail Specialist

Deidre Morris Regional Retail Specialist

Prischina Pillay Incentive Management

TBA Retail Marketing Manager

Peter Webb Market Analyst Multon Chueu Market Analyst Fatima Vally Research coordinator TBA Research Manager Ingrid Olivier Planning & Analysis Manager

Paul Sanders

Product Analyst TBA Product Analyst Lentswe Modingwane Product Planning Manager

Wilhelm Baard

Ast Product Manager Vinod Thomas Product Manager

Hennie Kotze Commercial vehicles

Ast Prod Manager Phiwi Khonelwayo Product Manager Palesa Gule Navara, Patrol Product Manager Nigel Kapswara Pathfinder, X-Trail Marketing Manager SUV

and LCVs TBA Product Manager Lara Gunning Tiida Product Manager Nadia Trimmel (Maternity leave) Product Manager TBA Ast Prod Manager

Jenifer Mokoma Marketing Manager

Passenger Odette du Preez Almera, Murano, 350Z, Micra

Internet Communication Desire Read Brand Support Manager

Glyn Demmer Events & SponsorshipsPenelope Jeffrey GM Marketing & Planning

Gerhard Fourie

(12)

NSA mission and goals

MINDSET ACTIONS

VISION

Nissan: Enriching People’s Lives

NISSAN WAY The power comes from inside

Focus is the customer Driving force is value creation

Measurement is Profit MISSION

Nissan provides unique and innovative automotive products and services that deliver superior

measurable values to all stakeholders in alliance with Renault.

Figure 4 Nissan Corporate mission and vision (www.nissan.co.za)

NSA has set up its own mission and goals to reach its corporate mission and goals (figure 4). Its goal is ‘to be present in all market segments and gain a sustainable share in each segment’.

 NSA vision is to be a ‘Top Level Customer Orientated Company’.  This vision stands on three pillars;

1. Quality; Provide top quality as perceived by the customer

2. Cost; Global cost and competitiveness local & global benchmark supported by local content

(13)

NSA Products

NSA distinguishes two types of vehicles, the locally assembled CKD models and the imported CBU models. In total NSA is present in 17 segments with 18 models and more than 150 model derivatives (see Table 1).

Market Competing Segment NSA Model

Passenger B Micra

Passenger C

Almera, Tiida Passenger Sports 350Z Passenger Small SUV X-Trail Passenger Medium SUV Pathfinder Passenger Large SUV Patrol Passenger Cross Over SUV Murano LCV 0.5 ton PU 1400 bakkie LCV >1 ton PU Patrol LCV SC Hi-line QW LCV SC Workhorse QW LCV Extended Cab QW LCV Double Cab QW, Navara LCV Panel Van Primastar MCV Panel Van High Roof Interstar MCV

Panel Van Standard

Roof Interstar

MCV 1t-2t FC CC Cabstar

Table 1 NSA Model Line Up 2005-2006

(LCV= Light Commercial Vehicle, MCV= Medium Commercial Vehicle) NSA Production

NSA currently produces three Complete Knocked Down (CKD) models at its Rosslyn plant, namely the Nissan Almera passenger car (substituted since February 2006 by Tiida), the 1400 bakkie and the QW Hardbody pick up ranges. In 2004 NSA announced that it had achieved 65% local content in its vehicles. The Rosslyn plant exports cars to Europe, Singapore, Australia and New Zealand.

Besides the three locally produced CKD models, NSA completes its line up with Completely Build Up (CBU) import vehicles from various Nissan plants around the world.

The production lead-times at NSA are around 2-3 months for locally produced models and 2-4 months for imported CBU models. Problems occur however that end customers do not know three months in advance what vehicle they want to buy. The dealers do not have that kind of information either. Therefore NSA needs either forecast their demand or keep an extra safety stock on all models.

NSA Sales

NSA has four main sales channels, which are dealers, government, rental and singles. The dealer channel has the highest profitability. NSA has his own franchise licensed dealerships, which are

(14)

categorised in small, medium and large dealerships based on sales volumes. It has also exports sales, which fall out of this research scope.

Each channel has also his own characteristics. Dealerships mostly order low volumes frequently. Government and rental sales orders are less frequent, but have relative high volumes. The single channel is low volume sales to the own NSA fleet for employees and promotional vehicles. Overview of the Market

The South African Automotive Industry is divided in 4 general markets (Figure 5); passenger vehicles (PV), Light Commercial Vehicles (LVC), Medium Commercial Vehicles (MCV) and Commercial Vehicles over 8.5 ton (CV Over 8.5t). This split is based on the weight of the vehicles. Nissan is only present in the PV, LCV and MCV markets, therefore in this thesis these markets are considered to be the total market.

Figure 5 Ideal planning & forecast levels of NSA

Each general market is further subdivided in segments, which can be further split up into bodyshapes. In total the South African automotive market has 24 segments (including two segments defined as ‘others’).1 Which vehicles model belongs to what segment raises lot of discussion among the OEMs, and therefore is reviewed by the NAAMSA statistics committee annually. Each Original Equipment Manufacturer (OEM) can use its own segmentation for internal use. However, all OEMs who join NAAMSA are committed to report their sales according to agreement upon NAAMSA segmentation. The NAAMSA distinguishes five main sales channels, namely dealer, government, rental, single and exports.

Beside OEMs who join the NAAMSA, there is also a significant group of OEMs who will not report to NAAMSA. They are referred as the non-reporters, and have there own organisation AMH. Although they will not report according to NAAMSA standards, their sales volumes are available. TIV Markets Segments Model(groups) Model(derivatives)

(15)

NSA Historical Sales 1995-2005 11,4% 8,7% 6,9% 8,9% 8,7% 8,1% 7,5% 8,7% 9,2% 9,2% 7,9% 0 5.000 10.000 15.000 20.000 25.000 30.000 35.000 40.000 45.000 50.000 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Y e a r 0,0% 2,0% 4,0% 6,0% 8,0% 10,0% 12,0%

NSA Passenger NSA LCV NSA TIV

Figure 6 NSA Historical Sales & TIV market share 1995-20052

Nissan is the 5th best selling brand in the South African market with 7.9% of the total automotive sales in 2005. This is a decline of levels achieved in the 1990s. After Nissan’s take-over of Automotors NSA market share deteriorated considerably since 2004, when the market share was still at 9.2% (figure 6). Recently NSA cannot keep up with the market growth (see table 2), and it has been unable to follow Toyota’s leadership of the overall market and VW’s lead in the passenger car segment.

On the commercial vehicle market, however, NSA is still strong with 15.4% of the LCV market only next to Toyota and General Motors, and with 4.8% of the PV market share in 2005.

NSA Growth % TIV Growth % NSA Growth % TIV Growth % 1996 -20,1% 4,3% 2001 -0,3% 7,5% 1997 -25,8% -6,5% 2002 10,6% -4,9% 1998 10,4% -14,4% 2003 10,4% 4,8% 1999 -8,8% -5,7% 2004 21,8% 22,0% 2000 8,0% 15,3% 2005 7,8% 25,7%

Table 2 NSA volume growth versus SA Total Industry Volume(TIV) growth 1996-2005

(16)

1.4 Forecasting at NSA; an introduction

Forecasting at NSA means the estimation of the demand for (new) cars over a certain time period (Schönsleben 2004; Interviews NSA 2005). The new vehicle demand is determined on model derivative level in units, whereby market share is also taken into account. The time period to be forecasted is not very clear on paper. It varies between monthly forecast, the Retails Sales Plan (RSP), till a 5 year ahead forecast, the Mid-Term Plan (MTP). Beside the RSP and MTP, NSA also forecasts a 1 year Business Plan (BP) and the 1 year Carflow/RSP version 1, which is a draft from the BP.

Within the current forecasting process three time horizons can roughly be distinguished;

 The short term forecast over 1 to 5 months. This short period is relatively long due to the fact that NSA has to take into account long lead-times, which vary from 2-3 months for the CKD’s and 3-5 months for the CBU’s. Over these periods will be forecasted on model per derivative level per month (per Stock Keeping Unit). The short term forecast will be updated every month. From the monthly forecasts a production schedule will be derived for the CKD’s, which is divided in a weekly and daily planning. With the Carflow/RSP orders are placed for de CKD parts and CBU’s, whereby the different lead times will be taken into account.

 The medium term forecast covers all 12 months of the current financial year3_{. For the}

medium term NSA has two forecasts; the BP and the Carflow/RSP version 1. The difference between these two is that the BP is a commitment to the GOM region and NML Japan, and the Carflow/RSP version 1 is used as an operational forecast document. In reality this means that the forecasts in the BP are always very low and conservative to be able to achieve the commitment and related bonuses. The Carflow/RSP version 1 is the more realistic and higher forecast, which is updated every month with adjustments when necessary. Officially the Carflow/RSP version 1 will be updated quarterly, resulting in version 1 to 4. But in practice its seems to be necessary every month.

 The long term forecast covers a period up to 6 years, in which a forecast is issued on model level per month. This document is also called the Mid-term plan (MTP) which includes trends, political development, economical factors, environmental analysis and planned new model introductions from Nissan and their competitors.

1.4.1 NSA forecasting process

NSA has no officially forecasting process wherein the roles and responsibilities of all stakeholders are defined, although there is a figure in which all functions are represented. With interviews and observations I have mapped the current forecasting process at NSA.

Forecasting&research is primarily responsible for the forecasting process. Although, the real forecasting at the moment is done by the individual product managers. The product managers research and analyse the automotive market and the segments in which their model is positioned. The input they are using consist most of the time of historical sales data, new product (launches) information, and customer research. Besides this information they also use the input from the

(17)

Carflow/RSP version 1 to 4, which include the forecast for the whole year, and for the total market. Sales planning responsibility is to collect all individual product forecasts from the product managers, and process this into one document, the RSP. Forecasting&research analyses the stock levels and stock parameters per product and makes adjustments in the product forecasts when necessary. After this the RSP forecast will go to SCM who will look if adjustments are necessary due to production constraints and send the (adjusted) RSP back. The possibly adjusted RSP will be judged by F&R and (when agreed) issued after approval by corporate management and stakeholders. The RSP is connected to the whole sale plan, and the WSP is connected to the production plan (see figure 7).

RSP WSP Production Requirement Detailed Constraint Checking Dealer Stock NML Requirement

Demand Supply Match

Proposed Demand Supply Matching

Yard Stock

Feasible Production Plan drives Yard Stock

Supply Request

(WSP + Min Stock Requirement) drives Production Requirements Resulting Yard Stock cover evaluated against target.

Evaluate associated risk and opportunity. Production Requirement may be adjusted if Yard Stock cover deviation from target is deemed unacceptable.

Figure 7 Demand and Supply match at NSA ( source Nissan South Africa)

During the planning process many official and informal meetings take place. The official meetings are;

- Product review meeting (PM’s, channelheads, sales planning) - Carflow/RSP review meeting (all stakeholders in the FC process) - Carflow indent meeting (approval by all stakeholders)

Communication is of major importance in the forecasting process to get to an agreement of the final forecasts. Tightly connected with this are the arguments for the forecast numbers by all individual stakeholders (PM, channelheads, SP, SCM, corporate management).

At the moment there are no or hardly any profound arguments for the forecast numbers. There is a lot of communication going between stakeholders, but without any documentation. The same counts for the official meetings, there are no records kept of these meetings. This causes unclearness about the outcomes of these meetings.

(18)

1.4.2 How is forecast accuracy measured?

NSA is measuring many KPI’s and there is a lot of data available. Forecast accuracy is not measured at the moment, although there has been KPI’s set up in the past for demand forecast accuracy and data input accuracy.

NSA does look on a monthly basis to the variance between BP, Carflow/RSP and actual sales. This is recorded to monitor the extent to which NSA is on track according to BP and Carflow/RSP version 1. When there is a variance, and this is almost always, NSA will manage with the help of these numbers to adjust the next Carflows/RSP. None of the product managers or departments are getting feedback or confronted with the accuracy of their forecasts.

To get an indication of the issued problem situation and perhaps to detect struggles in certain segments and/or sales channels, the first step to be taken is to research how accurate the current demand forecast were in the past. Therefore the a KPI for demand forecast accuracy at NSA will be used.

1.4.3 Forecast accuracy results from NSA

Forecast accuracy can be measured at different aggregation levels over different time horizons. Figure 9 is showing the forecast accuracy over the monthly retail sales plan(RSP) measured against actual sales for both NSA CKD and CBU models (see appendix 3 for the forecast accuracy for CBU and CKD over FY 2004 and FY 2005).

Looking at the results and the graphs the following conclusions can be drawn;

 Overall high level numbers on forecast accuracy are not very useful. A distinction should be made between forecast to high and forecast to low.

 Measuring forecast accuracy over 1 month is useful, but it should especially be measured over the real forecast period of 2-3 to 3-5 months.

 There is a accuracy problem for NSA, regarding the many outliers.

 Outliers occur in every segment, which is a first indication for a structural forecast problem (Kahn 2003)

Forecast accuracy error % Carflow version 1-4 vs. actual 2005

-40,0% -20,0% 0,0% 20,0% 40,0% 60,0% 80,0% 100,0% 120,0%

Jan Feb Mrt Apr Mei Jun Jul Aug Sep Okt Nov Dec

Month A c c u ra c y e rr o r %

(19)

1.5 Initial Management question

NSA core process is manufacturing and selling vehicles. Demand forecasting and planning is a core process in this, especially due to the fact NSA is importing most of its model line up from overseas plants. The lead-time of these orders is between 2-4 months depending on the models. Local production lead-time of CKD’s is between 2-4 months. Roel de Vries, M&S managing director at NSA, stated that the current demand forecasting process and its performance measurements are not working the way they should be. While NSA is facing high stock levels on certain models, there is no clear view of future demand. There is also no clear view of how the current demand forecasting process should and or could perform due to missing a clear process and the lack of a clear performance measurement. Therefore NSA wants to know how its current demand forecasting process and measurement looks, and which improvements can be made to come to a more accurate, reliable and practical demand forecasting process.

Therefore, the following research objectives have been formulated together with Roel de Vries and Paul Sanders, senior manager of F&R.

‘Analyse and map the current forecasting procedures/processes for Nissan South Africa, and provide recommendations/solutions which will lead to an more accurate, reliable and practical forecasting model/method for the F&R department.’

1.6 Conclusion

In this chapter an overview is given of the general research environment and the initial research objectives are presented. Main elements of the research environment are the NSA supply chain, the South African automotive industry and NSA’s organisational structure. In this further research the focus will be mostly on NSA Marketing&Sales department, and the factors NSA is able to influence.

The main research frame roughly consists of three phases; first phase will be a pre research. Second phase is a diagnosis based on the pre research. The result of the diagnosis will be the input for a re design in phase three. The research will end with an overall conclusion and the first implementation results. The next chapter will present a more detailed research set up. First an intermezzo will follow to elaborate briefly on the turbulent South African (automotive) environment.

(20)

Intermezzo development of South Africa and their Automotive Industry

While South Africa and their Automotive Industry play a significant role in this Thesis, this intermezzo will briefly give an overview of both their latest and upcoming future developments. South Africa is a rapidly developing country, and is trying to get rid of it’s ‘third world’ status. The South African economical, political and cultural development received a boost after the release of Nelson Mandela and following on that the end of the Apartheid in 1994. South Africa has now a relative stable economy and a relatively stable democratic government. This attracts many foreign investors and raises the local business and consumer confidence levels.

The South African government started to restructure the whole country with several kind of social and economical programs. For the SA automotive industry the government agreed upon a new Motor Industry Development Program (MIDP) in 1995. This program was the follow up on several other programs which where not very successful, while the automotive industry was still low volume with high unit costs. The MIDP would end in 2002, but due of it’s overall success it is postponed to 2012.

The latest boom in new car sales started around 2003-2004, and brought new vehicles sales to amazing records which nobody had foreseen. The boom still continues. Main causes for this boom, and the general macro-factors which influence new car sales, are;

* Stable and strong Rand

* Growing Real Disposable Income * Lowest Prime Interest Rate ever * Low Inflation

* High Business and Consumer confidence * High Gross Domestic Product Growth

* Emerging and the Emerged Black Middle Class * Good Overall Infrastructure

* SA regulation; New Taxi Program * Many New Model Introductions

* New Car Buyer Cycle of 18 months (ABSA Bank)

The SA automotive industry is still developing and has much growth potential with the upcoming black middle class and also due of some big events coming up like the soccer World Cup 2010 and the rugby World Cup 2011 both hosted by SA. On the other side there are also some threats for South Africa and it’s Automotive Industry in particular, these are;

* Highest AIDS infection rate in the world * Shortage of high educated/skilled people * Local inflation

* High debt-to-income ratio’s for SA Households * Increase of new vehicles prices due of strong Rand

* Substantial increase of raw material prices (e.g. steel prices) * SA regulation; Change in Car Allowance Income Tax dispensation

* Upcoming trend of reduced trade-in values undermining new vehicles affordability

(21)

Chapter 2 Research Design

This chapter is mainly concerned with an in depth problem analysis of the background of the research and the initial management question. The performed problem analysis is first described and elaborated, as the justification for the further research. Then, the chapter focuses on the role of the demand forecasting process at NSA. This is followed by the theoretical framework wherein this research is performed. Finally, the structural concept of the diagnosis and the further structure of the whole research are discussed together with the presentation of the conceptual model.

2.1 Problem analysis

The first step of this research was performing a pre research annex problem analysis. The original initial management question was the input for the pre research at NSA. The main objective of the problem analysis was;

‘Gain insight in the connection of the mentioned problem with the underlying problems to formulate a relevant, goal efficient and researchable problem definition/research question’ The problem analysis should give answers to the following questions (De Leeuw 1996); 1 Who is using the demand forecast at NSA?

2. Which are the problems caused to them by missing a clear defined demand forecasting process and performance measurement?

These questions are answered by reviewing the relevant literature, existing NSA documents, observations and conducting interviews. The focus has been first on the research (sub) system (marketing&sales department), the inventarisation of the organisational problem (demand forecasting process and forecast performance measurement) and the connection between these two(functional integration).

Organisational (sub)system

Both the organisational system NSA and the sub system marketing&sales have been described. The function of forecasting is located in the forecasting&research department. Figure 9 presents other relevant sub systems.

(22)

Organisational Problem

The organisational problem was formulated by the director of M&S and the senior manager of F&R. NSA does not has a proper forecasting process, although F&R does provide a monthly RSP and a yearly BP. Thereby many people and departments (particularly senior management, SCM, product marketing and F&R.) have the feeling that the forecasts annex plans are lacking accuracy. This became clear after an SWOT analysis (see appendix 1). Other issues include the lack of a clear process, lack of clear roles and responsibilities, and thus unrealistic plans.

Functional Integration

The forecasts are used by many stakeholders (figure 9). All stakeholders use the forecast for their own purposes, which causes many islands within the organisation regarding forecasting. Main problems caused to the stakeholders are stress, confusion and disagreement about the forecasts and planning. For NSA not having a proper forecasting process is causing higher inventory cost, uncertainty in decision making and lower work satisfaction (NSA Interviews 2005). The main involvement of the stakeholders is listed below;

 Product Planning - creates MTP with own forecasts

 F&R (former Salesplanning department) - creates RSP with forecast input, creates forecasts

 Finance – calculates profitability based on forecast  SCM - creates carflow with RSP input

 Manufacturing- carflow leads production

 Corporate Planning - input for strategy / goalsetting  Senior Management - decision making

 Product Marketing - decision making, creates own forecast and gives it to F&R  Sales - target set/based partly on forecasts

 Dealer Development - input for development/dealer capacity requirements  Dealer - indirect by allocation on RSP

(23)

2.2 Problem Statement

2.2.1 Main Research Question

The themes and related elements of the theoretical framework and conceptual model are the input for the main research question. By describing and analysing NSA on these themes and elements a clear view can be presented regarding NSA’s current demand forecasting process and forecast accuracy. By comparing NSA demand forecasting process with the world class forecasting framework of Moon (2004) gaps can be determined for further improvements and or recommendations.

In the previous chapter the following research questions have been defined;

Which are the current forecasting methods and processes at Nissan South Africa, and which improvements can be made to be able to produce an more accurate and measurable demand forecast for Nissan South Africa and the South African market?

2.2.2 Research objective

The objective of this research is to identify problem areas in the current demand forecasting process at NSA, and what improvements can be made to solve these problem areas to achieve a well working demand forecasting process with satisfactory accuracy results. By exploring NSA as a whole, the in and outputs of , and the demand forecasting process itself a clear insight can be gained for improvements and making recommendations.

2.2.3 Research Delimitation

In conducting this research, certain delimitation’s are taken into account. The study is mostly based on the perspective of the case study company, Nissan South Africa. The decisions and recommendations made in this paper are based on this company-situation information. Nissan South Africa and South Africa are relatively short on (forecasting) skills. To minimalise these influences several external stakeholders are involved, like economic consultants, local universities and banks.

2.2.4 Research Outline

The outline of this research (figure 10) starts with an introduction to the case company and its operational environment, followed by a pre research and the research set up (Chapters 1-2). The next step is carried out a review of the literature relating to the thesis topic, demand forecasting in the automotive industry (chapter 3). Based on the outcome of this literature study a conceptual model, investigative questions and research method will be presented in chapters 4 and 5.

The framework for the diagnosis and analysis part of this research will be presented in chapter 6-9. After the diagnosis and analysis the recommendations will be presented further worked out in the redesign (Chapter 10). This research report will end with an overall conclusion and reflection on the research (Chapter 11).

(24)

Figure 10 Research Outline

Thesis set-up/Time Frame

The research has started in September 2005, and the estimated submission date will be the end of August 2007. The actual intern period lasted 9 months. This research was supposed to be submitted before September 2006. Due to several reasons this date could not be achieved, one of these being a job offer by Nissan Europe as a reward for my research and other work for Nissan South Africa related to this thesis.

Due to the pioneering element in this research, I have agreed with NSA to focus firstly on inventarisation of the current situation and come up with recommendations. When time and circumstances permitting, the research could be extended to problem solving and implementation of the recommendations.

September October November December January February March April May June July

1. Problem Statement 2. Detailed research set-up 3. Information collection 4. Analysis

5. Reporting

(25)

Chapter 3 Theoretical Framework

This chapter will present the theoretical framework of this research. Most of the used theories and articles will be elaborated in the relevant chapters they appear. A world class forecasting framework will be the input for the conceptual model, which will be used to describe, analyse and provide recommendations for the current situation at NSA (Moon 2004)). The framework itself will be explained in chapter 4.

Theoretical framework

In previous chapters demand forecasting at NSA is already described. This resulted in a need to map the demand forecasting process of NSA, and investigate the causes of the inaccurate forecasts.

The reason for this research was that NSA does not have a clear insight view in the forecasting process and the accuracy of it. NSA therefore thinks that improvement is possible regarding the forecast process and accuracy. From several interviews it appeared that there is still unclearness around the current forecasting process and who is actually forecasting. Beside this there are many stakeholders involved by the forecasting process. Hereby it can be questioned if the right people are involved and responsible in the right phase of the process. Therefore more information and insight is needed in the organisation of the demand forecasting process. Furthermore, it is not clear whether or not the right indicators and measurements are used. This should be relevant to the issue of why the demand forecasts are inaccurate. To address this, it is necessary to collect information on measuring of demand forecast accuracy.

In Section 2.3 demand forecasting at NSA has been introduced. Section 2.3 described briefly that forecasts are generated by processing several different type of data, and for processing these data different unclear methods are used. Question is which methods are used? And are the right methods used?

The usage of different types of data also raises the questions of what kind of data is used and if this is correct and relevant data? And how accurate is this data? Therefore there should also be more information be gained on forecasting methods, as well as forecasting input data and sources.

To summarise, it is relevant to gain insight with help of a literature study on the following subjects;

- Measuring of demand forecast accuracy - Organisation of demand forecasting

- Methods of demand forecasting

- Input data&sources of demand forecasting

Methods of demand forecasting will be reviewed in this chapter. Other relevant theoretical subjects are reviewed and/or discussed were necessary in the chapters itself.

(26)

Measurement of demand forecasting

Jain(2000) describes that there are numerous ways to measure forecast accuracy, but too often three of them are effectively used in a business environment;

1. Mean Absolute Percentage Error(MAPE) 2. Standard deviation

3. Behaviour of distortion/deviation

Methods of demand forecasting

In literature many forecast methods are described and in which way they can be used. All forecast methods can be roughly divided into 2 groups of methods; qualitative (judgmental) and quantitative forecasting methods. The quantitative methods can be further classified into time-series and causal methods. Finally all forecasting methods can be assigned to one of these three groups; time-series, causal or judgmental (Jain 2000; Makridakis et al.; 1978).

The time-series forecasting methods use data from the past and project these into the future. Most common methods in this group are the moving average, weighted average, exponential smoothing decomposition and the Box-Jenkins/Rule-Based method. One important assumption for those methods is that the future data will have the same pattern as the past data.

Causal methods are also based on historical data, but they focus on so called drivers, which are responsible for the creation and influence from these historical data. There are many causal methods, but the most well known and common used is the regression method. For using this causal method, there should always be a stable relationship between the driver(s) and the effect. Judgmental methods are based on expert opinions, or individual experts. They are appropriate especially when there’s none or insufficient historical data available such as in new product introductions. Most well known and common judgmental forecasting methods are intentions, conjoint analysis, judgmental bootstrapping and the Delphi method.

The accuracy of a demand forecasting method is strongly depending on the (right) choice of the method considering the forecast circumstances. According the theory this choice is depending on a few aspects as listed below (Jain 1995; Jain 2000).

1. Detail Level 2. Time Horizon 3. Accuracy 4. Costs

5. Availability of historical data 6. Product life cycle

7. Size of promotion

8. Size of the customer market

9. Possibility of product categorisation 10. Product seasonality

(27)

Chapter 4 Conceptual model and main research

question

This chapter will introduce the World Class Framework of forecasting (Moon 2004). It will integrate the presented themes from the theoretical framework into one conceptual model. The main research question and investigative questions will be based on this conceptual framework.

4.1 Conceptual model

The framework of world class forecasting presents four main themes, which are forecasting performance measurement, organisation of demand forecasting, input and data sources, and demand forecasting methods (Moon 2004). All these themes are represented in the conceptual model (Figure 11)

Figure 11 Conceptual Model

For the purpose of this research, the demand forecasting process will have the central focus. This is also the main area pointed out by the stakeholders in the problem analysis and a SWOT analysis (see appendix 1). To integrate the four themes and include other relevant elements the conceptual model is based on a framework of ‘World Class Forecasting (Moon 2004) and the input-output model (De Leeuw, 2000).

The input-output framework is applied to have a broader view around the demand forecasting process, hereby the demand forecasting process is regarded as a black box (De Leeuw, 2000). 4.1.1 Input

The input of the demand forecasting process are basically the demand and the demand forecast indicators. The input consist of the characteristics of the demand, expert knowledge and historical data. The historical data can be further sub divided into historical, causal, planning and alternative data.

4.1.2 Environment

The environment is placed at the top of the conceptual model and also influences the demand forecasting process. The environment is an influential factor of the demand forecasting process, but is not regarded as input due to its high level variables like population growths, social interest and political preference which cannot be influenced by NSA directly.

(28)

4.1.3 Demand forecasting process

The demand forecasting process is placed in the centre of the conceptual model. The content of the process box is based on the “World Class Forecasting” framework (Moon, 2004), which distinguishes four main themes; forecasting approach, forecasting systems, forecasting performance management and functional integration of forecasting. All themes are of equal importance and they should be in balance with each other.

4.1.3.1 Forecasting Approach

The forecasting approach is a major pillar in the demand forecasting process. The approach is shaping the internal forecast environment and therefore will also create boundaries in the demand forecasting process. Part of the approach is the forecasting method. The forecasting method is in many previous forecast studies regarded to be the forecasting process, together with the forecasting system (Moon 2004).

The forecast approach further contains the forecast orientation, conceptualisation of historical demand, differentiation of forecast entities, forecasting hierarchy, level forecast technique sophistication, the relationship with forecasting and planning.

4.1.3.2 Forecasting Systems

Forecasting systems are the facilitator of the demand forecasting process and can make the process much easier, more effective and efficient. Companies often think that a forecasting system itself will solve all their forecasting issues. This often seems to be a huge misperception (Moon 1998; Moon 1999).

The forecasting system will be described (and analysed) with the topics systems integration with other systems, handling of reporting, maintenance of historical data, systems handling of performance measurement and system infrastructure investments.

4.1.3.3 Functional Integration of Forecasting

The functional integration of forecasting has been well documented in the forecasting literature. One of the best examples is the formal Collaborative Forecasting and Replenishment systems (Schönsleben 2004). It is of major importance in a ‘World Class Forecasting’ system. The functional integration can be described and analysed with the three C’s; Communication, Co-ordination and Collaboration.

4.1.3.4 Forecasting Performance Measurement

Forecasting performance measurement closes the demand forecasting process, and is used to improve the process and learn from its outcomes. Its place in the demand forecasting process can be argued, while it is measuring the output of the process. The forecasting performance measurement forms the end of the process and is an input at the same time, therefore it is part of the process.

The forecasting performance measurement is described and analysed with discussing two topics; the metrics and the feedback information on forecasting performance.

4.1.4 Forecasting Output

The forecasting output is the outcome of the demand forecasting process and is positioned at the right in the conceptual model. Obviously the output is the demand forecast itself, but the demand

(29)

4.2 Research question

The themes and related elements of the theoretical framework and conceptual model form the input for the main research question. By describing and analyzing NSA on these themes and elements a clear view can be given on NSA’s current demand forecasting process and output accuracy. By comparing NSA demand forecasting process with a world class forecasting framework, gaps can be determined for further improvements and or recommendations.

In the previous chapter the following research question was defined;

Which are the current forecasting methods and processes at Nissan South Africa, and which improvements can be made to be able to produce an more accurate and measurable demand forecast for Nissan South Africa and the South African market?

4.3 Investigative Questions

To answer the above research questions, a literature study has been carried out. Specifically, it focuses on which factors are influencing the demand forecasting process. By diagnosing the situation at Nissan South Africa on these factors it is possible to detect opportunities for improvement and provide recommendations to achieve these.

To perform an efficient and effective diagnosis it is necessary to split the main research question into bounded and researchable investigative questions based on the literature study, conducted interviews and the conceptual model.

Input

1. To what extent is NSA using enough and relevant input data in the demand forecasting process?

Demand Forecasting Process

2. To what extent is NSA using a suitable forecasting approach?

2.1 What is NSA’s forecast orientation?

2.2 How does NSA conceptualise its historical demand?

2.3 Is NSA differentiating within its forecast entities?

2.4 Is there any level of forecasting hierarchy?

2.5 How sophisticated is the NSA forecast technique?

2.6 What is the relationship between forecasting and planning?

2.7 What is the level of training and documentation of the forecasting process?

3. To what extent is NSA using a proper forecasting system?

3.1 Is there an integration of corporate systems?

3.2 How are the forecasts reported?

3.3 How is the (historical) data stored?

3.4 Are there any infrastructure investments?

4. To which extent is NSA’s demand forecasting process functional integrated?

4.1 What communication is used by NSA?

4.2 What co-ordination is used by NSA?

4.3 What collaboration is there at NSA?

5. To what extent is NSA measuring and learning from forecasting performance?

5.1 What metrics are used?

(30)

Outputs

6. To what extent are the outputs of NSA’s demand forecasting process satisfying?

Recommendations

7. Which gaps in NSA forecasting process can be determined and need improvement? 8. How does NSA ideal demand forecasting process look?

(31)

Chapter 5 Methods and techniques

In this chapter the research methods and techniques adopted in this research will be discussed. This chapter will describe the research strategy, research type and method, the data collection process, and the interpretation and analysis of the research findings.

Research strategy

The strategy of this research is to perform the research in an efficient, effective and pleasant manner, by working in a co-operative and structured way to gain mutual benefits.

In the methodological justification the methods of executing this research are described in more detail. To do so, first the type of research is identified. Subsequently, documentation used, research methods and analysis methods are considered.

Research type

This thesis is a practical research project because it will generate data and new insights, which are useful for Nissan South Africa. Insight is generated in sales forecasting processes. Benchmarking against the‘ World Class Forecasting’ model from Moon (2004), recommendations are proposed for improvement at the sales forecasting/planning processes at Nissan South Africa.

Documentation

In this section the sources of information to collect data are identified. De Leeuw (1996) identifies six main sources to collect data. These are:

1. Documents (library, archives, and secondary sources) 2. Media (papers, magazines, radio, television)

3. Databases

4. Reality (the field). 5. Boosted reality

6. Experiences of researchers

In general four main sources of information are used in this thesis; documents, databases, media and the reality (the field).

 Documents refer to useful theoretical articles, books and reports.

1. Internal reports at NSA; Carflow planning documents, flash reports and the Hyperion user manual provided by NSA.

2. Academic articles, textbooks and industrial reports, which provide relevant theories and concepts, are gathered from the University Library of Groningen and the University of Pretoria.

 The electronic databases of NAAMSA and NSA are used to learn more about the organization, the planning activities, the country and the job roles.

Actual sales, marketing plans etceteras in a given period are stored in different databases of NSA and external agencies (e.g. Econometrix and MSA). In case of no database available, NSA management and external parties provided the information.  Media is another main source of information, more specific the Internet. With the help of

(32)

 Reality (the field) is used by visiting dealerships, external stakeholders, the NSA manufacturing plant, attending marketing&sales related meetings, conducting interviews with all stakeholders and travelling around the country.

Research Method

In the previous section the investigative questions are formulated. Input for all investigative questions are the literature study, interviews with stakeholders and observations. All investigative questions are related to the conceptual model, and will be answered by first describing the situation followed by a current state analysis based on the framework on ‘World Class Forecasting’ (Moon 2004) to determine the current state situation at NSA. The framework from Moon captured past and latest forecast research results into one theoretical best–practice model. Given the fact NSA is not having much forecasting processes in place, this framework will be a good and helpful analysis tool. The current state situation will be the input for the recommendations to achieve an improved demand forecasting process. All investigative questions are answered by performing a literature study, conducting interviews and observations.

Interpretation of research findings

Interpretation of the research findings, as well as the current state analysis, is a difficult issue in this research due to the high qualitative aspects and the situation at NSA. Thereby NSA did not do much on forecasting previously, and did not have a proper documentation as well on their planning annex forecasting processes. Therefore it is not possible to fully rely on the quantitative and theoretical information. By conducting interviews with both internal and external stakeholders at different organizational levels it is tried to capture this issue.

(33)

Chapter 6 Input for the Demand Forecasting Process

Out of the theoretical review appeared that historical data is an important source for the demand forecasting process. In this chapter 4 types of data will be reviewed on their sources and how it is used at NSA. Beside data, expert knowledge is also an important input source for demand forecasting and will be reviewed as well. First a general overview of the types of data in the demand forecasting and marketing planning process.

Figure 12 The use of (marketing) data during demand forecasting (Makridakis 2000)

6.1 Characteristics of NSA demand

 Variety of products (complexity and uncertainty of demand)

For mapping the variety of products it is important that products can be determined separated from their customers orders (Bertrand et al., 1998). The complexity of demand is about the influence of the product variety on NSA demand forecasting process. This is reflected by the conceptualisation of the Customer Order Decoupling Point (CDOP). The CDOP is the point from where the customer order becomes allocated to a specific unique customer (De Leeuw 2000). The CDOP for NSA normally lays at the dealerships and the NSA stockyard, while NSA is producing make to stock and sometimes allowing direct customer orders as well. The last orders will have a longer order-to-delivery-to-customer lead time.

For this research, we only look at new vehicles, which can be split up into various models, model derivatives and colours where the customer can choose from. There is also a distinction in lead time between CBU and CKD models. The huge product variety can make a demand forecasting process very difficult, while every product has its own demand characteristics. A solution to minimise this problem is to make product groups with general demand characteristics.

(34)

The variety of customers will influence the organisation of demand. A few big customers requires a different approach than many small ones in regard to information processing capacity and co-ordination (Bertrand et al., 1998). The former will require in general less information processing capacity and co-ordination.

NSA demand is received by four sales channels; dealers, government, rental companies and single/fleets sales. The dealer channel is the main sales channel. It is split up into 3 main regions (East, West and Central). NSA dealerships (144 in total) are categorised in small, medium and large dealerships. This categorisation is used for incentive purposes, not for forecasting or allocation planning (Interview NSA 2005).

 Predictability of demand (availability of information/data, time series analysis, background of the time series, forecast methods&forecasts, customer analysis, conclusion) >forecastability reduces uncertainty

Predictability of demand is the extent to which it is possible to make accurate predictions of demand. A high predictability equals a low complexity. A predictability of demand which seems to be high, but requires lot of efforts, equalling a high complexity. When the predictability seems to be low, the complexity is (paradoxically) low as well (De Leeuw 2000).

Historical Data

Historical data can be split up into NSA internal historical data and external historical data, as will be discussed later on.

NSA has access to an enormous amount of historical data regarding new car sales, going back over more than 10 years. The data can be accessed via two external databases, one online system called RGT-AutoStats and one desktop version called RGT-MSA. These databases are updated on a monthly basis. NSA also has an own database called Essbase, an integrated software used by SCM, Finance and F&R. NSA is also storing all the reports they produce with these data, like monthly sales reports on a marketing server.

The data is available at many different levels, ranging from market totals till manufacturers sales per model and derivative. There is even the possibility to request data on engine size, fuel type and (since January 2005) colour.

Beside historical data on new car sales, NSA has access to various other historical data like Customer Survey Index (CSI) data, Marketing data, economical and environmental data. Only the CSI and Marketing data is stored by NSA, and all the other data is stored at the providing resource organisations like Econometrix (economic consultant) and banks. Historical external data like economical and environmental data is bought in from specialised agencies like Econometrix and banks. NSA is receiving this daily, weekly and monthly in the format of electronical mailings and online access to reports.

Forecasting in the South African Automotive Industry

Forecasting in the

South African Automotive Industry

‘A case study at Nissan South Africa’

Forecasting in the

South African Automotive Industry

‘A case study at Nissan South Africa’

Master Thesis on behalf of the Rijksuniversiteit Groningen; Faculty of

Management and Organization

The Author is responsible for the contents of this thesis;

the copyright of this thesis belongs to the author

Abstract

Preface

Table of contents

Chapter 1 Introduction and initial management question

1.1

Demand forecasting and the Automotive Industry

1.2

Nissan Japan(NML)

1.3

Nissan South Africa(NSA)

1.4 Forecasting at NSA; an introduction

Demand Supply Match

1.5

Initial Management question

1.6

Conclusion

Chapter 2

Research Design

2.1

Problem analysis

2.2

Problem Statement

2.2.1 Main Research Question

2.2.2 Research objective

2.2.3 Research Delimitation

2.2.4 Research Outline

Chapter 3

Theoretical Framework

Theoretical framework

Chapter 4

Conceptual model and main research

question

4.1

Conceptual model

4.2

Research question

4.3

Investigative Questions

Chapter 5

Methods and techniques

Chapter 6

Input for the Demand Forecasting Process

6.1

Characteristics of NSA demand