The forecasting performance of Company X has been identified as the core problem of this research. Hence, the remaining part of this research will focus on improving the forecasting performance of Company X. In order to do so, we define the research design for solving this action problem in this section.
The description of the problem and the goal of this research according to the previous paragraph lead to the following main question:
RQ: How can Company X improve its liability by reducing (and avoiding further creation of) the excess inventory of materials and components at the EMSs?
To be able to answer this research question, the following sub questions will be answered by the approach mentioned per question.
SQ 1: What is the current situation of Company X regarding demand forecasting?
a) What are the current processes regarding demand forecasting at Company X?
b) What are possible methods for measuring the demand forecasting performance?
c) What is the current forecasting performance of Company X?
d) How can the effect of demand forecasting performance on excess inventory be determined?
e) What are the consequences of inaccurate forecasting for the inventory for Company X?
Sub question b and d are answered in chapter 3 by performing a literature review. Section 3.1 first provides some general information about demand forecasting. Section 3.2 gives answer to sub question b, by discussing several forecasting performance metrics. Hereby we make a distinction between metrics for measuring the accuracy and for measuring the bias. In section 3.3 sub question d is answered, by describing how the link between the demand forecasting performance and the inventory levels can be quantified. In chapter 4 the information acquired by the literature review is applied to Company X. This chapter starts with a description of the forecasting processes of the different business units in section 4.1. Hereby sub question a is answered. The processes are schematically represented using flowcharts. The different processes are compared and their main differences and similarities are discussed. The information for describing the forecasting processes is acquired by performing interviews with the employees of the different business units that are responsible for the forecasting task. These interviewees know the most about how the processes actually take place, because they are most closely involved. Three Company X business units are included in the scope of this research, so three interviews have been performed. For both the business unit Business Unit B and Business Unit A, two employees were interviewed to obtain a complete and reliable view of the process. For the business unit Business Unit C, only one employee has been interviewed, since their forecasting task is performed by only this employee.
In section 4.3 the forecasting performance of Company X is assessed by measuring both the accuracy and bias, whereby sub question c is answered. Prior to this, we perform an analysis after the demand patterns of the Company X products in order to judge how hard it is to forecast these products. This analysis and its results are presented in section 4.2. Section 4.3 contains a description of the forecasting performance analysis. The data used for this analysis are the actual and forecasted demand for the year 2018, derived
from Company X’ ERP-system. Earlier data about the forecasted demand is not available and at the time
this analysis was performed, information about the actual demand in 2019 was not known yet. We measured the current forecasting performance by using the most appropriate metrics found in literature.
In section 4.4 we attempted to describe the link between the forecasting performance and the excess inventory levels by applying the information found in literature. However, it turned out that due to the lack of available data, this analysis would not provide valuable outcomes, so is excluded from the research. Due to this, we have no clear answer to sub question e. This has no major consequences for the further research, since the results would not constitute an important input for the further course of the research. The added value of the analysis would be helping with creating awareness within the organization and different business units around the importance of accurate forecasts.
SQ 2: Which possible improvements regarding reducing (and avoiding further creation of) excess inventory exist for Company X?
a) What are possible methods for increasing the demand forecasting performance?
b) What can Company X do to improve their demand forecasting performance?
c) What are other possible improvements for Company X to reduce and avoid further creation of the
excess inventory at the EMSs?
Sub question a is answered by performing a literature review, which is elaborated in section 3.4. Sub question b and c are answered in chapter 5. We answer sub question b in section 5.1, by providing a road map to an improved forecasting performance. This roadmap is based in the information found in literature and the analyses performed after the demand patterns, the current forecasting processes and the current forecasting performance of Company X.
Sub question c is answered in section 5.2. This is divided in two parts: reactive and proactive improvements. The reactive improvements contain possibilities for reduction of the current excess inventory and by the proactive improvements Company X can avoid further creation of excess inventory. Sources of these suggested improvements are the methods found in literature and conversations with employees of Company X.
Literature review
With this chapter, phase 4 (solution generation) of the MPSM has been elaborated. First some general information about forecasting and its terminology is explained in section 3.1. Section 3.2 contains a description of forecasting performance measurements, whereby sub question 1.b is answered. Thereafter, in section 3.3, the effect of forecasting errors on the inventory levels are described. This section gives the answer to sub question 1.d. In section 3.4, we discuss different forecasting models, which provides the answer to sub question 2.a. Finally, in the section 3.5, the conclusions are drawn.