EFFECTS OF MULTIPLE ADJUSTMENTS IN SUPPLY CHAIN FORECASTING ON FORECAST ACCURACY Banusha Aruchunarasa 208025X Degree of Master of Science Department of Transport and Logistics Management University of Moratuwa Sri Lanka December 2021 EFFECTS OF MULTIPLE ADJUSTMENTS IN SUPPLY CHAIN FORECASTING ON FORECAST ACCURACY Banusha Aruchunarasa 208025X Thesis/Dissertation submitted in partial fulfillment of the requirements for the degree of Master of Science in Supply Chain Optimization Department of Transport and Logistics Management University of Moratuwa Sri Lanka December 2021 i DECLARATION OF ORIGINALITY I declare that this is my own work, and this thesis/dissertation does not incorporate without acknowledgement any material previously submitted for a degree or diploma in any other University or institute of higher learning and to the best of my knowledge and belief, it does not contain any material previously published or written by another person except where the acknowledgement is made in the text. Also, I hereby grant to the University of Moratuwa the non-exclusive right to reproduce and distribute my thesis/dissertation, in whole or in part in print, electronic or other medium. I retain the right to use this content in whole or part in future works (such as articles or books). Signature: Date: 19/03/2022 ii STATEMENT OF THE SUPERVISOR The above candidate has carried out research for the Degree of Master of Science under my supervision. Name of the supervisor: Dr. H.N. Perera Signature of the Supervisor: Date: 19th March 2022 iii ABSTRACT Behavioral supply chain management is a subdiscipline within behavioral operations management that is growing rapidly. Judgmental adjustments of forecasts are considered part of this domain given the salience of forecasts to the smooth functioning of a supply chain. System-generated forecasts are frequently modified in the industry by forecasting professionals for numerous purposes. Accurate forecasts are significant to supply chain management and efficient organizational planning. Multiple adjustments occur when forecasts are subjected to more than one adjusted in its life cycle. Multiple adjustments are one of the key forecasting issues which impact forecast accuracy. Despite this, multiple adjustments to forecasts remain a not well-addressed research gap in academia. There are very few preliminary studies that investigate multiple adjustments to forecasts. Thus, to investigate the effect of multiple adjustments to forecasts to enhance forecast accuracy in the SC, the researcher employed a laboratory experiment with four different treatments to measure the forecasters’ behavior specifically on multiple adjustments to forecasts. 194 undergraduate and MBA students were recruited as participants for the experiment. In the Control Group, forecasts with first adjustments were observed while other treatments investigate how the participants would perform when they do subsequent adjustments with different levels of information availability. The authors found that multiple adjustments to forecasts significantly improve forecast accuracy. This expands the knowledge of multiple adjustments to forecasts to industry and academic professionals. Moreover, the provision of relevant information related to the previous adjustment allows the forecasters to perform better. The authors suggest the industries to increase information visibility among supply chain partners to have accurate forecasts and subsequent results in supply chain optimization. The results emphasize the importance of industry exposure and understanding the practical situations for a forecaster to improve his/her decision-making regarding judgmental adjustments. This study stresses the supply chain management-related degree programs to provide industry exposure to students to understand the practical implications of forecasting and other supply chain issues. Further works in this avenue, such as developing a forecasting model by integrating multiple adjustments and investigating the impact of the black-box effect in multiple adjustments are encouraged. Keywords: Forecasting, Judgmental adjustments, Multiple Adjustments, Laboratory Experiment, Behavioral Supply Chain Management iv ACKNOWLEDGEMENT In the first place, I intend to convey my heartfelt gratitude to Dr. Niles Perera, my post- graduate research supervisor. I have been privileged to get supervised by a supervisor like him who directed me from the very start of my MSc to the end. Further, I am thankful for his knowledge, motivation, guidance, and commitment throughout the degree program. I am sincerely grateful to my external advisor Prof. Dilek Onkal of Northumbria University, United Kingdom for her guidance, motivation, and support in the methodology development of my research study. I intend to express my gratitude to the post-graduate research coordinator of the Department of Transport and Logistics Management Dr. T. Sivakumar for his guidance and support for the progress reviews at each step of the degree program. My profound gratitude should go to Senior Prof. Amal S. Kumarage, Former Head and Founder of the Department of Transport and Logistics Management of University of Moratuwa, and the current Head of the Department, Prof. A.A.D.A.J. Perera for providing me the opportunity to follow the degree program. Further, I would like to thank Dr. A.I.T. Gamage and Mr. H.H.H.R. Chamara for their support and guidance throughout the experimental design and data collection for the research study. I am also thankful to the undergraduate and post-graduate students of the Department of Transport and Logistics Management and SLIIT Academy for their participation in the experiment. I wish to appreciatively acknowledge the Senate Research Grant No. SRC/LT/2020/20 of the University of Moratuwa, Sri Lanka for funding my research. I would also like to thank all the academic and non-academic staff of the Department of Transport and Logistics Management for their support. Finally, I am truly grateful to my family, friends, and team members of the Center for Supply Chain, Operations and Logistics Optimization (SCOLO) for their advice, encouragement, and emotional support throughout the research study. v TABLE OF CONTENTS DECLARATION OF ORIGINALITY i STATEMENT OF THE SUPERVISOR ii ABSTRACT iii ACKNOWLEDGEMENT iv TABLE OF CONTENTS v LIST OF FIGURES vii LIST OF TABLES viii LIST OF EQUATIONS viii LIST OF ABBREVIATIONS ix 1. INTRODUCTION 1 2. LITERATURE REVIEW 4 2.1. Behavioral operations in Supply Chain Management 4 2.2. Forecasting 5 2.2.1. Judgmental Forecasting 7 2.2.2. Human adjustments in Forecasting 8 2.2.3. Multiple adjustments in Forecasting 10 2.3. Research Hypothesis Development 11 3 METHODOLOGY 13 3.1 Research Approaches 13 3.2 Laboratory experiment 14 3.3 Attributes of Laboratory experiments 15 3.3.1 Effective experimental design 15 3.3.2 Context 17 3.3.3 Subject pool 17 3.3.4 Incentives 19 3.3.5 Infrastructure and logistics 19 3.4 Experimental Design 19 3.5 Mechanics of the experiment 21 3.5.1 Task description 22 3.5.2 Basic Information 23 3.5.3 Plan 23 3.5.4 Feedback panel 27 3.6 Method and Procedure of experiment 28 3.7 Course credit selection for treatments 30 vi 3.7.1 Random assignment of treatments 30 3.8 Sample 32 4 DATA ANALYSIS 34 4.1 Forecast accuracy improvement and multiple adjustments 35 4.2 Impact of Information sharing in multiple adjustments 39 4.2.1 Forecast Value Addition 40 4.3 Impact of Industry exposure in forecast accuracy 42 4.4 Size and Direction of the adjustments 46 4.5 Feedback Analysis 47 4.5.1 Prior knowledge on multiple adjustments 47 4.5.2 Importance placed on system-generated forecast 48 4.5.3 Importance placed on the forecast with the First adjustment 49 4.5.4 Importance placed on historical information 50 4.5.5 Level of trust on the system 51 5 DISCUSSION 52 5.1 Multiple adjustments and forecast accuracy 52 5.2 Forecast accuracy improvement through information sharing 53 5.3 Impact of industry exposure on forecast accuracy of multiple adjustments 53 5.4 Additional findings of the study 54 6 CONCLUSION 57 REFERENCES 60 vii LIST OF FIGURES Figure 2-1: Year-wise distribution of research papers 4 Figure 3-1 - Task description panel 22 Figure 3-2 - Cover story of the experiment 22 Figure 3-3 - Basic information panel 23 Figure 3-4 - Countdown clock in plan panel 24 Figure 3-5 - Historical information - Control Group 24 Figure 3-6 - Plan tab (Control Group) 26 Figure 3-7 - Plan tab (Treatment 1) 27 Figure 3-8 - Plan tab (Treatment 2) 27 Figure 3-9 - Plan tab (Treatment 3) 27 Figure 3-10 - Feedback panel 28 Figure 3-11 - Sequence of the experiment 29 Figure 3-12 - Landing page of fence link 31 Figure 3-13 - Redirection to treatment 31 Figure 3-14 - Random assignment of treatment 31 Figure 3-15 - Gender 32 Figure 3-16 - Subject group 32 Figure 3-17 - Age distribution 33 Figure 4-1 - Sample size of treatments 34 Figure 4-2 - QQ normal plots - H1 36 Figure 4-3 - Violin plot of MAPE by Treatment 38 Figure 4-4 - Violin Plot Groups 45 Figure 4-5 - Awareness of multiple adjustments 48 Figure 4-6 - Importance placed on the System-generated Forecast 48 Figure 4-7 - Importance placed on Forecast with the first adjustment 49 Figure 4-8 - Importance placed on Historical information 50 Figure 4-9 - Trust on System 51 viii LIST OF TABLES Table 3.1 - MAPE of each SKU in data set 21 Table 3.2 - Difference between treatments 25 Table 3.3 - Details of marks 30 Table 3.4 - Average marks of students 30 Table 4.1 - Summary of Observations 35 Table 4.2 - R output of Shapiro Wilk test 36 Table 4.3- Dunn (1964)’s test- H1 37 Table 4.4 - Dunn (1964)’s test H2 40 Table 4.5 - FVA comparison 41 Table 4.6 - Summary of Observations in Groups 42 Table 4.7 - Shapiro Wilk test - Groups 43 Table 4.8 - Bartlett test H3 44 Table 4.9 -R output of Kruskal-Wallis’s test H3 44 Table 4.10 - Summary of forecast adjustments 46 Table 6.1 - Findings of the Study 58 LIST OF EQUATIONS Equation 3-1 - Marks calculation 30 Equation 4-1 - MAPE calculation 34 Equation 4-2 - FVA calculation 41 Equation 4-3 - Forecast Adjustment 46 ix LIST OF ABBREVIATIONS ANOVA – Analysis of Variance BSCM – Behavioral Supply Chain Management BSC – Behavioral Supply Chain CG – Control Group FSS – Forecast Support System FBA – Forecast by Analogy FVA – Forecast Value Addition MAPE – Mean Absolute Percentage Error MBA – Master of Business Administration MRP – Material Resource Planning RBF – Rules Based Forecasting SC – Supply Chain SCM – Supply Chain Management SD – Standard Deviation SKU – Stock Keeping Unit SLIIT – Sri Lanka Institute of Information Technology SRC – Senate Research Committee UK – United Kingdom QQ – Quantile – Quantile plot