
AGENT-BASED MODELLING OF SOCIAL RISK AMPLIFICATION DURING PRODUCT CRISES Yun Liu Supervisors: Jerry Busby, Stephan Onggo Department of Management Science Management School Lancaster University Submitted in partial fulfilment of the requirements for the degree of Doctor of Philosophy January 2018 DECLARATION This thesis is my own work, and it has not been submitted for the award of a higher degree elsewhere. Yun Liu January 2018 ACKNOWLEDGEMENTS I would like to thank my supervisors, Dr. Jerry Busby and Dr. Stephan Onggo, for their tremendous support, patient guidance, and encouragement throughout my PhD study. I have been extremely lucky to have two supervisors who have serious and rigorous attitudes toward academic research, who were committed to meeting me weekly, who always gave very detailed comments on my work, and who constantly trained me to think logically and critically. Their incisive suggestions and insights helped me at each stage of my research. I really appreciate all their contributions of time and effort to make my PhD an amazing experience. My sincere gratitude goes to members of my review panel: Prof. John Boylan, Prof. Konstantinos Zografos, and Dr. Roger Brooks, from whom I received particularly helpful feedback. I would also like to thank my viva voce examiners, Dr. Edmund Chattoe-Brown and Dr. Roger Brooks, for their time, interest, and insightful questions. I am indebted to my friends who participated in the pre-test of my survey questionnaire and helped improve it. Thanks also go to those who made time to complete the survey online. Special mention goes to our PhD Programme Coordinator, Ms Gay Bentinck, who provided administrative support and was always ready to help. Finally, I would like to thank my parents for their love and support in all my pursuits. ABSTRACT Public response to risk is socially shaped in a way that often over- or under-estimates expert risk assessments. One of the main theoretical tools to examine public risk perception is the social amplification of risk framework (SARF). This framework proposes a mechanism through which risk responses arise from interactions among various social actors, but past empirical work has been mainly concerned with correlations between structural variables rather than the mechanism of amplification and the process over time by which it develops. And more importantly, there has been quite limited modelling of risk amplification to date. This study aims to discover a way of formalising social risk amplification, to find out what are the necessary assumptions for modelling risk amplification, and to work out what consequences this modelling would predict. It is an attempt to model collective response to risks that are significant at a societal level but which materialise in a distributed way across a population. The natural heterogeneity of individual risk perceivers, the emergence of behaviour through interactions of social actors, and the complex feedback loops linking risk perception with risk related behaviour point to using an agent-based model as a modelling medium. The study is developed in the context of product contamination scandals such as the recent cases in China of contaminated milk products. One of the important features of contamination crises is that product recall has become an increasingly inevitable part and is often a key element in risk communication during such crises. Yet recalls send ambiguous signals about the misconduct of the organization in question: they clearly indicate some kind of failure, and possibly negligence, in the product that are associated with a risk of significant harm; but they also suggest that the organization is concerned with consumers’ welfare. The model that was developed is based on the principle that risk perceivers have to assimilate risk through the risk beliefs of others, their direct experience of a risk, and communications about the risk from organizations (including their product recall decisions) and the media. And it is based on the principle that, as well as discovering the nature of a risk, risk perceivers also make judgments about wrongfulness (which Freudenburg called recreancy) – and this also shapes the strength of risk responses. The model is partially calibrated with a consumer survey carried out in the context of a Chinese milk contamination scandal that took place in recent years. Simulation results from the model show that public risk perception grows progressively toward an exogenous peak before it immediately decays, and that there is a relatively high residue of concern after the crisis is resolved. The objectivity of media coverage appears to be inversely related to risk amplification: a media that simply follows public opinion is associated more strongly with exaggerated risk perceptions than an objective one. A sensitivity analysis indicates that the initial conditions, objective risk level, duration of contamination, and variation of recreancy perception are the most significant influences on the degree of social amplification. This knowledge helps prioritize data collection for future research and identify important aspects that particularly require managerial attention. The main contribution of this study is to develop a process of modelling social risk amplification that consists of three steps of increasing contextualisation. The first step involves a basic model that captures social risk amplification as a general theory relative to all kinds of risk event. The second step contextualises this model specifically for product recall crises. It involves extracting agent decision rules from the literature on product recall, based on statistical associations found in empirical work on recall crises. And the third step contextualises the model for a specific population. It involves calibrating the relative importance of different information sources for the heterogeneous agent population using a survey of Chinese consumers responding to a milk contamination crisis. One important insight from the process of modelling risk amplification is that SARF is not sufficient for modelling particular crises. It seems essential that modelling of SARF should involve a clearly defined context in which risk responses arise. Keywords: social risk amplification; agent-based modelling; product recall; recreancy; media CONTENTS 1 INTRODUCTION................................................................................................................. 1 1.1 Research problem ............................................................................................................. 1 1.2 Theoretical foundation ..................................................................................................... 2 1.3 Research context .............................................................................................................. 3 1.4 Research objective............................................................................................................ 4 1.5 Layout of thesis ................................................................................................................ 5 2 LITERATURE SURVEY ..................................................................................................... 6 2.1 Theoretical background of risk amplification .................................................................. 6 2.2 Empirical evidence of risk amplification ......................................................................... 8 2.2.1 The actors in risk amplification ................................................................................. 9 2.2.2 Contributory effects ................................................................................................. 14 2.2.3 The different contexts in risk amplification ............................................................ 16 2.2.4 The different methodologies by which risk amplification has been studied ........... 19 2.3 Modelling of risk amplification...................................................................................... 22 2.4 Conclusions .................................................................................................................... 24 3 RESEARCH CONTEXT.................................................................................................... 27 3.1 2008 Chinese milk scandal ............................................................................................. 28 3.2 Nongfu Spring water event ............................................................................................ 31 3.3 Gutter oil scandal ........................................................................................................... 33 3.4 Summary ........................................................................................................................ 39 4 RESEARCH DESIGN ........................................................................................................ 41 4.1 Research questions and objectives ................................................................................. 41 4.2 Process and methods of modelling ................................................................................. 42 4.2.1 Choice of agent-based modelling ............................................................................ 42 4.2.2 Overall process of modelling ................................................................................... 46 4.2.3 Validation procedures .............................................................................................
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages231 Page
-
File Size-