Testing Choice Theory Using Discrete Choice Experiments in Swiss Energy Policy Matteo Mattmann ii Promoter: Prof. Dr. Roy Brouwer Co-Promoter: Dr. Ivana Logar Thesis committee: Prof. Dr. Wouter Botzen Prof. Dr. Michael Getzner Dr. Jürgen Meyerhoff Prof. Dr. Ståle Navrud Prof. Dr. Rolf Wüstenhagen Cover Design: Åsa Frölander ISBN: 978-3-906327-95-2 iii VRIJE UNIVERSITEIT Testing Choice Theory Using Discrete Choice Experiments in Swiss Energy Policy ACADEMISCH PROEFSCHRIFT ter verkrijging van de graad Doctor aan de Vrije Universiteit Amsterdam, op gezag van de rector magnificus prof.dr. V. Subramaniam, in het openbaar te verdedigen ten overstaan van de promotiecommissie van de Faculteit der Bètawetenschappen op dinsdag 17 oktober 2017 om 11.45 uur in de aula van de universiteit, De Boelelaan 1105 door Matteo Mattmann geboren te Zürich, Zwitserland iv promotor: prof.dr. Roy Brouwer copromotor: dr. Ivana Logar This PhD thesis was funded by the Swiss Federal Institute of Aquatic Science and Technology (Eawag), and is part of the Competence Center for Research in Energy, Society, and Transition (SCCER CREST). v “Durchaus studiert, mit heißem Bemühn. Da steh ich nun, ich armer Tor! Und bin so klug als wie zuvor;” J.W. von Goethe in Faust vii Summary Testing Choice Theory Using Discrete Choice Experiments in Swiss Energy Policy The "Swiss Energy Strategy 2050" proposes to phase-out nuclear power gen- eration and expand renewable sources of energy. Hydropower, an established source of energy in Switzerland, is expected to be one of the key renewables that will be further expanded. In this policy context, this PhD thesis aims to test axioms and assumptions underlying microeconomic choice theory by applying discrete choice experiments (DCE). A DCE is conducted among a representative sample of Swiss respondents and elicits their preferences for an expansion of hydropower. This dissertation contributes to the existing literature by examin- ing how public preferences for expanding hydropower production are linked to public perception of (avoiding) nuclear risks. To this end, hydropower as well as nuclear risks are included in the DCE. This thesis begins with a quantitative meta-analysis of the existing stated preference literature that estimates the non-market values of hydropower exter- nalities. The results of the meta-analysis are used as inputs in designing the DCE. The results of the meta-analysis suggest that deteriorations in vegetation, land- scape, and wildlife are valued negatively, while there is only limited evidence for a significant positive willingness-to-pay (WTP) for mitigating these negative externalities. The avoidance of greenhouse gas emissions proves to exert a sig- nificant positive influence on welfare estimates, but no significant impacts on aesthetic and recreational amenities can be detected. The meta-analysis also re- veals that no stated preference studies so far have considered the link between preferences for renewable sources of energy and nuclear risks. The data obtained from the DCE are used to answer this dissertation’s main research questions. These focus on the standard choice-theory assumptions of viii certain and known preferences and the axioms of continuity and monotonicity. Furthermore, the role of multiple reference points in the framework of prospect theory is investigated. More specifically, the common and idiosyncratic determinants of choice cer- tainty, consistency, and monotonicity are investigated. In contrast to the existing literature, these three concepts are analyzed simultaneously based on the same sample of respondents. The results show that there are significant differences between the choice behavior of certain and uncertain respondents as well as be- tween consistent and inconsistent respondents. Moreover, gender and choice- task complexity prove to be common predictors of choice certainty, consistency, and monotonicity. This thesis also investigates the standard economic axiom of continuous pref- erences in the context of attribute-non-attendance (ANA). A novel methodology to assess ANA is presented based on the monitoring of the respondents’ visual information acquisition process using mouse-tracking. No significant model im- provement is found when including such a visual measure of ANA compared with the standard approach based on stated ANA information. Nevertheless, choice models based on visual ANA result in a slight improvement over choice models that do not take ANA into account and over choice models that use in- ferred ANA information. Finally, the dependence of preferences on (multiple) reference points, a key assumption in prospect theory, is tested. Non-status quo related reference points, associated with comparative risks shown on risk ladders, are expected to affect parameter estimates and welfare measures for a change in hydropower and nu- clear risk. The study confirms the importance of multiple reference points, and shows that, besides the status quo, these other reference points also influence respondents’ choices and welfare measures in DCEs. The results of this thesis support the need for a holistic view on energy policy accounting for the direct and indirect externalities of alternative energy sources in both research and policy. ix Acknowledgements First and foremost I would like to thank my supervisors Ivana Logar and Roy Brouwer. It has been a very supportive, friendly, and uncomplicated collabora- tion with both of you. Your comments and ideas often perfectly complemented each other. Ivana, a special word of thanks to you for the infinite amount of time you have invested in giving me very valuable, extensive, and precise feedback on work in progress at various stages. Roy, special thanks to you for your enthu- siasm and for keeping the big picture in sight at times when no light at the end of the tunnel seemed visible. Thank you also for putting me up on various visits to Amsterdam and Waterloo in Canada, and for connecting me with the people in your respective teams. Many thanks also belong to the members of the reading committee for the review of this dissertation: Wouter Botzen, Michael Getzner, Jürgen Meyerhoff, Ståle Navrud, and Rolf Wüstenhagen. Special thanks are due to Mehmet Kut- luay, my connection to the Vrije Universiteit in Amsterdam and the team at IVM, for the motivating and valuable exchange. Thank you Noémie Neverre for the excellent French translation of my survey. I would also like to thank conference attendees in Nancy, Zurich, Cork, and Athens, as well as the organizers and members of the Competence Center for Research in Energy, Society and Transi- tion (SCCER CREST), of which this thesis is part of. This research project was funded by the Swiss Federal Institute of Aquatic Science and Technology (Eawag), and I would like to thank Eawag. A number of great people made the institute an interesting and enjoyable place to work. First of all, there are my team- and office-mates Paola and Markus. Thank you Paola for the humor you brought to our office and for your wise advice regarding finishing a PhD. Thank you Markus for the interesting and useful discussions. Thank you dear other members of ESS: Alex, Alice, Bernhard, Caroline, Fridolin, Jasmine, Mara, Mario, Mika, Mirella, Pauline, Philipp, Simon J., Simon M., and Ulrike. Last but not least, thank you Maja for all the food supply. Thank you. xi Contents Summary vii Acknowledgements ix 1 Introduction 1 1.1 Background ............................... 1 1.2 Main objective, hypotheses, and research questions ........ 3 1.3 Data collection and econometric analysis .............. 6 1.4 PhD thesis outline ............................ 8 2 Hydropower Externalities: A Meta-Analysis 11 2.1 Introduction ............................... 11 2.2 Study selection and characteristics .................. 15 2.3 Meta-model ............................... 19 2.3.1 Heterogeneity, heteroskedasticity, and non-independence 19 2.3.2 The meta-regression models .................. 20 2.4 Selection and definition of variables ................. 22 2.5 Results .................................. 27 2.5.1 Descriptive statistics ...................... 27 2.5.2 Meta-regression results .................... 29 2.5.3 Cross-validation ........................ 33 2.6 Conclusions and discussions ...................... 35 2.A Studies included in the meta-analysis ................ 38 3 Choice Certainty, Consistency, and Monotonicity 41 3.1 Introduction ............................... 41 3.2 Choice certainty, consistency, and monotonicity in DCEs ..... 44 3.3 Econometric Models .......................... 48 xii 3.4 Case-study description ......................... 50 3.4.1 Discrete choice experiment .................. 50 3.4.2 Elicitation of choice certainty, consistency, and monotonicity 52 3.4.3 Sampling procedure and choice experiment design .... 53 3.5 Results .................................. 55 3.5.1 Descriptive results ....................... 55 3.5.2 Swait-Louviere test results ................... 57 3.5.3 Logit model results ....................... 62 3.6 Discussion and conclusions ...................... 66 4 Attribute non-Attendance in Discrete Choice Experiments 69 4.1 Introduction ............................... 69 4.2 Attribute non-attendance ........................ 71 4.3 Case-study description and experimental design .......... 77 4.4 Elicitation of stated and visual ANA ................. 79 4.5 Econometric models .......................... 81 4.6 Results .................................. 84 4.6.1 Descriptive statistics .....................